article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
in cells , a small variety of expression of some protein may influence cell metabolism seriously . in synthetic biology , many models have been presented to describe the metabolic network @xcite . according to these models , it often needs to express a certain kind of protein ( especially enzyme ) in a specific intensity .
one of the widely used ways to do this is to adjust the nucleotide sequence of the corresponding promoter @xcite .
a promoter is a region of dna that initiates transcription of a particular gene @xcite , see fig .
[ figschematic ] . in gene expression ,
the genetic information coded in nucleotide sequence of dna should be firstly transcribed into message rna ( mrna ) , which is performed by enzyme rna polymerase ( rnap ) @xcite .
usually , the transcription process begins with the binding of rnap to one specific upstream region of the target gene , which is called promoter @xcite .
experiments show that , with different promoters , the protein production rates , or the strengths of gene expression , will be different @xcite .
therefore , the rate of gene transcription to mrna is regulated by the nucleotide sequence of promoter ( for simplicity , in this study , promoters are assumed to be constitutive , i.e. transcription rate of the corresponding downstream gene is not influenced by transcription factors , for related discussions about translation factors one can see @xcite ) . due to the requirement of genetic engineering and synthetic biology , the production rates of certain proteins , especially some enzymes ,
should be regulated detailedly .
one of the ways to attain this aim is to choose specific promoter sequence to get needed rate of transcription ( another efficient way is to choose specific ribosome binding site sequence , i.e. rbs sequence , to regulate the rate of translation @xcite ) . in order to achieve this goal ,
many promoter libraries corresponding to large - scale strength of gene expression have been built experimentally @xcite . since it has been experimentally verified that the activity of a promoter can be reused among different kinds of proteins @xcite , these promoter libraries are valuable to the regulations of metabolic networks involving many different kinds of proteins .
the establish of promoter libraries is of great important to the development of synthetic biology .
however , reliable mathematical models to describe the relationship between promoter strength and its nucleotide sequence are much desirable . since such models can not only reduce the experimental expense in building further promoter libraries , but also can help us to understand the mechanism of promoter during gene transcription .
more importantly , such models will make it easier to get needed promoter with specific expression strength , and then will be valuable in genetic engineering and synthetic biology . in the last twenty years , many studies have been done to build quantitative relationship between promoter strength in gene expression and its nucleotide sequence .
it has been discovered early that , in _
e. coli _ promoters , the -10 region hexamer and -35 region hexamer are strongly conserved , and they are much important for determining the expression strength of promoter @xcite . but
recent experimental data shows that promoter strength also depends on nucleotide types in the spacing region of promoter sequence @xcite , and may also depend on the discriminator region .
where spacing region is the promoter sequence region between the most conserved -10 region and -35 region , and discriminator region is the sequence region between the -10 region and transcription start site , see fig .
[ figschematic](d ) .
therefore , it becomes much difficult to build reasonable theoretical models to describe the relationship between promoter sequence and its strength , since there may be too many factors or variables which may affect the promoter strength . in @xcite ,
one modular position weight matrix model is presented to evaluate the contribution of promoter sequence to its strength . where promoter score , which correlates with protein - dna binding energy and consequently correlates with promoter strength , is obtained as one linear combination of scores for each active promoter sequence modular and an additional penalty term for nonoptimal modular , with the sequence modular scores obtained by basic principles of statistical physics .
similar idea has also been used in @xcite to try to understand the promoter strength from its nucleotide sequence . meanwhile , in @xcite , based on a large number of experimental data for strengths and sequences of _ e. coli lac _
promoters , an adapted energy matrix for rnap binding to promoter is statistically determined .
based on this energy matrix , one thermodynamic model is designed in @xcite to predict promoter strength from its nucleotide sequence . on the other hand , by using support vector regression method and the distribution of specific nucleotides at each position of promoter sequence , which is obtained in @xcite and based on 168 _ e. coli _
promoter sequences , one strength prediction skill of _ e. coli _ promoter from microarray data is provided in @xcite .
they found that several non - consensus nucleotides in the -10 region and -35 region of promoter sequences act positively on the promoter strength , while certain consensus nucleotides have only a minor effect on the strength .
although many related studies have shown that the consensus sequences ( -35 and -10 regions ) are most essential to determine the strength of promoter @xcite , and actually for the sake of simplicity most of the existing models are based on such assumption , recent experimental data for promoter strength obtained by wang s study group in @xcite indicates that nucleotide types in spacing region of promoter sequence are also not neglectable . with different spacing sequences but keeping -10 region and -35 region unchanged , the expression strength of promoter may vary between 31 and 105 ( in unit relative intensity of red fluorescent per od@xmath0 , i.e. rirf / od@xmath0 ) . for simplicity , this study assumes that gene transcription will initiate as soon as one rnap binds to its upstream promoter , and the nucleotide sequence of gene is not too long .
then for low concentration of rnap , which assures that the elongation of mrna ( i.e. the motion of rnap along dna ) will not be jammed @xcite , the transcription rate will mainly be determined by the binding rate of rnap to promoter .
biophysically , rnap binding rate to promoter can be roughly written as @xmath1 .
where @xmath2 is one rate constant which depends on rnap concentration and other experimental environments , @xmath3 is boltzmann constant , @xmath4 is the absolute temperature , and @xmath5 is the free energy barrier of rnap binding to promoter which is mainly determined by the promoter sequence . generally , the energy barrier @xmath5 may depend on the secondary structure of promoter , and may also depend on the concentrations of transcription factors for nonconstitutive promoters .
but this study assumes that @xmath5 can be completely determined by the nucleotide sequence of promoter , and the main focus in the following is to find one reasonable method to get energy barrier @xmath5 from the nucleotide sequence of promoter .
one immediate idea to get energy barrier @xmath5 is to assume that @xmath5 can be approximated by one linear combination of energy barriers @xmath6 contributed by each nucleotide in the promoter sequence .
here @xmath7 is the index of position of nucleotide in promoter sequence .
this idea is similar as the ones previously used in @xcite .
the main difference between the idea here and the previous ones is as follows . in the previous methods , such as the one used in the modular position weight matrix model in @xcite ,
@xmath5 is assumed to be determined only by some so called active modules , including -35 region , -10 region , discriminator region , transcription start region , and the contribution from other suboptimal regions is only included as one penalty term which depends only on their lengths of nucleotide sequences . as have been mentioned previously , recent experimental data presented in @xcite indicates promoter strength also changes with the nucleotide types in spacing region .
so we need to calculate the energy barrier contribution @xmath6 from nucleotide in any position @xmath7 of promoter sequence .
there are several possible ways to get @xmath6 , which base more or less on basic principles of statistical physics @xcite .
for example , the energy barrier contribution of one nucleotide at position @xmath7 with type @xmath8 ( @xmath9 or @xmath10 ) can be obtained as @xmath11 @xcite . where @xmath12 is the number of nucleotide @xmath8 at position @xmath7 in the aligned sequences , @xmath13 is the total number of promoter , and @xmath14 is the theoretical probability of finding each type of nucleotide ( usually @xmath15 is used ) , and 0.005 and 0.02 are just two other model parameter values .
such method may be theoretically sound , but our calculations indicate it is not reasonably good to describe the experimental data of promoters in @xcite .
if the types of nucleotide at different positions of promoter sequence are independent to each other , then due to principles of statistical physics , the probability that there is one base @xmath8 at position @xmath7 is related to the free energy @xmath16 by @xmath17 . by replacing probability @xmath18 with frequency @xmath12 ,
we then get @xmath19 . where the frequency @xmath12 is obtained from the 168 promoters of _ e. coli _ compiled in @xcite .
the total energy barrier @xmath5 for one given promoter sequence can then be obtained by @xmath20 , with @xmath8 the nucleotide at position @xmath7 .
however , this model can not fit to the recent experimental data presented in @xcite well , see fig .
s1(a ) in supplemental material @xcite . in recent paper @xcite , based on experimental data of _ e. coli lac _
promoters , one @xmath21 parameter matrix @xmath22 is built , which is used to describe the interaction of rnap with promoter region from position -1 to position -75 . where the matrix element @xmath23 represents the contribution to this interaction from having a base @xmath8 at position @xmath24 in the promoter sequence .
using this parameter matrix @xmath22 , the total energy barrier for rnap binding to one promoter can be known , and so the corresponding expression strength can be obtained .
this parameter matrix has been employed by brewster _
_ in @xcite to build one theoretical model to predict the strength of promoter , and its accuracy has been validated in the design of promoter with specific strength .
however , our test shows that such method is not satisfying when it is employed to describe the recent experimental data presented in @xcite , see fig .
s1(b ) in supplemental material @xcite .
one common characteristic of the three methods discussed above is that , the energy barrier @xmath5 of rnap binding to promoter is assumed to be one linear combination of energy barriers @xmath16 contributed by each nucleotide in promoter sequence .
the failures of these methods in describing the recent experimental data presented in @xcite implie that this assumption of independence and additivity of energy barriers may not be generally true , though it may be approximately reasonable in some special cases and has been validated previously for transcription factor binding sites of promoter @xcite .
therefore , this assumption needs to be modified to be more reasonable .
one immediate way to do this is to assume that the energy barrier @xmath5 of rnap binding to promoter is one linear combination of the ones contributed by all nearest - neighbor ( nn ) nucleotide groups in promoter sequence . for convenience
, this study assumes that the energy barrier contributed by one nn nucleotide group is equal to the energy obtained by * nn model * in the study field of nucleic acids @xcite . which has actually been used in some software packages , such as nupack @xcite , to calculate the folding free energy of nucleic acid sequence to determine its secondary structure . in * nn model * , the total energy of one given nucleic acid sequence is obtained as the summation of energies contributed by each nn nucleotide groups , and one additional term according to the initial nucleotide , see the corresponding values listed in fig .
s1(d ) @xcite .
for example , the energy of sequence cgttga at temperature 37@xmath25c is obtained as @xmath26 ( excerpted from ref .
@xcite ) . with one additional parameter
to indicate the average level of experimental environments , the fitting results of this model are plotted in fig .
s1(c ) @xcite .
unfortunately , this model is also not satisfactory .
the failure of the above nn method to predict promoter strength , i.e. the failure in calculating energy barrier @xmath5 of rnap binding to promoter , may due to following two reasons .
( 1 ) in the nn method , no difference is included among energy contributions from nucleotide groups in -10 region , -35 region , discriminator region , and spacing region . as mentioned above
, previous studies have shown that the -10/35 region may be more important for determining promoter strength .
so the nucleotide ( group ) in -10/-35 region should contribute more to the rnap binding energy @xmath5 .
( 2 ) the energy barrier @xmath5 may also depend on large nucleotide groups , at least on nucleotide groups with three adjacent nucleotides . in the following ,
we will test new models which do not have at least one of the above two weaknesses .
firstly , we test the model which includes nucleotide position explicitly , where the energy barrier @xmath5 of rnap binding to promoter is assumed to be linear combination of @xmath16 and @xmath27 . here
@xmath16 is the energy barrier contributed by one nucleotide @xmath8 at position @xmath7 , and @xmath27 is the energy barrier contributed by one nearest - neighbor nucleotide group @xmath28 with nucleotide @xmath8 at position @xmath7 and nucleotide @xmath29 at position @xmath30 . for convenience , in the following
, this model is called position2 model .
one main difficulty in such model is , compared with the known experimental data , there are too many model parameters . for example , if the promoter sequence comprises of 35 nucleotides , then there will be @xmath31 model parameters . to avoid the overfitting problem , we used the partial least squares ( pls ) regression to get parameter values . where the principal component number , i.e. the number of independent model parameters , is chosen by 10-fold cross - validation . with the chosen principal component number ,
both the mean residual of all promoter strength between measured values and theoretical values , and the mean residual in 10-fold cross - validation are reasonably low , see fig .
s2(a , b ) in supplemental material @xcite . in this study ,
the experimental data includes a total of 422 promoters , with their nucleotide sequences and strengths presented in refs .
@xcite respectively . due to the different experimental environments used in measuring promoter strengths , including temperature , rnap type and concentration , and lots of other conditions in transcription and translation processes , in this model as well as other models used in the study ,
extra constants are added to total energy barrier @xmath5 to stand for these differences . for experimental data from different references ,
these extra constants will be different , and their values are also obtained by pls regression . meanwhile , to know if there are real differences between the energy barrier contributions from nucleotides in -10 region , -35 region , discriminator region , and spacing region , another extra constant is added to distinguish the length of promoter spacing region .
the possible lengths of spacing region in the experimental data used in this study are 16 , 17 and 18 . with seven principal components ,
the fitting results of the position2 model are presented in fig .
s2(c ) @xcite . where the mean residual between the experimental data and theoretical values is around 61 ( in arbitrary unit ) , see fig .
s2(a ) in supplemental material @xcite . from the model coefficient values plotted in fig .
s2(d , e , f ) @xcite , one can see that the contributions to total energy barrier @xmath5 from -10 region , -35 region , and discriminator region are larger than those from the spacing region for promoters with any length of spacing region . here
the model coefficients are obtained by an inverse transform from the values of the seven principal components .
one may argue that one of the reasons that the above nn model is not good enough to describe the relationship between promoter strength and sequence is that , the values of @xmath32 given by ref .
@xcite for energy barrier contribution from nearest - neighbor nucleotide group @xmath28 may not be accurate enough ( see also the values listed in fig .
s1(d ) in supplemental material @xcite ) , or they may not be generally right for any nucleic acid sequence . in order to exclude this reason , we have tested one generalized model , called group2 model , in which the energy barrier @xmath5 of rnap binding is assumed to be one linear combination of @xmath33 and @xmath32 .
different with the above nn model , @xmath33 and @xmath32 here are obtained by pls regression . however , from the plots in fig .
s2(a , b ) @xcite , one can see that the group2 model is actually less accurate than the above position2 model . as mentioned previously ,
another possible reason for the failure of nn model to describe the relationship between promoter sequence and strength is that , the promoter strength may also depend on large nucleotide groups ( with at least three adjacent nucleotides ) , but not only on independent nucleotides and nucleotide groups with two neighboring nucleotides .
we may need to point out that , models with nucleotide groups of size only 3 implicitly include the cases of size 1 and 2 .
but , for the sake of comparison of contributions to the total energy barrier @xmath5 of rnap binding to promoter , from nucleotide groups with different sizes , our model includes all possible nucleotide groups with size 1 , 2 , and 3 . for convenience , such model
is called group3 model . in group3 model , there are also too many unknown model parameters which need to be fitted from experimental data , altogether @xmath34 . therefore , pls regression is also used in the data fitting process to determine model parameter values , in which the principal component number ( i.e. the number of independent model parameters ) is determined by reducing both the mean residual between promoter strengths from experimental data and from theoretical model , and the mean residual of promoter strengths in 10-fold cross - validation , see fig .
s3(a , b ) in supplemental material @xcite . with eight principal components ,
the fitting results are presented in fig .
s3(c ) @xcite . from the model coefficient values plotted in fig .
s3(d ) @xcite , one can see that contributions to total energy barrier @xmath5 of rnap binding from nucleotide groups with three adjacent nucleotides are relatively larger than those from nucleotides or nucleotide groups with two neighboring nucleotides .
based on the above two test models , the position2 model and the group3 model , we conclude that the energy barrier @xmath5 of rnap binding to promoter depends on nucleotide groups with three adjacent nucleotides .
meanwhile , the contributions of nucleotide ( groups ) to @xmath5 from different regions of promoter sequence are different
. therefore , it seems more reasonable to use one combined model to describe the relationship between promoter strength and sequence , in which both nucleotide groups with size up to three and their positions in promoter sequence are explicitly considered .
undoubtedly , there will be too many unknown parameters in such combined models .
for example , if each promoter has 35 nucleotides and there is no missing nucleotide in promoter sequence , the number of model parameters will be @xmath35
. meanwhile , as mentioned previously , during data fitting extra parameters are also added due to the difference of length of promoter spacing region and the difference of experimental environment to measure promoter strength .
therefore , pls regression should also be used to determine the values of model parameters . with reasonable low value of
mean residual between theoretical model predictions and experimental data of promoter strengths , and mean residual of 10-fold cross - validation , the principal component number of this combined model is chosen to be seven , see fig .
s3(a , b ) in supplemental material @xcite . with these seven principal components ,
the theoretical results of this combined model ( for convenience , this model is called position3 model ) are plotted in fig . [ figposition3](a ) .
the mean residual between theoretical predictions and experimental measurements of promoter strengths is about 59 ( in arbitrary unit ) , see fig.s3(a ) . from the plots in fig .
[ figposition3](b , c , d ) , one can easily see that the nucleotide groups in -10 region , -35 region , and discriminator region have more effects on promoter strength .
meanwhile , the effects of nucleotide groups with three adjacent nucleotides are also nonnegligible , see fig .
s4 in supplemental material @xcite . using the position3 model and genetic algorithm ,
four promoter libraries are calculated ( see fig .
[ figlibraries1 ] , and the excel files in supplement materials for the corresponding promoter sequences ) , which correspond to the four experimental environments under which the data used in this study are obtained @xcite .
these libraries will be helpful for the synthesis of promoter with specific expression strength . using the same methods , promoter libraries corresponding to any other experimental environments
can also be theoretically built , but with one extra step during which the model constant corresponding to the given experimental environment should be determined firstly by initial promoter samples . combining our model here for promoter strength with the calculation method for ribosome binding site sequence @xcite , the gene expression , including transcription and translation ,
can be regulated to given strength detailedly . in conclusion , in this study one theoretical model for describing the relationship between promoter strength and sequence
is presented .
our study shows that the nucleotides in -10 region , -35 region , and the discriminator region have more effects on promoter strength than those in spacing region
. meanwhile , promoter strength depends more on nucleotide groups with three adjacent nucleotides than on single nucleotides and nucleotide groups with two neighboring nucleotides . using our model , promoter libraries with wide range of expression strength
are theoretically obtained .
the lengths of promoter spacing region in * ( b , c , d ) * are 16 , 17 , and 18 respectively . the plots in * ( b , c , d ) * imply that , generally nucleotides in -10/-35 region and discriminator region are more important for determining the strength of promoter , see also fig .
s2(d , e , f ) @xcite .
, title="fig:",width=566 ] +
|
promoter strength , or activity , is important in genetic engineering and synthetic biology .
evidences show that a constitutive promoter with certain strength for one given rna can often be reused for other rnas .
therefore , the strength of one promoter is mainly determined by its nucleotide sequence .
one of the main difficulties in genetic engineering and synthetic biology is how to control the expression of certain protein in one given level .
one usually used way to achieve this goal is to choose one promoter with suitable strength , which can be employed to regulate the rate of transcription and then leads to needed level of protein expression . for this purpose , so far , many promoter libraries have been established experimentally .
however , theoretical methods to predict the strength of one promoter from its nucleotide sequence are desirable . since such methods are not only valuable in the design of promoter with specified strength , but also meaningful to understand the mechanism of promoter in gene transcription .
in this study , through various tests one theoretical model is presented to describe the relationship between promoter strength and its nucleotide sequence .
our analysis shows that , promoter strength is greatly influenced by nucleotide groups with three adjacent nucleotides in its sequence .
meanwhile , nucleotides in different regions of promoter sequence have different effects on promoter strength .
based on experimental data for _ e. coli _ promoters , our calculations indicate , nucleotides in -10 region , -35 region , and the discriminator region of promoter sequence are more important than those in spacing region for determining promoter strength . with model parameter values obtained by fitting to experimental data ,
four promoter libraries are theoretically built for the corresponding experimental environments under which data for promoter strength in gene expression has been measured previously .
|
meshless methods have recently emerged as numerical techniques for electromagnetic modelling @xcite .
these methods do not require a predefined mesh , and use points scattered in the problem domain avoiding the need of information on the position among them .
one of the most popular meshless method , the smoothed particle hydrodynamics ( sph ) @xcite has been recently reformulated by the authors for computational electromagnetics ( cem ) problems .
maxwell s equations , which relate the electric and magnetic fields by means of a time dependent system of partial differential equations(pdes ) are considered @xcite . due to the coupling nature of the electric and magnetic field components , the points are in such a way that the electric points should be surrounded by those of magnetic field and vice - versa .
the method named as spem is applied in approximating the space field variables of the time domain maxwell s curl equations by using a kernel representation working with a cluster of scattered nodes .
the differential operators applied to the field variables are transmitted to a kernel function which characterizes the spatial discretization . despite the good spatial properties of spem , the numerical integration based on explicit method for time evolution , introduces a severe constraint .
in fact , as for the conventional grid based methods , spem has to satisfy the courant - friedrich - levy stability condition @xcite .
as well known , this condition becomes highly restrictive when the geometry resolution forces the use of much finer space discretization than that is needed to solve for waves away from material irregularities or interfaces .
this problem is firstly investigated in the paper in spem framework , in order to determine the largest allowed time increment . moreover , in technical literature , a great interest is addressed in finding numerical schemes which allow computational advantages together with unconditional stability @xcite .
otherwise , an alternate implicit direction method is proposed for the time evolution of the electromagnetic field in the spem framework by preserving the leapfrog marching on in time of the explicit integration scheme . by considering a 2-stage splitting approach ,
the leapfrog implicit formulation is carried out by suitably handling the formulas related to two adjacent time steps . by working with an implicit scheme ,
a sparse linear system is generated and it has to be solved at each time step .
the system matrix involves the kernel derivatives , and it can be assembled only once at the beginning of time integration process .
details on the matrices used in the problem are investigated .
the resulting algorithm is called as laf - spem method .
accuracy degradation , due to a lack of particle consistency , is also taken into account : a numerical corrective strategy , which allows to restore the consistency , without any modification of the smoothing kernel function and its derivatives , is employed @xcite .
test problems are simulated to validate the proposed methodology .
simulations running with a time step larger than that satisfying the cfl condition has been found to remain stable , so improving the conventional spem algorithm .
the paper is organized as follows . in section 2
basic ideas on smoothed particle hydrodynamics method are recalled in order to provide the original spem numerical scheme in solving time - domain maxwell s curl equations .
investigations about the stability criterion are also included . in section 3
the laf - spem method is described and simulation results are reported and discussed in section 4 also by comparing with standard fdtd computations and existing exact solution .
a brief conclusion complete the paper .
the sph scheme has been adopted by the authors to compute the time - domain maxwell s curl equations by using particle approximation , so obtaining spem method @xcite . for the reader convenience ,
a brief summary of spem fundamentals is reported in the following .
let consider the standard maxwell s curl equations in time domain , in a linear medium in the compact curl notation : @xmath0 @xmath1 where @xmath2 and @xmath3 are the electric and magnetic vectors fields , @xmath4 is the conduction current density vector , @xmath5 as the time , @xmath6 the derivative operator , @xmath7 the medium permittivity , permeability , and conductivity , respectively , which are supposed to be constant .
the sph method is based on an integral representation of the unknown function @xmath8 : @xmath9 + where @xmath10 is the problem domain and @xmath11 is the dirac delta function .
the delta function is approximated by a smoothing kernel function @xmath12 depending on the spatial coordinates and on the smoothing length parameter @xmath13 .
the smoothing kernel function must verify the following properties : 1 .
@xmath14 2 .
@xmath15 on a sub - domain of @xmath16 and @xmath17 outside ; 3 .
@xmath18 4 .
@xmath19 function value for a particle is monotonically decreasing + with increasing distance away from the particle .
the discrete spatial formulation of equations ( [ eq1 ] ) has been obtained by using sph particle approximation : the integral representation is approximated over a set of particles placed in the problem domain .
the particles are the points in which the fields components are computed at each time step , by using the information belonging to the neighbouring ones .
the smoothing length _
h _ defines the size of the support domain of each particle @xcite .
particular attention has to be paid on the choice of _
a too small _ h _ can lead to an inaccurate solution because not enough particles may be placed inside the influence area of a fixed particle @xmath20 ; a too large smoothing length , can smooth out local properties and the particle approximation suffers too . from now on , for each referred fixed particle @xmath20 , an own smoothing space length @xmath21 will be considered .
the sph particle approximation of ( [ eq2 ] ) can be expressed as a linear combination of translates of a kernel basis function : @xmath22 where _ np _ is the total number of particles in the influence domain of the fixed one @xmath20 and @xmath23 measures the surrounding media at the position of particle @xmath24 . _
np _ is strictly related to the choice of the smoothing length @xmath21 . in computing the two interlaved maxwell s curl equations , particles for the electric field ( @xmath25-nodes ) and magnetic field ( @xmath26-nodes )
must be considered in the spatial domain . due to the electric and magnetic coupling nature of maxwell s equations , spatial staggering of _ e_-nodes and _ h_-nodes
is required .
e_-nodes and the _ h_-nodes are arranged so that each _ e_-node is surrounded by _
h_-nodes and each _
h_-node is surrounded by _
the electric field update depends on the magnetic particles in the neighbourhood of the fixed electric particle and viceversa . in solving ( @xmath27 ) ,
fields spatial derivatives have to be approximated by means of sph .
the differential operators on the fields components are transmitted to the smoothing kernel function : so the kernel approximation allows differential operations to be determined from the values of the function and the derivatives of the kernel , rather than the derivatives of the function itself .
however , this is really true only when the influence domain of a fixed particle is placed entirely inside the problem domain ; when the influence domain overlaps with the geometry boundary the smoothing kernel _
w _ is truncated , and a resulting non - zero surface integral has to be opportunely treated in order to avoid the corruption of the numerical solution @xcite . in the following , for the sake of simplicity , and in agreement with the numerical results reported in section 4 ,
a lossless medium is considered .
thus , the two interleaved maxwell s curl equations ( [ eq2 ] ) result in spem formulation as follows : @xmath28 \delta v_{j } \end{array}\end{aligned}\ ] ] @xmath29 \delta v_{j}\\ \end{array}\end{aligned}\ ] ] @xmath30 \delta v_{j}\\ \end{array}\end{aligned}\ ] ] @xmath31 \delta v_{j}\\ \end{array}\end{aligned}\ ] ] @xmath32 \delta v_{j}\\ \end{array}\end{aligned}\ ] ] @xmath33 \delta v_{j}\\ \end{array}\end{aligned}\ ] ] where @xmath34 and @xmath35 denote the positions of the _ i_-th and _ j_-th particle for the electric and magnetic field component , respectively ; @xmath36 ( @xmath37 ) is the number of neighbouring particles of the _ i_-th fixed electric ( magnetic ) particle .
the first derivative operator is indicated as @xmath38 , _ x , y , z _ as the spatial coordinates .
a central finite difference algorithm is used to approximate time derivatives in ( @xmath39 ) and ( @xmath40 ) .
a leapfrog scheme requiring only explicit updates , has been used for advancing the solution in time : the electric field is computed at whole time step and the magnetic field at the half time step @xcite . by indicating with @xmath41 the time step and with @xmath42 the generic step
, a complete temporal step of ( [ eq1 ] ) evolves from _
n_-1/2 to _ _
n__+1 , by involving both electric and magnetic field updates .
the marching - on in time is actually from the step _
n_-1/2 to the step _ _
n__+1/2 when updating magnetic field while is from step _
n _ to the step _
_ n__+1 when updating electric field : @xmath43 \delta v_{j } \end{array}\end{aligned}\ ] ] @xmath44 \delta v_{j}\\ \end{array}\end{aligned}\ ] ] @xmath45 \delta v_{j}\\ \end{array}\end{aligned}\ ] ] @xmath46 \delta v_{j}\\ \end{array}\end{aligned}\ ] ] @xmath47 \delta v_{j}\\ \end{array}\end{aligned}\ ] ] @xmath48 \delta v_{j}\\ \end{array}\end{aligned}\ ] ] as well - known , the explicit time integration scheme is subjected to the cfl condition , that for the em problem is expressed as : @xmath49 where @xmath50 is the propagation velocity of the physical phenomenon in the medium , _ d _ is the geometric dimension of the problem , and @xmath51 is the minimum points spacing in the problem domain . in spem ,
the previous constraint depends on the smoothing length value ; moreover , this last is strictly related to the spatial particles distances .
therefore it requires the time step to be proportional to the smallest spatial point resolution : @xmath52 the previous condition guarantees the numerical stability , but it can makes the spem impractical to use , especially when the distance between the two closest nodes becomes very small . in this section
, some efforts are made to obtain a better maximum allowable time step .
namely , referring to the first case study reported in section 4 , a propagation of a 2d @xmath53 wave is considered . in this case , the discrete formulation of relations ( [ eq6 ] ) and ( [ eq7 ] ) are compactly re - written , as follows : @xmath54\\ \end{array}\end{aligned}\ ] ] @xmath55 @xmath56 + where @xmath57 are the vectors of the field components and @xmath58 as the matrices collecting the spatial field derivatives computed with spem formulation and the physical parameters of the problem domain . by substituting ( [ eq9 ] ) and ( [ eq10 ] ) in ( [ eq8 ] ) , i.e. : @xmath59\textbf{e}_{z}^{n}+\delta t
[ \textbf{th}_{y}^{n-1/2}-\textbf{uh}_{x}^{n-1/2}]\\ \end{array}\end{aligned}\ ] ] + the evolution in time is described by the relation @xmath60 , where : + @xmath61 + @xmath62 + in order to @xmath63 be bounded , as n @xmath64 , the time increment must satisfy the following relation [ 27 ] : @xmath65 + where @xmath66 is the maximum eigenvalue of the matrix @xmath67 .
the ( [ eq12 ] ) gives an upper limit to the increment of the time step getting away from the minimum particles spacing .
furthermore , the restriction on @xmath41 limits the physical applications yet , so that in the following an unconditionally stable time advance equations has been introduced .
the original spem method has been modified by preserving the leapfrog evolution in time of the explicit time integration scheme .
in this section the leapfrog adi - fdtd method @xcite is generally described and re - formulated for generating the implicit leapfrog formulation of spem . in the adi - fdtd method , the temporal stepping from _
n _ step is obtained by splitting the integer time step into two half - time steps . by indicating with _ e _ and _ h _ the fields components with respect to a generic coordinate in the _ x , y , z _ geometric system , the finite difference time domain scheme involves the following general relations , in which the actual field value is formally expressed as a generic function of the fields at previous time steps : @xmath68 @xmath69 @xmath70 @xmath71 + for the marching - on in time of the electric field components , equation ( [ eq16 ] ) is substituted in ( [ eq15 ] ) and ( [ eq16 ] ) is substituted in ( [ eq15 ] ) , so obtaining : @xmath72 @xmath73 in ( [ eq17 ] ) and ( [ eq18 ] ) first and second order spatial field derivatives hold and tridiagonal linear systems have to be solved to generate @xmath74 and @xmath75 .
an adi - fdtd time step is so described by means of relations ( [ eq17 ] ) and ( [ eq14 ] ) , and relation ( [ eq18 ] ) and ( [ eq16 ] ) , respectively . by considering the temporal step for the fields evolution from step @xmath76 to @xmath77 and obtaining @xmath78 from ( [ eq16 ] ) , the following relation holds : @xmath79 by substituting ( [ eq19 ] ) into ( [ eq15 ] ) the electric field component is generated by means of : + @xmath80 let now consider equation ( [ eq17 ] ) for the marching - on in time of the electric field from _
n _ to @xmath81 : + @xmath82 and by using ( [ eq20 ] ) , the previous relation can be re - written as : + @xmath83 + the leapfrog equation describing the marching - on in time of the electric field component is so obtained .
let now consider relation ( [ eq15 ] ) for the half time step from _
_ n__+1/2 to _
_ n__+1 : @xmath84 + by substituting it into ( [ eq16 ] ) , the following relation holds @xmath85 + for the half time step from _
n _ to _ _ n__+1/2 relation ( [ eq13 ] ) is considered : + @xmath86 + by substituting this last into ( [ eq14 ] ) .
the @xmath87 component holds : + @xmath88 + this last relation is subsituted in ( [ eq23 ] ) so obtaining the following : + @xmath89 + equations ( [ eq21 ] ) and ( [ eq23 ] ) give rise to laf scheme for marching - on in time of the generic magnetic field component _ h_. in ( [ eq21 ] ) and ( [ eq23 ] ) only first and second spatial field derivatives are involved .
the electric and magnetic fields are staggered in time and updated in one full time step , iteratively .
the complete formulation of maxwell s curl equations derived from eqs ( [ eq21 ] ) and ( [ eq24 ] ) is reported in the following by approximating the spatial first and second derivatives by means of sph particle approximation : @xmath90\delta v_{j } \end{array}\end{aligned}\ ] ] @xmath91\delta v_{j } \end{array}\end{aligned}\ ] ] @xmath92\delta v_{j } \end{array}\end{aligned}\ ] ] @xmath93\delta v_{j } \end{array}\end{aligned}\ ] ] @xmath94\delta v_{j } \end{array}\end{aligned}\ ] ] @xmath95\delta v_{j } \end{array}\end{aligned}\ ] ] where @xmath96 , @xmath97 , @xmath98 , @xmath99 .
+ the equations ( 25)-(30 ) can be expressed in a compact matrix notation as follows : @xmath100 + where : @xmath101 @xmath102 and the non zero entries are : @xmath103 @xmath104 the matrices employed in * a*,*b*,*c*,*f * are sparse block matrices and the sparsity depends on the influence domain of each particle .
each row of the matrices refers to a fixed particle , center of a kernel function , and the non zero elements correspond to the particles surrounding the fixed one .
the matrix @xmath105 ( @xmath106 ) is generated from the smoothing kernel functions regarding the _
e_-particles , i.e. each kernel is centered on @xmath34 and has @xmath107 as neighbourings . in a similar fashion
the matrix @xmath108 ( @xmath109 ) is related to the smoothing kernel functions centered on @xmath110 and has @xmath35 as neighbourings . in all these matrices
the second derivatives are involved .
the matrices @xmath111 ( @xmath112 ) regard smoothing kernel centered on @xmath34 but involving @xmath35 as neighbourings .
analougously , @xmath113 ( @xmath114 ) regards smoothing kernel fixed on @xmath110 but involving @xmath107 as neighbourings . in all these last matrices
the first derivatives are involved .
when kernel vanish , the limit configuration of the explicit formulation in time is obtained . in the following ,
some matrices configurations are reported .
equally distributed particles are considered generating a structured block matrices ( figs.1,2 ) . in this case
the matrices are banded and the bandwidth is with the width of the local support of the kernel function . the smoothing length _ _
h__=@xmath115 must be opportunely chosen to achieve satisfactory approximation field values , giving rise to a sparse and solvable system .
the approach in which the support size remain fixed for increasingly denser set of data is very inefficient with associate system matrices increasingly denser . in the simulations ,
the smoothing length is scaled so that peaked basis functions are with densely spaced particles and flat basis functions are with coarsely spaced particles .
+ in fig.3 the error in the computation is reported by considering 100 equally distributed points over the whole problem domain . for the electric field in the @xmath53 case ,
the error is measured in @xmath116 in comparison with the finite difference time domain solver .
the best behaviour is with @xmath117=0.7075 . in tab.1 results
are reported by considering the optimal smoothing length found , to give an accurate solution for this problem .
the sparsity of the matrices increases by increasing the problem size . [ cols="^,^",options="header " , ] * table 1*. results with regular particle distribution by increasing the particles density n over the problem domain in the computation of the matrix @xmath105 + [ fig : fig1 ] for a regular particle distribution , title="fig:",width=377 ] [ fig : fig2 ] in @xmath118 of figure 1,title="fig:",width=377 ] [ fig : fig3 ] in figs.4,5 the block matrix @xmath105 is depicted for irregular particle distributions , by varying the number of particle in the problem domain . fig.4 is with 81 particles and a sparsity of about 80 % is generated .
fig.5 refers to 625 particles with sparse matrix of about 96 % of zero elements .
[ fig : fig4 ] for an irregular 9x9 particle distribution , title="fig:",width=377 ] [ fig : fig5 ] for an irregular 25x25 particle distribution , title="fig:",width=377 ] the kernel derivatives in computing the derivatives of the field components can be performed in a pre - processing stage , and all the matrices employed in the computation can be assembled out of the temporal loop . moreover , by considering the smoothing length independent from the evolution in time , the structure of the matrix not depends on the temporal step yet .
in this section , some different case studies have been performed and results are discussed to assess the proposed meshless implicit leapfrog numerical method . at first , a transverse electric @xmath119 ) mode in air is firstly taken into account . in this case , equations are reduced as follows : @xmath120 a regular particles distribution is built up , such as in a fdtd spatial grid . for problem with regular distribution ,
nodes are placed as in the point - matching procedure ; for the general case , a voronoi decomposition @xciteor others technique can be adopted . in particular , in the paper when an irregular particles distribution is set , in order to avoid particle inconsistency which can lead to a loss of accuracy the consistency restoring adopted in @xcite , is used .
the fdtd simulation is considered for comparison . the perfect matching layer ( pml ) technique is used in simulating the open spatial domain @xcite . in this way , also a good treatment of the surface integral truncation problem , as reported in section 2 , is carried out for laf - spem computation .
the actual time step , satisfies the cfl condition , so as in the fdtd approach , i.e. : @xmath121 where @xmath122 is the spatial step . in figure 6 fdtd and laf - spem spatial profiles along y direction , ten cells distant from the middle in the x direction in a 2d domain ,
100x100 number of cells are reported ; a time varying shaped source , @xmath123 , placed at the middle point of the domain is considered in simulating a @xmath53 field in air .
the comparison shows a satisfactory agreement .
+ [ fig : fig6 ] field is simulated in air .
@xmath124 field ( v / m ) spatial profiles at time step 70 , along y direction ten cells distant from the middle in the x direction , title="fig:",width=453 ] as a further validation , an axial symmetric cylindrical domain in air is considered , with @xmath125 ( @xmath126 as constitutive parameters ) , with the following boundary and initial conditions : @xmath127 + @xmath128 + the domain is discretized with 10x10 irregular distributed particles obtained by randomly positioning the particles near the original regular position in the cylindrical domain .
the actual time step is set equal to @xmath129 . as shown in figure 7
the simulation remains stable also for a final time corresponding to 175 time steps , and a satisfactory approximation has been achieved .
[ fig : fig7 ] bounded by a perfect electric conductor : comparison between laf - spem results and the exact solution at time step 175 .
the laf - spem time step is 4 times that satisfying the cfl condition.,title="fig:",width=453 ] in order to better assess the validity of the proposed approach , a more realistic example has been carried out . a sectorial ( 2-d ) perfectly electric conducting ( pec )
horn antenna excited by a sinusoidal voltage in a transverse magnetic @xmath119 computational domain has been simulated and compared with fdtd results [ 16 ] @xcite .
the computational domain is truncated with a pml absorbing boundary condition .
the horn is modelled by a perfect electric conductor ( pec ) material .
the spatial cell size is equal to 0.0025 m , the time step is equal to 4.23 ps , the excitation frequency is equal to 9.84 ghz .
[ fig : fig8 ] ) holds .
the reported relative error is related to fdtd results.,title="fig:",width=453 ] in figure 8 , laf - spem @xmath130 field results , in a 2-d domain with 100x100 regular particles distribution , are reported , at time step 198 .
the excitation section is also reported in the figure .
the actual time step , @xmath41 , satisfies the cfl condition ( [ eq33 ] ) . in comparison with fdtd simulation
, a good agreement has been obtained . for a better comparison with fdtd results , in figure
9 fdtd and laf - spem @xmath131 field spatial profiles along the direction containing the longitudinal axis of the horn antenna at the same time step of figure 6 , are reported .
[ fig : fig9 ] ) holds .
, title="fig:",width=453 ] in order to show the capability of the proposed approach to run with a time step greater than that obtained by satisfying the cfl condition , a simulation with a time step equal to @xmath132 cfl is carried out .
also in this case , an irregular particles distribution has been also adopted by randomly positioning the particles near the original regular position in the domain .
as shown in figure 10 , the simulation remains stable also for a final time corresponding to 200 time steps , and a satisfactory approximation has been achieved .
[ fig : fig10 ] field ( v / m ) results obtained with fdtd and laf - spem at the same final time , along the direction containing the longitudinal axis of the antenna .
the laf - spem time step is 4 times that satisfying the cfl condition.,title="fig:",width=453 ]
in this paper the sph meshless method has been composed with a leapfrog adi - fdtd method for time evolution of electromagnetic time - domain transient propagation problems .
the new schemeimproves the original spem method , already developed by the authors .
the method is very promising because avoids grid generation in space and results unconditionally stable in time . in this way
, realistic simulations can be performed with reasonable computational efforts .
the computational tool is assessed by using different numerical simulations also employing irregular particles distribution .
perfect matching layer technique is used in simulating open spatial problems and a consistency restoring approach is introduced .
the proposed methods may have good perspectives of extensive applications .
+ g. ala , e. francomano , an improved smoothed particle electromagnetics method in 3d time domain simulations , international journal of numerical modelling : electronic networks , devices and fields 25 ( 2012 ) 325 - 337 .
g. ala , e. francomano , smoothed particle electromagnetics modelling on hpc - grid environment , applied computational electromagnetics society ( aces ) journal 27 ( 3 ) ( 2012 ) 229 - 237 .
g. ala , g. di blasi , e. francomano , a numerical meshless particle method in solving the magnetoencephalography forward problem , international journal of numerical modelling : electronic networks , devices and fields 25 ( 2012 ) 428 - 440 .
g. ala , e. francomano , a. tortorici , a. spagnuolo , a meshless approach for electromagnetic simulation of metallic carbon nanotubes , j. mathematical chemistry 48 ( 1 ) ( 2010 ) 72 - 77 .
g. ala , e. francomano , a multisphere particle numerical model for non - invasive investigations of neuronal human brain activity , progress in electromagnetics research letters,36 ( 2013 ) 143 - 153 .
r. araneo , s. celozzi , analysis of the shielding performance of ferromagnetic screens , ieee transactions on magnetics 39 ( 2 ) ( 2003 ) 1046 - 1052 . t. kaufmann , y. yu , c. engstrom , z. chen c. fumeaux , recent developments of the meshless radial point interpolation method for time - domain electromagnetics , international journal of numerical modelling : electronic , networks , devices and fields , 25 ( 2012 ) 468 - 489 .
cooke , m. botton , t.m .
antonsen , b. levush , a leapfrog formulation of the 3-d adi - fdtd algorithm , international journal of numerical modelling : electronic networks , devices and fields 22 ( 2009 ) 187 - 200 .
y. duan , s.j .
lai , t. huang , coupling projection domain decomposition method and meshless collocation method using radial basis functions in electromagnetics , progress in electromagnetics research letters 5 ( 2008 ) 1 - 12 .
a. elsherbeni , v. demir , the finite - difference time - domain method for electromagnetics with matlab simulations , schitech , 2009 .
j. fang , a. parriaux , m. rentschler , c. ancey , improved sph methods for simulating free surface flows of viscous fluids , applied numerical mathematics , 59 ( 2 ) ( 2009 ) 251 - 271 .
t. kaufmann , c. fumeaux , r. vahldieck , the meshless radial point interpolation method for time - domain electromagnetics , ieee mtt - s international microwave symposium , usa , 2008 , 61 - 64 .
fulk , d.w .
quinn , an analysis of 1-d smoothed particle hydrodynamics kernels , journal of computational physics , ( 126 ) ( 1996 ) 165- 180 .
hart , s.j .
cox , k. djidjeli , compact rbf meshless methods for photonic crystal modelling , journal of computational physics , 230 , 12 ( 1 ) ( 2011 ) , 49104921 .
lai , b.z .
wang , y. duan , application of the rbf - based meshless method to solve 2-d time domain maxwell s equations , in : international conference on microwave and millimeter wave technology , 2008 , 749 - 751 .
lai , b.z .
wang , y. duan , meshless radial basis function method for transient electromagnetic computations , ieee transactions on magnetics 44 ( 10 ) ( 2008 ) 2288 - 2295 .
g. r. liu , m.b .
liu , smoothed particle hydrodynamics a mesh - free particle method , world scientific publishing , 2003 .
g. r. liu , mesh free methods : moving beyond the finite element method , world scientific publishing , 2003 .
liu , g.r .
liu , smoothed particle hydrodynamics ( sph ) : an overview and recent developments , archives of computational methods in engineering 17 ( 1 ) ( 2010 ) 25 - 76 .
x. liu , b.z .
wang , s. lai , element - free galerkin method for transient electromagnetic field simulation , microwave and optical technology letters 50 ( 1 ) ( 2008 ) 134 - 138 .
monaghan , an introduction to sph , comput .
48 ( 1988 ) 8996
monaghan , j.c .
lattanzio , a refined particle method for astrophysical problems , astron .
astrophys . 149
( 1985 ) 135143 .
monaghan , smoothed particle hydrodynamics , annu .
. astrophys . 30 ( 1992 )
moussa , b.b .
: on the convergence of sph method for scalar conservation laws with boundary conditions , methods and applications of analysis 13 ( 1 ) ( 2006 ) 29 - 62 .
t. namiki , a new fdtd algorithm based on alternating - direction implicit method , ieee transactions on microwave theory and techniques , 47 ( 10 ) ( 1999 ) 2003 - 2007 .
k. shanazari , n. rabie , a three dimensional adaptive nodes technique applied to meshless - type methods , applied numerical mathematics , 59 ( 6 ) ( 2009 ) 1187 - 1197 .
d. soares jr , time - domain electromagnetic wave propagation analysis by edge - based smoothed point interpolation methods , journal of computational physics,234 ( 2013 ) 472486 .
sullivan , electromagnetic simulation using the fdtd method , ieee press , 2000 . y. yu , z. chen , a 3-d radial point interpolation method for meshless time - domain modelling , ieee transactions on microwave theory and techniques 57 ( 8) ( 2009 ) 2015 - 2020 .
y. yu , z. chen , towards the development of an unconditionally stable time - domain meshless method , ieee transactions on microwave theory and techniques 58 ( 3 ) ( 2010 ) 578 - 586 .
x. zheng , z. chen , j. zhang , a finite - difference time - domain method without the courant stability conditions , ieee microwave and guided wave letters 9 ( 11 ) ( 1999 ) 441 - 443 .
|
numerical solution of equations governing time domain simulations in computational electromagnetics , is usually based on grid methods in space and on explicit schemes for the time evolution .
a predefined grid in the problem domain and a stability step size restriction must be accepted .
evidence is given that efforts need for overcoming these heavy constraints .
recently , the authors developed a meshless method to avoid the connective laws among the points scattered in the problem domain . despite the good spatial properties , the numerical explicit integration used in the original formulation of the method provides , also in a meshless context , spatial and time discretization strictly interleaved and mutually conditioned .
afterwards , in this paper the stability condition is firstly addressed in a general way by allowing the time step increment get away from the minimum points spacing . meanwhile , a formulation of the alternating direction implicit scheme for the evolution in time is combined with the meshless solver .
the formulation preserves the leapfrog marching on in time of the explicit integration scheme .
the new method , not constrained by a gridding in space and unconditionally stable in time , is numerical assessed by different numerical simulations .
perfect matching layer technique is used in simulating open spatial problems ; otherwise , a consistency restoring approach is introduced in treating truncation at finite boundary and irregular points distribution .
three case studies are investigated by achieving a satisfactory agreement comparing both numerical and analytical results .
adi leapfrog method , meshless methods , smoothed particles electromagnetics .
|
spin effects in high energy fragmentation processes have attracted much attention@xcite recently .
study of such effects provide useful information for the spin structure of hadron and spin dependence of high energy reactions . there
exist now two distinctively different pictures for the spin contents of the baryons : the static quark model picture using su(6 ) symmetric wave function [ hereafter referred as su(6 ) picture ] , and the picture drawn from the data for polarized deep inelastic lepton - nucleon scattering ( dis ) @xcite and su(3 ) flavor symmetry in hyperon decay [ hereafter referred as dis picture ] .
it is natural to ask which picture is suitable to describe the relationship between the polarization of the fragmenting quark and that of the produced hadron .
it has been pointed out in [ [ bl98 ] ] that measurements of @xmath1 longitudinal polarization in @xmath0 annihilation at high energies can provide useful information to answer this question .
calculations based on these two different pictures have been made in [ [ gh93 ] ] and [ [ bl98 ] ] .
a comparison of the obtained results with the aleph data@xcite obtained at lep , which was available at that time , was made .
the result shows that the su(6 ) picture seems to agree better with the data@xcite .
this is rather surprising since the energy is very high at lep and the initial quarks and anti - quarks produced at the annihilation vertices of the initial @xmath0 are certainly current quarks and current anti - quarks rather than the constituent quarks used in describing the static properties of hadrons using su(6 ) symmetric wave functions .
other models@xcite are also proposed which can give a description of the available data@xcite .
the available data@xcite are certainly still far from accurate and abundant enough to make a conclusive judgment of these different models .
it is therefore important and urgent to make further checks of this conclusion by performing complementary measurements . in this paper
, we formulate the calculation method of longitudinal polarization of different hyperons in @xmath0 annihilation at high energies proposed in [ [ gh93 ] ] and [ [ bl98 ] ] in a systematic way so that they can be generalized to other hyperons and/or other reactions .
we describe in detail the inputs and/or assumptions which have been used in the calculations .
we then present the calculated results for different hyperons based on these two pictures .
we present the results for the whole events as well as those for different particularly chosen event samples .
we show that corresponding measurements can be used as complementary method to check the above mentioned conclusion .
in particular , we discuss also how to distinguish such models from the kind of models proposed in [ [ kotz98],[ma2000 ] ] . in section 2 , we outline the calculation method and present the results for @xmath1 polarization . in section 3
, we present our results for other hyperons .
we recall that@xcite , according to the standard model of electroweak interactions , quarks and anti - quarks produced at the annihilation vertices of @xmath0 are longitudinally polarized .
such longitudinal quark polarization can be transferred to the produced hadron which contains the initial quark thus lead to longitudinal polarization of hadron in the inclusive process @xmath2 .
measuring the polarizations of the produced hadrons , we can study the relation between the spin of the fragmenting quark and that of the produced hadron which contains that quark . in this aspect ,
the @xmath3 hyperons are most suitable candidates among all the hadrons .
this is because these hyperons decay weakly so that their polarizations can easily be obtained from the angular distributions of their decay products .
we now outline the calculation method of the longitudinal polarization @xmath4 of different hyperon @xmath5 in the inclusive process @xmath6 .
since the longitudinal polarization @xmath4 of the hyperon @xmath5 in the inclusive process @xmath6 originates from the longitudinal polarization @xmath7 of the initial quark @xmath8 ( where the subscript @xmath9 denotes its flavor ) produced at the annihilation vertex of the initial state @xmath0 , we should consider the @xmath5 s which have the following different origins separately .
\(a ) hyperons which are directly produced and contain the initial quarks @xmath10 s originated from the annihilations of the initial @xmath11 and @xmath12 ; \(b ) hyperons which are decay products of other heavier hyperons which were polarized before their decay ; \(c ) hyperons which are directly produced but do not contain any initial quark @xmath10 from @xmath0 annihilation ; \(d ) hyperons which are decay products of other heavier hyperons which were unpolarized before their decay .
it is clear that hyperons in groups ( a ) and ( b ) can be polarized .
those in group ( a ) are polarized @xcite since the polarization of the initial quark @xmath10 can be transferred to such hyperon @xmath5 s in the fragmentation process .
we denote the probability for this polarization transfer by @xmath13 and call it polarization transfer factor from quark @xmath14 to hyperon @xmath5 , where the superscript @xmath15 stands for fragmentation .
hyperons in group ( b ) can be polarized since such hyperons can inherit part of the polarization of the parent hyperons in the decay process .
we denote the probability for this polarization transfer by @xmath16 and call it polarization transfer factor from hyperon @xmath17 to hyperon @xmath5 , where the superscript @xmath18 stands for decay .
in contrast , hyperons in groups ( c ) and ( d ) are unpolarized .
hence , we obtain that the polarization of the final hyperon is , @xmath19 here @xmath7 is the polarization of the initial quark @xmath10 ; @xmath20 is the average number of the hyperons which are directly produced and contain the initial quark of flavor @xmath9 ; @xmath21 is the polarization of the hyperon @xmath17 before its decay ; @xmath22 is average numbers of @xmath5 hyperons coming from the decay of @xmath17 hyperons which are polarized ; @xmath23 ) , @xmath24 , @xmath25 and @xmath26 are average numbers of hyperons in group ( a ) ,
( b ) , ( c ) and ( d ) respectively .
these different factors can be calculated in the following way .
longitudinal polarization of the initial quark @xmath10 is determined by the standard model for electroweak interactions .
the polarization comes from the weak interactions and the results for @xmath27 can be found e.g. in [ [ ar80 ] ] , i.e. , @xmath28 where @xmath29 is the angle between the outgoing quark and the incoming electron , the subscript @xmath9 denotes the flavor of the quark , and @xmath30 @xmath31 @xmath32 @xmath33 where @xmath34 and @xmath35 are the mass and decay width of @xmath36 ; @xmath37
@xmath38 @xmath39 @xmath40 are the vector and axial vector coupling constants of electron and quark to @xmath36 boson , and @xmath41 is the weinberg angle . to see the size and the @xmath29-dependence of @xmath7 , we show in fig.1 the numerical results of @xmath42 as a function of @xmath43 at lep i energy .
we see clearly that all the quarks produced at the @xmath0 annihilation vertices are longitudinally with significantly high polarizations .
we see also that the magnitude of the polarization of the down - type quarks ( @xmath44 and @xmath45 ) is large and varies little with the angle @xmath29 , while that of the up - type quarks ( @xmath46 and @xmath47 ) is a little bit smaller and has a relatively larger variation in the whole range of @xmath43 . the latter increases with @xmath43 from 0.58 at @xmath48 to 0.74 at @xmath49 . averaging over @xmath29
, we obtain , @xmath50 where @xmath51 is the angular variation of the unpolarized cross section . in fig.2
, we show @xmath52 as a function of the total @xmath0 center - of - mass energy @xmath53 .
we see that , at the lep i energy , i.e. @xmath54 , the polarizations of quarks have the maximum negative value , i.e. , @xmath55 for @xmath56 , and @xmath57 for @xmath58 and @xmath45 . from the lep i energy to the lep ii energy , the polarizations decrease a bit but it is still very large at the lep ii energy ( @xmath59 ) , where we have @xmath60 for @xmath56 , and @xmath61 for @xmath58 and @xmath45 . at lep
ii energy , the @xmath62 events , i.e. @xmath63 , are significant .
it takes about 10% of the whole events .
the polarizations for the initial quarks in such events are different from those in @xmath64 , and should be considered separately in the calculations . according to the standard model for electroweak interactions ,
the polarization of initial quark and that of initial anti - quark created in @xmath65 or @xmath66 decay vertex are equal to -1.0 and 1.0 , respectively .
there exists also another type of events , i.e. @xmath67 , which contribute at lep ii energy . but
this contribution is very small ( less than 1% ) .
we will not consider such events in the following calculations .
the fragmentation polarization transfer factor @xmath13 from the initial quark @xmath14 to hyperon @xmath5 is equal to the fraction of spin carried by the @xmath9-flavor - quark divided by the average number of quark of flavor @xmath9 in the hyperon @xmath5@xcite .
this fractional contribution to the hyperon spin from @xmath9-flavor - quark is different in the above - mentioned su(6 ) or the dis picture .
the results in the su(6 ) picture can easily be obtained from the wave functions . in the dis picture , the fractional contribution of quarks of different flavors to the spin of a baryon in the @xmath68 octet is extracted from @xmath69 obtained in deep - inelastic lepton - proton scattering experiments@xcite and the constants @xmath15 and @xmath18 obtained from hyperon decay experiments .
the way of doing this extraction is now in fact quite standard ( see , for example , the appendix in [ [ bl98 ] ] ) . the results for @xmath1 , @xmath70 and @xmath71 hyperons have been obtained in [ [ jaffe96 ] ] and [ [ bl98 ] ] . using exactly the same way
, we obtain also the results for @xmath72 . for completeness , we list the results for all the @xmath3 hyperons in table 1 . in the calculations ,
we take also the contributions from the decay of @xmath73 , i.e. the decuplet hyperons into account .
the polarizations of such hyperons in the su(6 ) picture can easily be calculated using the su(6 ) symmetric wave functions .
but , it is presently impossible to calculate them in the dis picture since no dis data is available for any one of the decuplet baryons . therefore , in the calculation we take them into account in the same way as those in the su(6 ) picture .
the decay polarization transfer factor @xmath16 is determined by the decay process and is independent of the process in which @xmath17 is produced .
the results for different decay processes are different and will be given for each individual hyperon in the following sections of this paper . after we obtain the results for @xmath7 , @xmath13 and @xmath16 , we can calculate @xmath4 if we know the average numbers @xmath20 , @xmath22 , @xmath25 , and @xmath26 for hyperons @xmath5 from the different origins .
these average numbers are determined by the hadronization mechanism and should be independent of the polarization of the initial quarks . hence , we can calculate them using the a hadronization model which give a good description of the unpolarized data for multiparticle production in high energy reactions . presently , such calculations can only be carried out using a monte - carlo event generator .
we use the lund string fragmentation model@xcite implemented by jetset @xcite in the following . among all the @xmath3 hyperons , @xmath1 is most copiously produced .
furthermore , the spin structure of @xmath1 in the @xmath74 picture is very special , which makes it play a very special role in distinguishing the su(6 ) and the dis pictures . in the @xmath74 picture ,
spin of @xmath1 is completely carried by the @xmath75 valence quark , while the @xmath46 and @xmath76 quarks have no contribution . since the initial @xmath75 quark produced in the annihilation of the initial @xmath0 takes the maximum negative polarization , @xmath77 obtained using the su(6 ) picture is the maximum among all the different models . in contrast , in the dis picture , the @xmath75 quark carries only about @xmath78 of the @xmath1 spin , while the @xmath46 or @xmath76 quark each carries about @xmath79 ( see table 1 ) .
the resulting @xmath77 should be substantially smaller than that obtained in the @xmath74 picture . comparing the maximum with experimental results provide us a good test of the validity of the picture .
@xmath1 is also the lightest hyperon , so the final @xmath1 s produced in high energy reactions contain contributions from the decays of many different heavier hyperons .
there are three octet , i.e. @xmath70 , @xmath80 and @xmath81 , and five decuplet , i.e. , @xmath82 and @xmath83 , hyperons that can decay to @xmath1 .
the polarization transfer in @xmath84 has been studied in [ [ gatto58 ] ] , the result for the polarization transfer factor @xmath85 is @xmath86 .
@xmath87 is a parity non - conserving decay and is dominated by s - wave .
the polarization transfer @xmath88 for this process is equal to @xmath89 , where @xmath90 can be found in review of particle properties [ [ c98 ] ] .
the decay process from the decuplet hyperon to octet hyperon and a pseudo - scalar meson @xmath91 such as @xmath92 or @xmath93 is dominated by the @xmath94 wave@xcite , and the octet hyperon will get the same polarization as that of the initial decuplet hyperon , i.e. , @xmath95 . for explicity , we list all these results for @xmath96 in table.2 .
there are also some contributions from the decays of open charm or open beauty baryons .
the polarization transfer factors in these decay processes are unfortunately unknown yet .
this is a theoretical uncertainty in the calculations .
however , according to the materials in review of particle properties@xcite , we do know that each of these open charm or open beauty baryons can decay to @xmath1 through many different channels .
the contributions to @xmath1 polarization in these different channels may be quite different .
we expect that the net contribution from all these channels can not be large .
we will just neglect it in our calculations . using jetset and pythia
, we obtain the average numbers @xmath97 , @xmath98 , @xmath99 , and @xmath100 at @xmath101 gev and those at @xmath102200 gev , respectively .
we show them in fig.3 and fig.4 .
the results at @xmath101 gev are of course the same as those in [ [ gh93 ] ] and [ [ bl98 ] ] . from these figures , we see clearly that contribution to @xmath1 from the events originating from the initial @xmath75 quarks play the most important role , in particular for large @xmath103 . for example , for @xmath104 , it gives about 70% of the whole @xmath1 s , while those from the events initiated by @xmath46 or @xmath76 take only 10% respectively .
using the results for the @xmath7 s , the @xmath105 s , and the @xmath16 s mentioned above , we obtain the longitudinal polarization of @xmath1 as shown in fig.5 .
a comparison of those results at lep i with the aleph data @xcite and the opal data @xcite shows that the data@xcite of both groups agree better with the calculated results based on the @xmath74 picture .
but , these available data@xcite are still far from accurate and enormous enough to make a decisive conclusion .
further complementary measurements are needed . at the lep ii energy ,
@xmath62 events take about 10% of the whole events .
we expect that such events give a even more significant contribution to @xmath1 polarization , since the polarization of the initial quark at @xmath65 or @xmath66 decay vertex is 100% , which is larger than those from the @xmath106 .
since there are two initial quarks and two initial anti - quarks in @xmath62 events , the energy of each of them should be much smaller than those in the @xmath106 events .
hence , such events should contribute mainly in relatively smaller @xmath103 regions .
adding all the different together , we obtain @xcite @xmath107 at lep ii energy shown in fig.5 .
we see that , compared with those obtained at the lep i energy , @xmath77 at lep ii energy is a little larger at small @xmath103 region , and is a little smaller at large @xmath103 region .
this is consistent with the above - mentioned qualitative expectations and can be checked experimentally . in the two models we discussed above
, @xmath1 polarization comes solely from the hyperons which contain the initial quark created at the @xmath0 annihilation vertex .
the contributions are very much different for the hyperons which contain the initial @xmath75 from those which contain the initial @xmath46 or @xmath76 quarks . there is also a big differences between the contribution from the hyperons which contain the initial quarks and that from those which do not .
it is thus clear that we can get a further check of the two pictures if we study the @xmath1 polarization in events which originate from the initial @xmath46 or @xmath76 quark or those which originate from the initial @xmath75 quark separately .
we should also get a very sensitive check to different pictures if we study only those @xmath1 s which contain the initial quarks . in this connection
, it should also be mentioned that there exist another different type of models@xcite for spin transfer in fragmentation processes .
in contrast to that described in subsection [ subs : method ] , in these models , no distinction is made between the hyperons which contain the initial quarks or those which do not contain the initial quarks .
some of them@xcite even do not distinct those which are directly produced or those which come from heavier hyperon decays .
in other words , in these models , there is no distinction between hyperons from the groups ( a ) , ( b ) , ( c ) and ( d ) .
it is simply assumed that a `` reciprocity relation '' is valid between the fragmentation function for the final hyperons and the corresponding quark distribution functions in the hyperons .
they are taken as proportional to each other , with a proportional constant which is common for quarks of different flavors .
the hyperon polarization can easily be calculated in such models if the spin - dependent quark distribution functions are known .
the obtained results depend obviously very much on the spin - dependent quark distributions , which are used as theoretical input .
since the spin dependent quark distributions in hyperons are still poorly known yet , the obtained results can be very much different from each other if different sets of quark distributions are used .
but , since the ( polarized ) fragmentation function from a given quark @xmath8 to a given hyperon @xmath5 depends only on the fractional momentum @xmath103 of the quark carried by the produced hyperon , the obtained polarization at a given @xmath103 should be completely the same if we study only those hyperons which contain the initial quarks or all of them in events originating from a given type of initial quark .
it is unfortunately impossible to make a complete separation of hyperons in these different groups in experiments .
however , it should be possible to separate the events into different subsamples , in each of which the hyperons from one group dominate . measuring @xmath1 polarization in such subsamples of events should give further complementary checks of the different models mentioned above . using the monte - carlo event generator jetset @xcite , we can study the various possibilities in this direction .
we note that , if a @xmath1 originates from the initial @xmath75 quark , the leading particle in the opposite direction should contain the @xmath108 produced in the @xmath0 annihilation vertex .
hence , we choose an event sample according to the following criteria : \(i ) @xmath1 is the leading in one direction ; \(ii ) the leading particle in the opposite direction is @xmath109 .
we expect that such @xmath1 s should mainly have the origin ( a ) mentioned in the subsection [ subs : method ] . in fig.6 , we show the results obtained from jetset for the average numbers of @xmath1 s of different origins .
we see that such leading @xmath1 s indeed mainly originate from the initial @xmath75 quark .
the contribution from the initial @xmath46 or @xmath76 quark is indeed substantially small .
it takes only about 3% of such @xmath1 s .
in particular , in @xmath110 region there is almost no contribution from the initial @xmath46 or @xmath76 quark at all . using eq.(1 ) , we calculated @xmath107 for the events under the conditions ( i ) and ( ii ) using the su(6 ) and the dis pictures .
the comparison of the obtained results with those obtained in the average events are shown in fig.7 .
we see that there is indeed a significant difference between the results obtained for such particularly chosen events and those for the average events in particular in the region of @xmath111 .
checking such differences by measuring @xmath107 for such events can be helpful in distinguishing the validity of different pictures for the spin transfer in fragmentation processes .
the production rates for other octet hyperons are smaller than that for @xmath1 so the statistic errors should be larger for the polarizations of these hyperons . on the other hand , decay contributions from heavier hyperons to these hyperons are also much less significant than that in case of @xmath1 .
hence , the contaminations from the decay processes are much smaller . these conclusions can easily be checked using a monte - carlo event generator for @xmath0 annihilation into hadrons . in fig.8
, we show the results obtained from jetset for these hyperons compared with those for @xmath1 .
we see that , the production rate for @xmath112 or @xmath80 is less than @xmath113 of that for @xmath1 . on the other hand ,
we see also that the contribution from heavier hyperon decays is also much smaller . for example , for @xmath112 s , the decay contribution takes only about @xmath114 of the total rate .
the situations for @xmath115 , @xmath80 , and @xmath81 are similar to that for @xmath112 .
most of them are directly produced .
hence , the theoretical uncertainties in the calculations for these hyperons are much smaller . the study of polarizations of these hyperons should provide us with good complementary tests of different pictures . in this section
, we calculate the longitudinal polarizations of these hyperons , i.e. , @xmath112 , @xmath115 , @xmath80 and @xmath81 , in @xmath0 annihilation at lep i and lep ii energies .
the calculations are quite similar to those for @xmath1 production .
the general procedure is the same and has been outlined in last section .
the polarization transfer factor @xmath13 are given in table 1 .
the situation for decay contributions is quite simple : there are only contribution from the corresponding decuplet hyperon decays .
more precisely , @xmath116 contains only decay contribution from @xmath117 and @xmath71 has the decay contribution from @xmath118 .
both decay processes are strong decays and are dominated by the p - wave .
the polarization transfer is the same as that in @xmath119 , i.e. @xmath120 .
first , although the spin structure of @xmath112 and that of @xmath115 or that of @xmath80 and that of @xmath81 are symmetric under the exchange of @xmath46 and @xmath76 ( c.f .
table 1 ) , their polarizations should be quite different from each other .
this is because the polarization of the initial @xmath46 and that of the initial @xmath76 produced at the @xmath0 annihilation vertices are quite different .
this can be seen clearly in fig.1 , where we see that the magnitude of the polarization of @xmath46 quark is smaller than that of @xmath76 quark .
second , from table 1 , we see that the contributions from the two different flavors in these hyperons ( @xmath121 , @xmath122 ) are quite different : they have different signs and different magnitudes , and the differences are also different in the su(6 ) or the dis picture .
this implies that their contributions to hyperon polarizations are also quite different .
they should have different signs and different magnitudes .
these differences make the situation very interesting .
we should have different expectations for different hyperons .
even the signs of polarizations of some hyperons in the two different pictures can be different from each other . using jetset and pythia , we calculated the different contributions to @xmath112 , @xmath115 , @xmath80 and @xmath81 from all the different sources discussed above at lep i and lep ii , respectively .
the results at lep i energy are shown in fig.9 .
those at lep ii energy are similar and are not shown . from this fig.9 , we see that the @xmath112 s mainly come from the initial @xmath75 and the initial @xmath46-quark , and that the @xmath115 s come mainly from the initial @xmath75 and the initial @xmath76-quark , and that these two contributions are comparable with each other in particular for @xmath115 s .
in contrast , the source for @xmath122 is very pure : they come predominately from the initial @xmath75-quarks .
all the others are negligible . using eq.(1 ) and
the results shown in fig.9 , we calculated the hyperon polarizations , @xmath123 , @xmath124 , @xmath125 and @xmath126 .
the obtained results are shown in fig.10 . from these results , we see clearly that the polarizations are different for these different hyperons , and that there are considerably large differences between the results using the @xmath74 picture and those using the dis picture . in particular , we see that @xmath123 in the @xmath74 picture has opposite sign to that in the dis picture .
but their magnitudes are relatively small .
the magnitude of @xmath125 and that of @xmath126 are considerably larger , in particular in the large @xmath103 region .
this is because the contribution to @xmath80 or @xmath81 is very clear .
they come mainly from the initial @xmath75-quark .
there is thus little theoretical uncertainty in calculating @xmath127 .
hence , although the statistics may be much worse than @xmath1 , there are still good reasons to study @xmath127 in @xmath128 at lep energies .
we thank gsta gustafson , li shi - yuan , wang qun , xie qu - bing and other members in the theoretical particle physics group of shandong university for helpful discussions .
this work was supported in part by the national science foundation of china ( nsfc ) and the education ministry of china .
99 see , for example , refs.[[jaffe91]-[ma2000 ] ] and the references given there .
jaffe , and ji xiangdong , phys .
lett . * 67 * , 552 ( 1991 ) ; nucl . phys . *
b375 * , 527 ( 1992 ) .
[ jaffe91 ] m. burkardt and r.l .
jaffe , phys .
70 , 2537 ( 1993 ) .
[ bj93 ] g.gustafson and j.hkkinen , phys . lett . *
b303 * , 350 ( 1993 ) .
[ gh93 ] r.l .
jaffe , phys . rev . *
d54 * , r6581 ( 1996 ) .
[ jaffe96 ] c. boros , and liang zuo - tang , phys .
* d57 * , 4491 ( 1998 ) .
[ bl98 ] a. kotzinian , a. bravar , d. von harrach , eur .
j. * c2*,329 - 337(1998 ) .
[ kotz98 ] b.q .
ma , i. schmidt , and j.j .
yang , phys .
rev . d*61 * , 034017 ( 2000 ) . [ ma2000 ] for a review of data , see e.g. , g.k .
mallot , in proc . of the 12th inter .
symp . on spin phys . ,
amsterdam 1996 , edited by de jager _
et al_. , world scientific ( 1997 ) , p.44 .
[ spin97 ] aleph - collaboration ; d. buskulic et al .
, phys . lett .
* b 374 * ( 1996 ) 319 .
[ aleph96 ] j.e .
augustin and f.m .
renard , nucl . phys . *
b162 * , 341 ( 1980 ) .
[ ar80 ] b. anderson , g. gustafson , g. ingelman , and t. sjstrand , phys . rep . * 97 * , 31 ( 1983 ) . [ agis83 ] t. sjstrand , comp .
. comm . * 39 * , 347 ( 1986 ) .
[ sjo86 ] r. gatto , phys . rev . *
109 * , 610 ( 1958 ) .
[ gatto58 ] particle data group , c. caso _ et al . , _ euro .
j. c*3 * , 1 ( 1998 ) .
[ c98 ] opal - collaboration ; euro .
j. * c2 * , 49 - 59 ( 1998 ) . [ opal98 ] f.e .
close and r. g. roberts , phys . lett .
* b316 * , 165 ( 1993 ) .
[ cr93 ] it should be mentioned that , at the lep ii energy , the magnitude of the polarization of the initial quark produced in @xmath64 can be influenced by the initial state radiation .
such initial state radiation reduces the effective energy of the @xmath0 at the annihilation vertex thus the polarization of @xmath7 in such events should take the value at the corresponding lower energy .
but , since @xmath7 varies only slowly with the energy , this effect can not be large .
we neglect it in the calculations .
.fractional contributions @xmath129 , @xmath130 and @xmath131 of the light flavors to the spin of baryons in the @xmath3 octet calculated using the @xmath74 picture and those obtained using the data for deep inelastic lepton - nucleon scattering and those for hyperon decay under the assumption that su(3 ) flavor symmetry is valid .
the first column shows the obtained expressions in terms of @xmath116 , @xmath15 and @xmath18 .
the @xmath74 picture results are obtained by inserting @xmath132 and @xmath133 into these expressions and those in the third column are obtained by inserting @xmath134 , obtained from deep inelastic lepton - nucleon scattering experiments [ [ spin97 ] ] , and @xmath135 obtained [ [ c98],[cr93 ] ] from the hyperon decay experiments . [
cols="<,^,^,^,^,^,^ " , ]
|
longitudinal polarizations of different kinds of hyperons produced in @xmath0 annihilation at lep i and lep ii energies in different event samples are calculated using two different pictures for the spin structure of hyperon : that drawn from polarized deep inelastic lepton - nucleon scattering data or that using su(6 ) symmetric wave functions .
the result shows that measurements of such polarizations should provide useful information to the question of which picture is more suitable in describing the spin effects in the fragmentation processes .
|
despite their apparently separated application areas , general relativity and quantum information are not disjoint research fields . on the contrary , following the pioneering work of alsing and milburn @xcite a wealth of works
@xcite has considered different situations in which entanglement was studied in a general relativistic setting , for instance , quantum information tasks in the proximity of black holes @xcite , entanglement in an expanding universe @xcite , entanglement with non - inertial partners @xcite etc .
entanglement behavior in non - inertial frames was first considered in @xcite where the fidelity of teleportation between relative accelerated partners was analyzed .
after this , occupation number entanglement degradation of scalar @xcite and dirac @xcite fields due to unruh effect @xcite was shown .
recent works studied the effect of the instantaneous wigner rotations and thomas spin precession on entanglement @xcite,@xcite .
the previous work @xcite on unruh effect for dirac field mode entanglement does not consider the spin of the parties .
hence , only two occupation numbers @xmath3 are allowed for each mode .
higher values of @xmath4 are forbidden by pauli exclusion principle . however , addressing the effect of unruh decoherence on spin entanglement can only be done by incorporating the spin of the parties in the framework from the very beginning . as a consequence ,
occupation number @xmath5 is also allowed .
this fact will affect occupation number entanglement which has to be reconsidered in this new setting .
for this purpose , in this work we will study the case of two parties ( alice and rob ) sharing a general superposition of dirac vacuum and all the possible one particle spin states for both alice and rob .
alice is in an inertial frame while rob undergoes a constant acceleration @xmath6 .
we will show that rob when he is accelerated respect to an inertial observer of the dirac vacuum would observe a thermal distribution of fermionic spin @xmath7 particles due to unruh effect @xcite .
next , we will consider that alice and rob share spin bell states in a minkowski frame .
then , the case in which alice and rob share a superposition of the dirac vacuum and a specific one particle state in a maximally entangled combination . in both cases
we analyze the entanglement and mutual information in terms of rob s acceleration @xmath6 .
finally , we will study the case when the information about spin is erased from our setting by partial tracing , remaining only the occupation number information . here
, entanglement is more degraded than in @xcite .
this comes about because more accessible levels of occupation number are allowed , so our system has a broader margin to become degraded .
this paper is structured in the following sections . in sections
[ sec2 ] , and [ sec3 ] we introduce the basic formalism and notation to deal with dirac fields from the point of view of an accelerated observer taking its spin structure into account . in section [ sec4 ]
we study how the minkowski vacuum state is expressed by an accelerated observer when the spin of each mode is included in the setting , discussing the implications of the single - mode approximation often carried out in the literature @xcite . also , we build the one particle state with spin @xmath8 in rindler coordinates and analyze the unruh effect when the spin structure is included . here
we discuss the necessity of tracing over rindler s region iv for rob , since it is causally disconnected from region i in which we consider rob s location . in section [ sec5 ]
we analyze how entanglement of spin bell states is degraded due to unruh effect .
we show that , even in the limit of @xmath9 , some degree of entanglement is preserved due to pauli exclusion principle .
then we analyze unruh effect on a completely different class of maximally entangled states ( like @xmath10 where @xmath8 and @xmath11 are @xmath12 component of spin labels ) comparing it with the spin bell states . in section [ sec6 ]
we show that the erasure of spin information , in order to investigate occupation number entanglement alone , requires considering total spin states for the bipartite system .
finally , our results and conclusions are summarized in section [ sec7 ] .
we consider a free dirac field in a minkowski frame expanded in terms of the positive ( particle ) and the negative ( antiparticle ) energy solutions of dirac equation notated @xmath13 and @xmath14 respectively : @xmath15 here , the subscript @xmath16 notates momentum which labels the modes of the same energy and @xmath17 is the spin label that indicates spin - up or spin - down along the quantization axis . @xmath18 and @xmath19 are respectively the annihilation operators for particles and antiparticles , and satisfy the usual anticommutation relations . for each mode of frequency @xmath16 and spin @xmath8 the positive and negative energy modes have the form @xmath20 where @xmath21 is a spinor satisfying the normalization relations @xmath22 .
the modes are classified as particle or antiparticle respect to @xmath23 ( minkowski killing vector directed to the future ) .
the minkowski vacuum state is defined by the tensor product of each frequency mode vacuum @xmath24 such that it is annihilated by @xmath18 and @xmath19 for all values of @xmath8 .
we will use the same notation as reference @xcite where the mode label will be a subscript inside the ket , and the absence of subscript outside the ket indicates a minkowski fock state . in this way , and as a difference with previous works , we will consider the spin structure for each mode , and hence , the maximum occupation number is two .
this introduces the following notation @xmath25 if @xmath26 the two particles state is not allowed due to pauli exclusion principle , so our allowed minkowski states for each mode of particle / antiparticle are @xmath27 consider that we have the following minkowski bipartite state @xmath28 with @xmath29 .
the subscripts @xmath30 indicate the modes associated with alice and rob respectively .
all other modes of the field are unoccupied that is to say that the complete state would be @xmath31$]. this state generalizes the bell spin states ( for instance , we have @xmath32 choosing @xmath33 ) or a modes entangled state ( for instance choosing @xmath34 ) . with this state
we will be able to deal with two different and interesting problems at once , 1 .
studying the unruh decoherence of spin entangled states and 2 . investigating the impact of considering the spin structure of the fermion on the occupation number entanglement and its unruh decoherence . later on , under the single mode approximation , we will assume that alice is stationary and has a detector sensitive only to the mode @xmath35 , and rob moves with uniform acceleration @xmath6 taking with him a detector sensitive to the mode @xmath36 .
an uniformly accelerated observer viewpoint is described by means of the well - known rindler coordinates @xcite . in order to cover the whole minkowski space - time , two different set of coordinates are necessary .
these sets of coordinates define two causally disconnected regions in rindler space - time .
if we consider that the uniform acceleration @xmath6 lies on the @xmath12 axis , the new rindler coordinates @xmath37 as a function of minkowski coordinates @xmath38 are @xmath39 for region i , and @xmath40 for region iv .
[ fig1 ] are hyperbolae and all the curves of constant proper time @xmath41 for the accelerated observer are straight lines that come from the origin .
an uniformly accelerated observer rob travels along a hyperbola constrained to region i , title="fig : " ] as we can see from fig .
1 , although we have covered the whole minkowski space - time with these sets of coordinates , there are two more regions labeled ii and iii .
to map them we would need to switch @xmath42 in equations , . in these regions @xmath41 is a spacelike coordinate and @xmath12 is a timelike coordinate .
however , the solutions of dirac equation in such regions are not required to discuss entanglement between alice and an accelerated observer since he would be constrained to either region i or iv , having no possible access to the opposite regions as they are causally disconnected @xcite .
the rindler coordinates @xmath43 go from @xmath44 to @xmath45 independently in regions i and iv .
it means that each region admits a separate quantization procedure with their corresponding positive and negative energy solutions @xmath46 and @xmath47 .
particles and antiparticles will be classified with respect to the future - directed timelike killing vector in each region . in region
i the future - directed killing vector is @xmath48 whereas in region iv the future - directed killing vector is @xmath49 .
this means that solutions in region i , having time dependence @xmath50 with @xmath51 , represent positive energy solutions , whereas solutions in region iv , having time dependence @xmath50 with @xmath51 , are actually negative energy solutions since @xmath52 points to the opposite direction of @xmath53 @xcite . as i and iv are causally disconnected @xmath54 and @xmath55 only have support in their own regions , vanishing outside them .
let us denote @xmath56 the particle annihilation and creation operators in region i and @xmath57 the corresponding antiparticle operators .
analogously we define @xmath58 the particle / antiparticle operators in region iv .
these operators satisfy the usual anticommutation relations @xmath59 where the subscript r notates the rindler region of the operator @xmath60 .
all other anticommutators are zero . that includes the anticommutators between operators in different regions of the rindler space - time .
taking this into account we can expand the dirac field in rindler coordinates analogously to : @xmath61 equations and represent the decomposition of the dirac field in its modes in minkowski and rindler coordinates respectively .
we can relate minkowski and rindler creation and annihilation operators by taking appropriate inner products @xcite .
the relationship between minkowski and rindler particle / antiparticle operators is linear and the coefficients that relate them are called bogoliubov coefficients : @xmath62 where @xmath63 and @xmath64 is a phase factor that will turn out to be irrelevant for our purposes .
notice that as we are working with two spatial - temporal dimensions and with massless dirac field , the relation between rindler modes and minkowski modes is given in .
we will discuss in the conclusions the implications of considering extra dimensions and massive fields , where minkowski modes are spread over all positive rindler frequencies @xcite .
notice from bogoliubov transformations that the minkowski particle annihilator @xmath18 transforms into a rindler particle annihilator of momentum @xmath16 and spin @xmath8 in region i and an antiparticle creator of momentum @xmath65 and spin @xmath66 in region iv ( in region iv all the magnitudes that are not invariant under time reversal change ) .
now that we have the relationships between the creation and annihilation operators in minkowski and rindler coordinates , we can obtain the expression of the minkowski vacuum state for each mode @xmath67 in rindler coordinates . for notation simplicity
, we will drop the @xmath16 label in operators / states when it does not give any relevant information , but we will continue writing the spin label .
it is useful to introduce some notation for our states .
we will denote with a subscript outside the kets if the mode state is referred to region i or iv of the rindler space - time .
the absence of subscript outside the ket will denote minkowski coordinates .
the @xmath68 label of particle / antiparticle will be omitted throughout the paper because , for the cases considered , a ket referred to minkowski space - time or rindler s region i will always denote particle states and a ket referred to region iv will always notate antiparticle states . inside the ket
we will write the spin state of the modes as follows @xmath69 which will notate a particle state in region i and an antiparticle state in region iv respectively , both with spin @xmath8 . we will use the following definitions for our kets @xmath70 and , being consistent with the different rindler regions operators anticommutation relations , @xmath71 @xmath72 now , it is useful to note that could be expressed as two - modes squeezing transformation for each @xmath16 @xcite @xmath73 where @xmath74\ ] ] so , analogously to @xcite , it is reasonable to postulate that the minkowski vacuum is a rindler two - mode particles / antiparticles squeezed state with opposite spin and momentum states in i and iv .
contrarily to @xcite , considering that the modes have spin , occupation number for each @xmath16 is allowed to be 2 , being higher occupation numbers forbidden by pauli exclusion principle . in the literature
the analysis is restricted only to one mode of the minkowski field , but we can restrict our analysis to some sector of the minkowski vacuum , defining for the particles sector @xmath75 such that the particle sector of can be written as in this fashion we are considering a discrete number @xmath4 of different modes @xmath76 , so minkowski vacuum should be expressed as a squeezed state in rindler coordinates which is an arbitrary superposition of spins and momenta .
this will be useful to discuss what would happen if we relax the single - mode approximation carried out often in the literature and let our detectors have a small mode spread .
@xmath77 where , here , the notation is @xmath78 with @xmath79 being @xmath80 and the symbol @xmath81 is @xmath82 1\ ; \text{otherwise } \end{array}\right.\ ] ] which imposes pauli exclusion principle constraints on the state ( quantum numbers of fermions can not coincide ) .
notice a pair of aspects of this notation for the multimode case .
first , in the series in all the possible orders of the operators are implicitly written .
due to the anticommutation relations of the fermionic operators , terms with different orderings of the creation operators are related , i.e. @xmath83
so , without loss of generality , we could choose not to write all the possible orderings of the operators in .
the difference between taking all the possible orderings of the operators into account and taking only one representant is a factor @xmath84 in the constants @xmath85 . from we
can also see that the coefficients @xmath85 are symmetric with respect to @xmath86 index permutations .
second , as there are only @xmath4 different modes @xmath87 , the last summation in equation has only @xmath88 terms due to .
these terms are all the different permutations of the creation operators for pairs of opposed spins for each mode
. there would be only one summand instead of @xmath88in the simplified notation where we do not write all the different permutations of the operators but only one representant .
it means that , in this simplified notation , the series of terms with @xmath4 pairs has the same summands as the series with the vacuum state ( i.e. only one ) . actually , in this notation
i.e . if we count all the different order permutations as only one the series with @xmath89 to @xmath90 has exactly the same number of summands as the series with @xmath91 to @xmath92 . to obtain restrictions on the values of the coefficients @xmath85 we demand that the minkowski vacuum has to be annihilated by the particle annihilator , @xmath93 . translating this into rindler coordinates we have @xmath94{\left| { 0 } \right\rangle}=0\ ] ] where the vacuum state should be expressed in rindler coordinates using .
as the elements form an orthogonal set , from we see that all the terms proportional to different elements of the set should be zero simultaneously , which gives the following conditions on the coefficients * @xmath95 as a function of @xmath96 + @xmath97 since equations , should be satisfied @xmath98 , we obtain that @xmath99 since @xmath96 does not depend on @xmath16 or @xmath8 .
we will denote @xmath100 . * @xmath101 as a function of @xmath102 + @xmath103 since equations , should be satisfied @xmath98 , we obtain that @xmath104 where @xmath105 does not depend on spins or momenta since @xmath102 does not depend on @xmath16 or @xmath8 , the only dependence of the coefficients with @xmath106 and @xmath107 is given by the pauli exclusion principle , this dependence comes through function .
in fact it is very easy to show that all the coefficients are independent of @xmath107 and @xmath106 apart from the pauli exclusion principle constraint. using the fact that @xmath96 does not depend on @xmath107 and @xmath106 and noticing that by applying the annihilator on the vacuum state and equalling it to zero , we will always obtain the linear relationship between @xmath108 and @xmath92 given below . * @xmath85 as a function of @xmath109 + @xmath110 we finally obtain that @xmath85 is a constant which can be expressed as a function of @xmath96 as @xmath111 and then , vacuum state can be expressed as @xmath112 where the only parameter not fixed yet is @xmath96 .
we can now fix @xmath96 by imposing the normalization of the minkowski vacuum in rindler coordinates @xmath113 , @xcite this condition can be written as @xmath114=1\ ] ] where @xmath115 and we have defined @xmath116 .
this expression gives ( formally ) the value of @xmath96 ( except for a global phase ) when considering the populated levels in an arbitrary number of modes of the field @xmath117^{-1/2}\ ] ] we can see that if we take the limit @xmath118 we recover the minkowski vacuum .
we will come back to the behavior at the limits below .
notice that the state is only normalizable in this discrete limit .
this comes about because the minkowski and rindler representations are not unitary equivalent ( there is not unitary operator connecting the two vacua ) .
it prevents us from taking the continuous limit in the expression but does not invalidate the treatment as equation can be considered as a superposition of a finite number of individual modes which are perfectly well defined @xcite . to address the problem of showing how the presence of spin degrees of freedom affects the entanglement between accelerated observers , it is useful to use the single mode approximation @xcite in the same way that in @xcite .
this is valid if we consider rob s detector so sensitive to a single particle mode in region i that we can approximate the frequency @xmath119 observed by alice to be the same frequency observed by rob , @xmath120 . as a consequence ,
the populated levels we are looking at are in this single frequency @xcite ( see also discussion in @xcite ) so we can consider the sums over @xmath106 in just like a sum of only one mode @xmath121 .
this is equivalent to restricting the analysis to the sector @xmath67 of the complete vacuum .
since the goal of this work is to show the effect of spin degrees of freedom on the entanglement for non - inertial observers , this approximation allows us to compare our results with previous literature on scalar and spinless fermion fields @xcite .
we have to notice that since the observer rob is accelerated , his possible measurements are affected by a doppler - like effect . given that the velocity of the observer is @xmath122 , the doppler effect will shift the sensitivity peak of the detector .
namely , if at the instant @xmath123 his detector is sharply tuned to a frequency @xmath124 , to compensate this doppler effect at some instant @xmath125 the detector should be tuned to the frequency @xmath126 .
this implies that any detector will eventually become insensitive to the populated levels of the minkowski field . in order to do the theoretical analysis in this work
, we can consider either that his detector can be sharply tuned to the frequency @xmath127 for each instant , or that rob has a set of individual detectors each one sharply tuned to the proper frequency for each instant .
carrying out this single - mode approximation , the minkowski vacuum for a single mode is a rindler two - mode particles / antiparticles squeezed state with opposite spin states in i and iv . considering that the modes have spin ,
occupation number is allowed to be 2 for each @xmath16 , being higher occupation numbers forbidden by pauli exclusion principle . as a consequence @xmath128 notice that @xmath129 is the analogous to @xmath96 , @xmath130 and @xmath131 are analogous to @xmath132 and @xmath133 respectively and @xmath134 is analogous to @xmath105 in the expression but considering only one representant of all the 2 possible orders for the pair . to obtain the values of the coefficients @xmath135 we demand that the minkowski vacuum has to be annihilated by the particle annihilator , @xmath136 . translating this into rindler coordinates we have @xmath137\left[v{\left| { 0 } \right\rangle_{i}\left| { 0 } \right\rangle_{iv}}\right.\\ * & + & \left.a{\left| { \uparrow } \right\rangle_{i}\left| { \downarrow } \right\rangle_{iv}}+b{\left| { \downarrow } \right\rangle_{i}\left| { \uparrow } \right\rangle_{iv}}+ c{\left| { { \uparrow\downarrow } } \right\rangle_{i}\left| { { \uparrow\downarrow } } \right\rangle_{iv}}\right]\end{aligned}\ ] ] which implies @xmath138\\ * \nonumber&&-e^{i\phi}\sin r\left[v{\left| { 0 } \right\rangle_{i}\left| { -s } \right\rangle_{iv}}-a\delta_{s\downarrow}{\left| { \uparrow } \right\rangle_{i}\left| { { \uparrow\downarrow } } \right\rangle_{iv}}\right.\\ * & & \left.+b\delta_{s\uparrow}{\left| { \downarrow } \right\rangle_{i}\left| { { \uparrow\downarrow } } \right\rangle_{iv}}\right].\end{aligned}\ ] ] this equation gives 4 conditions ( two for each value of @xmath8 ) ,
although only 3 of them are independent @xmath139 to fix @xmath129 we impose the normalization relation for each field mode @xmath140 , imposing this we finally obtain the values of the vacuum coefficients .
@xmath141 notice that comparing this result with expressions and , as we have truncated the series in , the value of @xmath129 will be different from the case when more than one mode is considered @xmath96 instead of @xmath129. if we restrict the series on @xmath142 to only one mode @xmath143 in , we obtain that @xmath144 and we get then proper values for @xmath145 and @xmath146 . since from @xmath147
, comparing @xmath129 with , we can see that , while under the single - mode approximation the limit of infinite acceleration leads to a finite distribution of the minkowski vacuum over rindler states , when considering the multimode rindler expression for the vacuum state the combined limit of @xmath148 and @xmath149 leads to a complete fading away of the amplitudes over all the rindler modes as @xmath150 , which may not be the case for finite @xmath6 .
this is beyond the scope of this article but we will discuss in the conclusion that it may have very strong implications on the entanglement of fermionic fields for accelerated observers .
so finally , under the single mode approximation , the minkowski vacuum state in rindler coordinates is as follows @xmath151 now we have to build the one particle ( of spin @xmath8 ) state in rindler coordinates .
it can be readily done by applying the minkowski particle creation operator to the vacuum state @xmath152 , and translating it into rindler coordinates : @xmath153\left[\cos^2 r{\left| { 0 } \right\rangle_{i}\left| { 0 } \right\rangle_{iv}}\right.\\ * \nonumber&&+e^{i\phi}\sin r\,\cos r\left({\left| { \uparrow } \right\rangle_{i}\left| { \downarrow } \right\rangle_{iv}}+{\left| { \downarrow } \right\rangle_{i}\left| { \uparrow } \right\rangle_{iv}}\right)\\ * & & \left.+e^{2i\phi}\sin^2 r\,{\left| { { \uparrow\downarrow } } \right\rangle_{i}\left| { { \uparrow\downarrow } } \right\rangle_{iv}}\right]\end{aligned}\ ] ] that means @xmath154 the three minkowski states @xmath155 correspond to the particle field of mode @xmath16 observed by alice . however , since rob is experiencing a uniform acceleration he will not be able to access to field modes in the causally disconnected region iv , hence , rob must trace over that inaccessible region as it is unobservable . specifically , when rob is in region i of rindler space - time and alice observes the vacuum state , rob could only observe a non - pure partial state given by @xmath156 that is @xmath157 but while alice would observe the vacuum state of mode @xmath16 , rob would observe certain statistical distribution of particles .
the expected value of rob s number operator on the minkowski vacuum state is given by @xmath158\end{aligned}\ ] ] substituting the expression we obtain @xmath159 using we obtain that @xmath160 where @xmath161 is the boltzmann s constant and @xmath162 is the unruh temperature . equation is known as the unruh effect @xcite , which shows that , for a two - dimensional space - time , an uniformly accelerated observer in region i detects a thermal fermi - dirac distribution when he observes the minkowski vacuum .
we obtain a factor 2 contrarily to ref .
@xcite due to the degeneracy factor @xmath163 .
in previous works @xcite it was studied how unruh decoherence affects occupation number entanglement in bipartite states as @xmath164where the figures inside the kets represent occupation number of alice and rob modes respectively , barring any reference to the spin of the field modes . here , where we have included the spin structure of each mode in our setting from the very beginning , it is possible to study the effects of acceleration in spin entanglement decoherence , which is different from the mere occupation number entanglement .
first of all , we build a general bipartite state that could be somehow analogous to state studied in @xcite , limiting the occupation number to 1 but including the spins of each mode .
@xmath165 with @xmath29 .
the subscripts @xmath30 indicate the modes associated with alice and rob respectively .
we will suppress the labels @xmath30 from now on , and we will understand that the first character in a ket or a bra corresponds to alice and the second to rob : @xmath166 this general setting allows us to study in this section what happens with spin entanglement under acceleration of rob and also what happens with the occupation number entanglement when considering sates analogous to but taking the spin structure into account
. it will also allow us to discuss , in section [ sec6 ] , the implications of tracing over spins and study only the entanglement on the occupation number degree of freedom compared with @xcite .
the density matrix in minkowski coordinates for the state is @xmath167 where n.d.h.c means non - diagonal hermitian conjugate , and represents the hermitian conjugate only for the non - diagonal elements .
computing the density matrix , taking into account that rob is constrained to region i of rindler space - time , requires to rewrite rob s mode in terms of rindler modes and to trace over the unobservable rindler s region iv . in appendix [ appen ]
we compute each term of in rindler s coordinates and trace over the unobserved region iv . using ,
, we can easily compute the density matrix for alice and rob from since @xmath168 , resulting in the long expression @xmath169+\mu\cos^3r\\ * \nonumber&&\times\big[\alpha^*{\left| { 0,0 } \right\rangle\!\left\langle { \uparrow,\uparrow } \right|}+\beta^*{\left| { 0,0 } \right\rangle\!\left\langle { \uparrow,\downarrow } \right|}+\gamma^*{\left| { 0,0 } \right\rangle\!\left\langle { \downarrow,\uparrow } \right|}\\ * \nonumber&&+\delta^*{\left| { 0,0 } \right\rangle\!\left\langle { \downarrow,\downarrow } \right|}\big]+\mu\sin^2r\,\cos r\big[\alpha^*{\left| { 0,\downarrow } \right\rangle\!\left\langle { \uparrow,{\uparrow\downarrow } } \right|}\\ * \nonumber&&-\beta^*\!{\left| { 0,\uparrow } \right\rangle\!\left\langle { \uparrow,{\uparrow\downarrow } } \right|}\!+\!\gamma^*\!{\left| { 0,\downarrow } \right\rangle\!\left\langle { \downarrow,{\uparrow\downarrow } } \right|}\!-\!\delta^*\!{\left| { 0,\uparrow } \right\rangle\!\left\langle { \downarrow,{\uparrow\downarrow } } \right|}\big]\\ * \nonumber&&+\cos^2 r\big[|\alpha|^2{\left| { \uparrow,\uparrow } \right\rangle\!\left\langle { \uparrow,\uparrow } \right|}+\alpha\beta^*{\left| { \uparrow,\uparrow } \right\rangle\!\left\langle { \uparrow,\downarrow } \right|}+\alpha\gamma^*\\ * \nonumber&&\times{\left| { \uparrow,\uparrow } \right\rangle\!\left\langle { \downarrow,\uparrow } \right|}+\alpha\delta^*{\left| { \uparrow,\uparrow } \right\rangle\!\left\langle { \downarrow,\downarrow } \right|}+|\beta|^2{\left| { \uparrow,\downarrow } \right\rangle\!\left\langle { \uparrow,\downarrow } \right|}\\ * \nonumber&&+\beta\gamma^*{\left| { \uparrow,\downarrow } \right\rangle\!\left\langle { \downarrow,\uparrow } \right|}+\beta\delta^*{\left| { \uparrow,\downarrow } \right\rangle\!\left\langle { \downarrow,\downarrow } \right|}+|\gamma|^2{\left| { \downarrow,\uparrow } \right\rangle\!\left\langle { \downarrow,\uparrow } \right|}\\ * \nonumber&&+\gamma\delta^*{\left| { \downarrow,\uparrow } \right\rangle\!\left\langle { \downarrow,\downarrow } \right|}+|\delta|^2{\left| { \downarrow,\downarrow } \right\rangle\!\left\langle { \downarrow,\downarrow } \right|}\big]+\sin^2r\\ * \nonumber&&\times\big[\left(|\alpha|^2+|\beta|^2\right){\left| { \uparrow,{\uparrow\downarrow } } \right\rangle\!\left\langle { \uparrow,{\uparrow\downarrow } } \right|}+\left(|\gamma|^2+|\delta|^2\right)\\ * \nonumber&&\times{\left| { \downarrow,{\uparrow\downarrow } } \right\rangle\!\left\langle { \downarrow,{\uparrow\downarrow } } \right|}+\left(\alpha\gamma^*+\beta\delta^*\right){\left| { \uparrow,{\uparrow\downarrow } } \right\rangle\!\left\langle { \downarrow,{\uparrow\downarrow } } \right|}\big]\\ * & & + \text{n.d.h.c.}\end{aligned}\ ] ] here the notation is the same than in the r.h.s . of : @xmath170 . notice that the state , which in minkowski coordinates is pure , gets mixed when the observer rob is accelerated . equation will be our starting point , from which we will study different entanglement settings and how unruh decoherence affects them . to begin with we
will compute how acceleration affects the entanglement of spin bell states when alice and rob share a maximally entangled spin state and rob accelerates . in minkowski coordinates that means choosing specific coefficients in , particularly , for bell states we should choose @xmath171 and the rest of the other coefficients equal to zero .
for such states in minkowski coordinates , the density matrix of alice and rob considering that rob undergoes an acceleration @xmath6 is obtained from : @xmath172\end{aligned}\ ] ] @xmath173\end{aligned}\ ] ] notice that , in this case , alice would have a qubit and rob would have a qutrit , since for his mode he could have three different possible orthogonal states : particle spin - up , particle spin - down and particle pair . to characterize its entanglement we will use the negativity @xcite normalized to one
( we can multiply it by a constant in order to have negativity equal to one for a maximally entangled state ) , therefore , to have negativity equal to 1 for a bell state we define it as twice the addition of all the negative eigenvalues of the partial transpose density matrix which consists on transposing rob s qutrits @xmath174\end{aligned}\ ] ] @xmath175\end{aligned}\ ] ] we can write @xmath176 matricially in the basis @xmath177 @xmath178 which have the same expression than @xmath179 in the basis @xmath180
. therefore the four bell states will have the same eigenvalues which are : @xmath181 since @xmath182 @xmath118 and @xmath147 so that @xmath183 is negative for all values of the acceleration .
this implies , using peres criterion @xcite , that the spin bell states will be always entangled even in the limit of infinite acceleration .
we can readily evaluate the entanglement at the limits @xmath184 and @xmath9 if we compute the negativity ( normalized to one for maximally entangled states ) , that is to say @xmath185 applied to our states we obtain that @xmath186 in the limit @xmath184 we obtain @xmath187 which is an expected result since @xmath184 is the inertial limit .
however , in the limit @xmath9 we obtain @xmath188 , which implies that spin entanglement degrades due to the unruh effect .
2 shows the negativity as a function of the acceleration of rob .
[ fig2 ] and @xmath130 share a maximally entangled state in minkowski coordinates .
red dashed line is mutual information for all the spin bell states .
black dotted line is mutual information for the minkowski state @xmath189 and blue solid line is negativity for both , bell spin states and @xmath189,title="fig : " ] the mutual information , which takes into account quantum and classical correlations , is @xmath190 where @xmath191 are the von neumann entropies of the partial state of alice and rob and @xmath192 is the entropy of the whole state . for and
, the partial states of alice and rob ( @xmath193 , @xmath194 ) can be expressed matricially as @xmath195 in the basis @xmath196 @xmath197 in the basis @xmath198 for all the bell states , and @xmath199 for @xmath200 in the basis @xmath201 and the same expression for @xmath202 in the basis @xmath203 .
the entropies of these states are @xmath204 and the mutual information is @xmath205 again we see that in the limit @xmath206 mutual information goes to 2 and in the limit of infinite acceleration it goes to 1 . the behavior of the mutual information as a function of @xmath6 is shown in fig . 1 . in @xcite it is discussed that pauli exclusion principle protects the on occupation number entanglement from decoherence , and some degree of entanglement is preserved even at the limit @xmath9 . here
we have obtained a similar result for the spin bell states , showing that spin entanglement is also degraded by unruh effect .
next , we will study the case in which alice and rob share a different class of maximally entangled state .
we consider that in minkowski coordinates we have @xmath207 which is a maximally entangled state that includes occupation number entanglement along with spin .
we study this kind of states as a first analog to the state considered in previous literature .
this state corresponds to the choice @xmath208 in equation .
the density matrix of such a state is @xmath209\end{aligned}\ ] ] notice the significant difference from the bell spin states ; considering that rob accelerates means that , this time , alice has a qubit and rob has a qu4it .
hence , negativity acts only as a measure of distillable entanglement , and does not account for the possible bound entanglement the system would have @xcite . since in rindler
coordinates the state is qualitatively different from the minkowski bell states , , it is therefore worthwhile to study its entanglement and the mutual information degradation as rob accelerates .
the partial transpose of @xmath210 is @xmath211\end{aligned}\ ] ] which is an @xmath212 matrix .
@xmath213 is diagonal by blocks with eigenvalues @xmath214 as we can see , @xmath215 is non - positive and @xmath183 is negative for all values of @xmath6 , therefore the state will always preserve some degree of distillable entanglement .
if we calculate the negativity we will obtain @xmath216 which means that for this case , distillable entanglement behaves equally than in the previous case , and negativity on fig .
1 is equally valid for this state .
finally we compute the mutual information of the system whose partial matrices are expressed as @xmath195 in the basis @xmath217 .
@xmath218 @xmath219 in the basis @xmath220 .
the eigenvalues of the whole system @xmath221 matrix @xmath222 are @xmath223 in this case the mutual information as a function of @xmath6 is not proportional to the negativity .
hence , it is different from the bell states , . as it can be seen in fig .
1 the value of mutual information for , and coincide at the limits @xmath224 , but are different in between , obtaining that @xmath225 . to conclude this section we stress that the same results will be obtained if the state @xmath226 in is replaced by any other 1 particle bipartite spin state @xmath227 .
the previous work @xcite on occupation number entanglement between accelerated partners ignored the spin structure of the dirac field modes .
it is not possible to straightforwardly translate a state like into mere occupation number states .
this comes about because for a state like the bipartite vacuum component does not have individual spin degrees of freedom as the other components do .
in other words , by including the vacuum state in the superposition , the hilbert space ceases to be factorable in terms of individual spin times particle occupation number subspaces . on the other hand ,
the bipartite vacuum is a well defined total spin singlet .
hence , the hilbert space is factorable with respect to the total spin of the system @xmath228 and the occupation number subspaces .
accordingly , to reduce the spin information in the general density matrix we will be forced to consider a factorization of the hilbert space as the product of the total spin and occupation number subspaces .
if we do such a factorization we could consider that we are not able to access to the information of the total spin of the system @xmath228 and then , we should trace over total spin degree of freedom .
the equivalence between the standard basis ( occupation number - individual spin ) and the new basis ( occupation number - total spin ) is given in equations and @xmath229 { \left| { 0,\uparrow } \right\rangle}={\left| { 01 } \right\rangle}{\left| { d_+ } \right\rangle } & { \left| { 0,\downarrow } \right\rangle}={\left| { 01 } \right\rangle}{\left| { d_- } \right\rangle}\\*[1.5 mm ] { \left| { \uparrow,0 } \right\rangle}={\left| { 10 } \right\rangle}{\left| { d_+ } \right\rangle}&{\left| { \downarrow,0 } \right\rangle}={\left| { 10 } \right\rangle}{\left| { d_- } \right\rangle}\\*[1.5 mm ] { \left| { \uparrow,{\uparrow\downarrow } } \right\rangle}={\left| { 12 } \right\rangle}{\left| { d_+ } \right\rangle}&{\left| { \downarrow,{\uparrow\downarrow } } \right\rangle}={\left| { 12 } \right\rangle}{\left| { d_- } \right\rangle}\\*[1.5 mm ] { \left| { \uparrow,\uparrow } \right\rangle}={\left| { 11 } \right\rangle}{\left| { t_+ } \right\rangle}&{\left| { \downarrow,\downarrow } \right\rangle}={\left| { 11 } \right\rangle}{\left| { t_- } \right\rangle}\\*[1.5 mm ] \end{array}\ ] ] @xmath230\\*[1.5 mm ] { \left| { \downarrow,\uparrow } \right\rangle}=\frac{1}{\sqrt{2}}{\left| { 11 } \right\rangle}\left[{\left| { t_0 } \right\rangle}-{\left| { s } \right\rangle}\right ] \end{array}\ ] ] where we are using the basis @xmath231 and the triplets , doublets and the singlet are denoted as @xmath232 if we rewrite the general state in this basis we obtain @xmath233 and the general state when rob is accelerated in terms of this new basis after reducing the information on the total spin by tracing over this degree of freedom is @xmath234 which results in a state in the occupation number basis whose entanglement decoherence could be studied and compared with the results in reference @xcite in which spin is ignored : @xmath235+\mu\cos^3r\left(\frac{\beta^*-\gamma^*}{\sqrt2}{\left| { 00 } \right\rangle\!\left\langle { 11 } \right|}\right.\\ * \nonumber&&\left.+\frac{\beta-\gamma}{\sqrt2}{\left| { 11 } \right\rangle\!\left\langle { 00 } \right|}\right)+(1-\mu^2)\big[\cos^2r{\left| { 11 } \right\rangle\!\left\langle { 11 } \right|}\\ * & & + \sin^2r{\left| { 12 } \right\rangle\!\left\langle { 12 } \right|}\big]\end{aligned}\ ] ] we can readily compute the partial transpose @xmath236 @xmath237+\mu\cos^3r\left(\frac{\beta^*-\gamma^*}{\sqrt2}{\left| { 01 } \right\rangle\!\left\langle { 10 } \right|}\right.\\ * \nonumber&&\left.+\frac{\beta-\gamma}{\sqrt2}{\left| { 10 } \right\rangle\!\left\langle { 01 } \right|}\right)+(1-\mu^2)\big[\cos^2r{\left| { 11 } \right\rangle\!\left\langle { 11 } \right|}\\ * & & + \sin^2r{\left| { 12 } \right\rangle\!\left\langle { 12 } \right|}\big]\end{aligned}\ ] ] whose eigenvalues are @xmath238 all the eigenvalues are non - negative except @xmath239 .
the negativity is , in this case , @xmath240 which depends on the proportion of singlet of the @xmath241 component in the state . when there is no singlet component ( @xmath242 ) the negativity is zero . indeed in the limit @xmath184 ( minkowskian limit ) @xmath243 that shows that the maximally entangled minkowski occupation number state ( negativity @xmath244 ) arises tracing over total spin when the starting state is @xmath245\ ] ] or , in the occupation number - total spin bases @xmath246\ ] ] that means that , for occupation number entanglement , the only way to have an entangled state of the bipartite vacuum @xmath247 and the one particle state @xmath241 of a dirac field is through the singlet component of total spin for the @xmath241 component . on the contrary , the state @xmath248\ ] ] will become separable after tracing over total spin due to the orthonormality of the basis , .
we have established that the minkowski maximally entangled state for occupation number arises after tracing over total spin in a state as .
now we will compute the limit of the negativity when the acceleration goes to @xmath45 in order to see its unruh decoherence and to compare it with the results for occupation number entanglement from @xcite . taking @xmath249 in @xmath250 therefore , for the maximally minkowski
entangled state we have @xmath251 , @xmath252 and the negativity in the limit is @xmath253 this result shows that when we are reducing the total spin information , looking at the occupation number entanglement alone , we see that it is more degraded by unruh effect than when we considered spin bell states in the previous section .
more importantly , the occupation number entanglement is more degraded than in @xcite , where the spin structure of the modes was considered nonexistent .
this happens because considering spin structure of each mode , occupation number 2 is allowed .
hence , pauli exclusion principle protection of the entanglement is weaker than in @xcite where the spin is not considered . the negativity dependence on the acceleration is shown in fig .
3 [ fig5 ] we can also compute the mutual information for the state as we did in the rest of the cases . its analytical expression is quite long and has no special interest , but we can see the dependence of @xmath254 for the minkowski maximally entangled state with the acceleration in fig .
3 , obtaining that @xmath255 and @xmath256
it is known @xcite that unruh decoherence degrades entanglement of occupation number states of fields . here
we have shown a richer casuistic that appears when we take into account that each dirac mode has spin structure .
this fact enables us to study interesting effects ( such as unruh decoherence for spin bell states ) and develop new procedures to erase spin information from the system in order to study occupation number entanglement . along this work
we have analyzed how a maximally entangled spin bell state losses entanglement when one of the partners accelerates .
we have seen that , while in minkowski coordinates alice and rob have qubits , when rob accelerates the system becomes a non - pure state of a qubit for alice and a qutrit for rob . in this case
spin entanglement for a dirac field is degraded when rob accelerates .
however some degree of entanglement survives even at the limit @xmath9 .
a first analog to the well studied state @xmath257 but including spin could be , for instance , @xmath258 .
this state , unlike the deceivingly similar spin bell states , becomes a qubit@xmath2qu4it when rob accelerates .
nevertheless , distillable entanglement degrades in the same way as for spin bell states .
we have also introduced a procedure to consistently erase spin information from our setting preserving the occupation number information .
we have done it by tracing over total spin .
the maximally entangled occupation number state is obtained from the total spin singlet after tracing over total spin .
finally we have shown that its entanglement and mutual information is more degraded than in @xcite where the spin structure of dirac modes was neglected . a reasonable physical argument for this result
is that , in our setting , occupation number 2 is allowed for the dirac field modes , and hence , there is a broader margin for entanglement degradation by unruh effect .
the thermal noise is obtained when dealing with a two dimensional space - time and massless fields .
mass gap and transverse degrees of freedom modify the counting statistics that is no longer given by thermal noise , but replaced by the so - called rindler noise @xcite , that the the space - time dimension . in this work
we were concerned with the specific issues associated to the spin degree of freedom , so the restriction to massless fields in two dimensional space - time adopted here , as well as the single mode approximation , allows a direct comparison with the previous works @xcite which considered massless spinless fields in 2d under those approximations . as a matter of fact , having more than 2 space - temporal dimensions and having massive fields may introduce relevant effects . allowing the possibility of having momentum off the acceleration direction and having massive fields we would obtain a spread of minkowski modes over rindler frequencies @xcite .
if we carry out the single - mode approximation , the spread of minkowski modes into rindler modes can be neglected even for higher dimensions @xcite , but if we want to relax such an approximation ( as for the discussion in the next paragraph ) , those effects should be considered in order to account for the entanglement in non - inertial frames .
it would be worthwhile to study those effects in future articles .
another very interesting point that deserves further study is the fact that when we consider more than one populated mode of the complete minkowski vacuum instead of the single mode approximation , the margin for unruh degradation increases as we could have in principle a larger number of levels that can be excited by the thermal / rindler noise .
one could think that these cases would be quite similar to the bosonic case @xcite where the margin for unruh decoherence is so broad that no entanglement survives at the limit of @xmath9 .
something similar would apply as well if we relax the single mode approximation allowing a small spread in both rob s detector and populated levels such that we consider a continuum of accessible levels .
these topics will be inspiration for future works .
the authors thank c. sabin and j. garcia - ripoll for the useful discussions during the elaboration of this paper . this work has been partially supported by the spanish mcin project fis2008 - 05705/fis .
e. m - m is partially supported by the csic jae - predoc2007 grant .
we are in debt to an anonymous referee , that has pointed out very interesting questions which helped us substantially improve this article .
to help with the calculations of the density matrix associated with it is useful to compute firstly the trace over iv on all the operators that compose . using equation we have @xmath259\otimes \text{h.c.}\end{aligned}\ ] ] tracing over iv : @xmath260 where notation is different in each side of the equality : bras and kets in l.h.s .
are referred to minkowski coordinates for alice and rob @xmath166 and in the r.h.s .
they are referred to alice s mode in minkowski coordinates and rob s mode in rindler s region i @xmath261 . in the same way ,
using expressions , we have @xmath262{\left\langle { s } \right|}\left(\cos r\,{_i\!\!{\left\langle { s ' } \right|}}\,{_{iv}\!\!{\left\langle { 0 } \right|}}\right.\\ * & & \left.+ \varepsilon'\ , e^{-i\phi}\sin r\,{_i\!\!{\left\langle { { \uparrow\downarrow } } \right|}}\,{_{iv}\!\!{\left\langle { s ' } \right|}}\right)\end{aligned}\ ] ] where @xmath263 if @xmath264 and @xmath265 if @xmath266 . now , tracing over iv @xmath267 notation here is the same than in .
again , using expression we get @xmath268{\left\langle { s_{3 } } \right|}\left[\cos r\,{_i\!\!{\left\langle { s_4 } \right|}}{\,_{iv}\!\!{\left\langle { 0 } \right|}}\right.\\ * & & \left.+\varepsilon_4e^{-i\phi}\sin r\,{_i\!\!{\left\langle { { \uparrow\downarrow } } \right|}}{\,_{iv}\!\!{\left\langle { s_4 } \right|}}\right]\end{aligned}\ ] ] and tracing over iv gives @xmath269 again , notation here is the same than in .
|
we investigate the unruh effect on entanglement taking into account the spin degree of freedom of the dirac field .
we analyze spin bell states in this setting , obtaining their entanglement dependence on the acceleration of one of the partners .
then , we consider simple analogs to the occupation number entangled state @xmath0 but with spin quantum numbers for @xmath1 . we show that , despite their apparent similitude , while the spinless case is always qubit@xmath2qubit , for the spin case acceleration produces a qubit@xmath2qu4it state .
we also introduce a procedure to consistently erase the spin information from our setting preserving occupation numbers .
we show how the maximally entangled state for occupation number emerges from our setting .
we as well analyze its entanglement dependence on acceleration , obtaining a greater entanglement degradation than in the spinless case .
|
[ h ] [ 0.26 ] + [ h ] [ 0.26 ] + the development of novel mobility technologies for cleaner vehicle emissions is essential to mitigate the current high levels of pollution seen on internal combustion engines .
one foreseeable technology involves the electrochemical conversion of energy using polymer electrolyte membrane fuel cells ( pemfc)@xcite . used in cars , fuel cells currently provide the necessary power needed to travel even long - distances .
however , the costly use of platinum ( pt ) as an electrocatalyst inside pemfc has triggered an a vast active search for cheaper pt - based alloys .
one proposed solution is to combine it with other late transition metals such as nickel ( ni ) , cobalt ( co ) , chromium ( cr ) , copper ( cu ) and iron ( fe)@xcite . the goal of the alloyed material is to improve the catalytic activity towards the oxygen reduction reactions ( orrs ) offered by commercial pt cathode catalysts without the scaling costs . in this respect , recent studies have shown that a five - fold decrease in the amount of pt currently used in pemfc stacks is needed , in terms of pt costs and scarcity , in order to reach levels for mass - production and commercialisation in light - duty vehicles @xcite . among these alloys
, a promising one corresponds to pt - ni .
recent experimental studies performed by stamenkovic _
shown an exceptional catalytic activity ( 10-fold ) towards orr of an extended pt@xmath1ni(111 ) surface compared to a monometallic pt(111 ) surface ; and up to 90 times than the carbon supported pt catalysts used in pemfc @xcite .
this was attributed to an unusual shift in the pt - ni _ d_-band centre along with a peculiar atomic segregation pattern involving a pt - rich outer layer .
furthermore , the maximisation of the large surface areas offered by nanoparticles ( nps ) makes them ideal as catalysts for fuel cells applications .
recent experimental works have focused on the preparation of small pt - ni nps ( @xmath2 5 nm ) having well - controlled octahedral geometries ( fcc - type atomic arrangements ) along with extensive ( 111 ) facets aiming for similar catalytic properties as the extended surface@xcite . from a theoretical view ,
computational monte carlo ( mc ) simulations have previously addressed the structural stability of cuboctahedral pt - ni nps@xcite . using many - body potential for particles ranging from 2.5 to 5 nm .
mc simulations at 600k have shown that pt - ni tend to form _ surface - sandwich _ structures , with a segregation pattern in which pt atoms are enriched the outermost and third shells while the ni atoms are enriched in the second shell .
these results suggest an economical catalyst design with pt atoms located at the outermost shells with ni occupying core positions .
similar mixed segregation patterns were obtained , though for small pt - ni clusters of less than 20 atoms , using a genetic algorithm global optimization approach via density functional theory ( dft ) calculations @xcite , while more recent work , we have shown that sub - nanometre gas - phase pt - ni clusters up to 55 atoms can bind @xmath3 too strongly ( above 1 ev ) due to substantial geometrical reconstruction of the metal - metal bridge underneath the molecule , making those clusters not ideal for catalysing the sluggish orr compared to a pt(111 ) surface@xcite .
more recent dft calculations performed by fortunelli and co - workers have reported a fully - dealloyed pt@xmath1ni@xmath4 particule ( @xmath5 8 nm . ) surfaces exhibiting triangulated surface arrangements , as a regular pt(111 ) , while reducing the rate - determining orr step significantly@xcite . in this work we performed extensive dft periodic calculations to quantify o@xmath0 adsorption on a variety of pt - ni clusters supported on mgo(100 ) as well as on their monometallic ( pt , ni ) counterparts . using the calculated adsorption energies ( e@xmath6 ) we are able to construct a o@xmath0 chemisorption map over supported pt - ni , pt and ni clusters , relevant to fuel cells and heterogeneous catalysis . furthermore , the interfacial low - index surface displayed at the cluster / oxide interface - where the cluster can expose both ( 100 ) and ( 111 ) facets in contact with the oxide interface - was also analysed . for the bimetallic case , attention
is placed on ni doping at clusters core sites ( _ i.e. _ core - shell clusters ) , though in some cases ni atoms are located inevitably located at surface sites due to cluster size ( to@xmath7 ) or geometrical arrangements ( dh@xmath8 ) .
overall , we identify the ( 111)/(111 ) and ( 111)/(100 ) cluster edges as those sites were calculated e@xmath6 values are the strongest ( above 1 ev ) which tend to have a low gcn value ( @xmath96 ) ; while fcc and hcp sites located at ( 111 ) facets have the weakest e@xmath6 values ( less than 0.64 ev ) involving gcn @xmath10 8 . as gcn can distinguish symmetrically equivalent sites , it allow us to rationalise calculated e@xmath6 values among different clusters since weaker interactions occur at sites where gcn numbers are higher . it has been recently shown that gcn further establishes a link between geometry , adsorption and activity@xcite .
we considered 4 different structural motifs ranging from fcc - type such as truncated octahedra ( to ) , cuboctahedral ( co ) , and non - crystallographic arrangements such as icosahedral ( ih ) and decahedral ( dh ) ; with sizes ranging from 25 up to 58 atoms .
these structures are constructed from highly symmetric models : 38(to)@xmath11 , 55(co)@xmath11 , 55(ih)@xmath11 and 75(dh)@xmath11 .
half - cuts were performed on these structures to obtain a new set of clusters with exotic cluster / oxide interfaces : 25(to)@xmath11 , 33(ih)@xmath11 and a 58(dh)@xmath11 .
here , the subscript @xmath12 stands for the miller indexes of the cluster - oxide interface initially exposed . in the following ,
both ( 100 ) or a ( 111 ) facets have been chosen .
pt - ni clusters display a deliberate pt@xmath13-ni@xmath14 chemical arrangement . depending on size and structural
, some ni atoms are find to occupy surface sites , for example pt@xmath15ni@xmath16(to)@xmath17 , pt@xmath18ni@xmath19(ih)@xmath20 as well as pt@xmath21ni@xmath22(dh)@xmath20 ( see figs . [
fig : sites1 ] and [ fig : sites2 ] ) .
after construction in the gas - phase , the metal clusters are subsequently placed over the mgo(100 ) substrate maximising the number of metal - oxygen ( m - o ) bonds .
an initial height of @xmath5 2 prior dft relaxations is employed between the cluster and the substrate ( optimal pt and ni atoms dft heights adsorbed on o - sites surface sites are 1.985 and 1.796 , respectively ) .
the substrate thus acts as a geometrical constraint affecting the features of each metallic clusters . as part of the cluster / oxide interfacial geometry characterisation
, we have counted the number of metal atoms at the interface .
furthermore , as a measure of the metal interface _ roughness _ , we have calculated the standard deviation of the corresponding interfacial atomic layer height , information which can be found in the supplementary material along with other calculated quantities .
we must stress that the considered supported pt , ni and pt - ni clusters , though locally dft - relaxed , they do not correspond to global minima in the potential energy surface ( pes ) .
however , they act as geometrical scenarios to understand their o@xmath0 adsorption properties . the plane - wave implementation of density functional theory ( dft ) within the quantum espresso ( qe ) code @xcite with the pbe exchange - correlation ( xc ) functional@xcite
. a combination of rappe - rabe - kaxiras - joannopoulos @xcite and vanderbilt@xcite ultrasoft pseudopotentials - including non - linear core corrections - are use to describe pt(5d@xmath236s@xmath24 ) , ni(3d@xmath234s@xmath24 ) , mg(2p@xmath253s@xmath243p@xmath26 ) and o(2s@xmath272p@xmath28 ) atoms . a kinetic energy cutoff for wave - functions and charge density of 45 and 360 ry ,
respectively is used . to improve convergence ,
a marzari - vanderbilt smearing value of 0.001 ry is used .
the pristine mgo(100 ) surface is modelled using a 6@xmath296 three - layer slab our calculated pbe bulk lattice constant ( 4.238 ) .
although this is slightly larger than the experimental value of 4.21 , is whoever , in close agreement with calculated pbe values ( 4.26 and 4.30 ) and hybrid ( pbe0 ) functionals ( 4.21 )@xcite .
the mgo(100 ) substrate has been considered on the basis that it is a often used in experiments , it is a well - characterized and rigid substrate strongly interacting with a metallic cluster and it can be accurately treated with standard ab - initio methods .
its checkerboard ( 100 ) surface allow us to analyse any effects between flat and corrugated oxide - cluster interfaces on the calculated e@xmath6 values .
the size of the supercell ( 17.9803 @xmath29 17.9803 @xmath29 29.2380 ) provides a sufficiently large vacuum region , necessary to avoid any spurious interactions between neighbouring periodic images .
all dft calculations have been performed spin - polarised .
the brillouin zone was sampled at the gamma point only .
empirical dispersion corrections are included via single - point calculations on pbe relaxed configurations using the dft+d functional , which includes a pairwise addition c@xmath30r@xmath25 correction term@xcite .
o@xmath0 adsorption ( chemisorption ) energies ( @xmath31 ) are calculated as total energy differences between the interacting configuration ( e@xmath32 ) , the lowest - energy supported bare cluster ( e@xmath33 ) , and the o@xmath0 molecule in the gas - phase : @xmath34 where positive e@xmath6 values indicate a exothermic o@xmath0 adsorption .
we also quantify the pt - ni , pt and ni clusters _ core _ and _ shell _ strain as the percentage difference between the average nearest - neighbour ( nn ) distance for those atoms occupying core ( d@xmath35 ) and shell ( d@xmath35 ) positions with respect to pbe calculated pt ( 2.812 ) and ni ( 2.487 ) lattice constants , while for pt - ni we considered the ni - pt@xmath1 phase with an experimental value of 2.718 , see ref.@xcite ( d@xmath36 ) .
negative values in eq .
[ eq2 ] indicate a compressive strain felt by the cluster : @xmath37 we employ another quantity to geometrically characterise the adsorption site , the generalized coordination number ( gcn )
. introduced by sautet and co - workers for monometallic pt systems the gcn is a new descriptor as robust as the _ d_-band center , which establishes a direct link between geometry , adsorption map and activity@xcite .
it to help us differentiate each adsorption site despite size , shape and composition of the cluster systems in order to highlight adsorption trends across a wide variety of structures .
the gcn can be defined as the coordination number at the adsorption site weighted by the overall coordination of the neighbouring atoms : @xmath38 where the sum runs over all neighbouring atoms _ n@xmath39 _ of the atom _
i _ and each neighbour is weighted according to its coordination number divided by the maximum coordination number of the adsorption site .
we keep _ cn@xmath40 _
= 18 as the o@xmath0 molecule is adsorbed at a bridge site ( on top of 2 pt@xmath13 atoms ) , as shown recently for larger supported pt - ni clusters@xcite .
+ [ h ] [ 0.2 ]
[ htb ] @cc@ [ 0.10 ] & [ 0.10 ] + [ 0.10 ] & [ 0.10 ] + [ 0.10 ] & [ 0.10 ] + due to the wide variety of pt - ni geometries and sizes we were able to identify several inequivalent sites for o@xmath0 adsorption .
these o@xmath0 adsorption sites have been grouped into different _ families _ as shown in both figs [ fig : sites1 ] and [ fig : sites2 ] .
prior to dft relaxations , the o@xmath0 molecule was always placed horizontally towards the cluster surface , ( _ bridging _ between two metal atoms ) , as recent studies theoretical studies show that this is the preferred adsorption configuration @xcite .
our analysis thus focuses mostly on the differences between the strongest and weakest o@xmath0 adsorption sites found for supported pt - ni clusters and use the bare pt and ni cases as references .
[ 0.35 ] + [ 0.2 ] + an o@xmath0 chemisorption map over pt , ni and pt - ni supported clusters is shown in fig.[fig : echems ] . here ,
calculated e@xmath6 values are plotted as a function of all inequivalent sites considered .
previously calculated pbe reference values for the o@xmath0 adsorption on the flat pt(111 ) are highlighted by the red area ( 0.46 up to 0.86 ev)@xcite , including experimental values ( 0.4 - 0.5 ev)@xcite , as well as for the stepped pt(321 ) surface ( 1.23 to 1.56 ev)@xcite . from fig.[fig : echems ] , we identify 4 pt - ni adsorption sites which lead to e@xmath6 values resembling those of the pt(111 ) surface : _ fcc _ , _ hcp _ , _ between edges _ , ( 111)/(100 ) and _ no vertex _ sites . in some cases , calculated e@xmath6 values are @xmath41 0.5 ev , particularly at _ fcc _ , _ hcp _ sites , where in most of the dft relaxations the o@xmath0 molecule undergoes fast dissociation .
we must stress we are only analysing the o@xmath0 adsorption trends on a variety of inequivalent sites , in order to understand o@xmath0 adsorption on a variety of pt - ni cluster structures .
this represents the first step in the complex oxygen reduction reaction ( orr ) which consisting of a series of protonation process of the dissociative products ( o , oh , ooh and h@xmath0o@xmath0 ) to form a final h@xmath0o product@xcite .
+ the relevance of ( 111 ) facets on cluster structures as candidates for low - barrier o@xmath0 dissociation has been recently discussed for truncated octahedral ( to ) gas - phase clusters at atom sizes 38 , 79 and 116 by jennings _ _ et al__@xcite . according to their work , stronger @xmath31 values are calculated for pure pt@xmath42 on both fcc and hcp sites ( 1.79 and 1.84 ev ) compared to a core - shell platinum - titanium pt@xmath43ti@xmath44 bimetallic cluster ( 0.38 and 0.74 ev ) . however , they reported near barrier - free dissociation at the fcc / hcp sites for pure pt clusters ( 0.00 and 0.04 ev ) with slightly larger barriers for the bimetallic case ( 0.62 and 0.34 ev ) .
the monometallic case behaviour was explained due to an easily distorted pt(111 ) facet - namely the central pt atom - facilitating o@xmath0 dissociation ; while an increase in the rigidity of this ( 111 ) facet was due to ti - alloying at the core , for the bimetallic case .
similar trends were also reported for larger to@xmath45 particles@xcite .
overall , the rigidity of the ( 111 ) was reported to decrease for 3d metals from the 4 - 8 group ( ti to fe ) , while similar distortions as of the pure pt cluster were reported for groups 10 - 12 ( ni , cu and zn ) .
+ we monitored in detail the metal - metal ( m - m ) bond length at the o@xmath0 adsorption site changes for the three different systems , fig .
[ fig : data ] ( a ) . upon o@xmath0 adsorption
, the pt - pt bond is stretched as a general trend due to a `` softer '' character of pt , in order to accommodate the incoming o@xmath0 molecule .
this is noticeable at edge sites , sites such as _ along edges _ , both ( 111)/(111 ) and ( 111)/(100 ) ; @xmath31 values of 1.60 - 1.94 ev . particularly at the to@xmath42 ( 111 )
facet , the central pt atom involving in o@xmath0 adsorption on pt - ni clusters is `` pulled up '' . for ni clusters ,
ni - ni bonds remain rather stable upon o@xmath0 adsorption , clustered around 2.40 - 2.70 , while it is the molecule o - o bond which is elongated - in some cases , stretched to the point of dissociation - as shown in fig .
[ fig : data ] ( b ) . having a pt shell ,
the m - m bond length of bimetallic pt - ni clusters resemble those values calculated for monometallic pt . from fig.[fig : data ] ( a ) , we notice that the m - m bond length in pt - ni , pt and ni have a `` l''-shaped distribution .
a linear behaviour is observed where shortening the m - m bond stands for an e@xmath6 increment ; when a critical compression is reached there is an abrupt change and small changes in the e@xmath6 are associated to very different m - m bond lengths .
however , m - m bond elongation is not the sole cause of larger e@xmath6 , but it is connected to both molecule o - o bond stretching and the adsorption site . fig .
[ fig : data ] ( b ) , shows only the o - o bond length range for molecular adsorption . from this plot , it seems to be two regimes for pt - ni : one where the e@xmath6 values increase from 0.18 up to nearly 1.2 ev while the o - o bond decreases ( 1.43 , down to 1.37 ) , namely namely at fcc / hcp sites for to / co structures and the _ no - vertex _ site of ih clusters . a second regime , starting roughly at 1.3 ev up to 2.1 ev ,
is observed where the o - o bond length increases from 1.37 up to 1.41 , mostly _ along edges _ and _ 5-fold vertex _ sites .
however , this behaviour is more erratic for monometallic pt clusters while ni clusters involve o - o bond lengths with e@xmath6 above 1.5 ev .
+ fig.[fig : data ] ( c ) shows both _
core _ and _ shell _ calculated strain within the cluster , as it has been recently further shown a direct link between a nanoparticle ( np ) reactivity and its compressive strain , particularly in bimetallic cases@xcite . _
strain for pt and pt - ni bimetallic clusters is rather large , particularly for the bimetallic case where compressive values can reach reach values around -10@xmath46 as this clusters are still small enough that have not yet developed bond lengths are those of the bulk .
ni clusters values show a modest compression ( @xmath41 -2@xmath46 ) , except for the to@xmath7(100 ) structure , where 12 ni atoms at the cluster / oxide interface need to match equal12 o surface sites , with average interface bond lengths of 2.52 , compared to the average ni - ni cluster bond length of 2.45 ; thus enlarging the internal ni - ni core distances leading to positive core strain values .
similar trends are seen for the ih@xmath47(111 ) , which due to its large and irregular cluster / oxide interface needs to stretch those ni - ni bonds at the interface and core positions .
yet , it is difficult to imply a direct correlation with the calculated @xmath31 values , except that we observe a trend where pt - ni clusters having a small number of atoms at the cluster / oxide interface ( @xmath41 8) tend to display a large core strain ; while those cluster with a larger number of interfacial atoms , _ core _ strain tend to decrease up to approximately -4@xmath46 . regarding _ shell _
strain - see fig.[fig : data ] ( d ) - we observe that pt - ni clusters display smaller strain values compared to monometallic pt clusters .
there is a small window , where fcc and hcp sites on pt@xmath43ni@xmath44to(111 ) as well as those for pt@xmath48ni@xmath49ih(111 ) have slightly less ( @xmath41 -1 @xmath46 ) _ shell _ , leading to e@xmath6 values lower than 0.5 ev .
+ bader charge transfer analysis was performed for both pt - ni , pt and ni clusters at _ best _ ( strong ) and _ worst _ ( weak ) o@xmath0 adsorption sites . in particular , fig .
[ fig : data ] ( e ) shows the amount of charge on the o@xmath0 molecule calculated for the pt - ni , pt and ni systems . as a general trend
, one can observe that the amount of charge transferred to the supported clusters varies as a function of interfacial metal atoms ( number of contact o surface sites ) , with less charge being transferred from the substrate to ni clusters , compared to pt and pt - ni clusters .
overall , the o@xmath0 molecule subtracts an small amount of charge ( @xmath5 0.35 e@xmath50 ) per o atom at pt - ni and pt sites , being larger for ni ( @xmath5 0.5 e@xmath50 ) , in line with having larger e@xmath6 values calculated for ni clusters .
this excess of charge on the o@xmath0 molecule creates a metal - superoxo ( m - o@xmath51 ) type intermediate ( one electron transferred)@xcite .
charge transfer is in line with both pt ( 2.28 ) , ni ( 1.91 ) and o(3.44 ) pauling electronegativities , and to those structural motifs and interfacial geometries which maximise the number of metal atoms in contact o - surface sites .
+ there is a modest tendency of pt - ni clusters to reduce the overall _ roughness _ of the ( 100 ) and ( 111 ) cluster / oxide interface layers , as shown in fig .
[ fig : data ] ( f ) . in this plot
, the pt metal - metal ( m - m ) bond length of the interface layer in pt - ni clusters results to be shorter than the average m - m bond length in the cluster ( pt - ni @xmath5 2.65 ) , in contrast with the pure cases ( pt @xmath5 2.73 , ni @xmath5 2.44 ) , where they show an elongation on the metal distances at the interface . even though low e@xmath6 values are calculated for to and co clusters in contact with the oxide through their large ( 100 ) facets at low _ roughness _ ( less than 0.1 @xmath46 ) , the dispersion of the overall values is too wide to provide any conclusive analysis suggesting a more local chemistry ruling o@xmath0 adsorption energies .
the role of the support in our work thus facilitates charge being transferred to the supported cluster ( which eventually is redistributed among the cluster and o@xmath0 molecule ) as well as it provide an `` anchor '' point for the cluster .
the substrates could eventually promote the controlled preparation of ( 111 ) facets as the pt - ni clusters grow in size , where tailored o@xmath0 adsorption ( and further dissociation ) can take place@xcite .
+ the inclusion of a semi - empirical dispersion correction ( dft+d ) via single - point ( scf ) calculations on all the inequivalent configurations involving o@xmath0 adsorption have , in most cases , modestly increased the calculated e@xmath6 values . in general , predicted pbe energetic ordering between adsorption sites was preserved , with only a minor energetic rearrangement among the strongest chemisorption sites .
full dft+d relaxations were also performed for all 5 chemisorption sites ( fig .
[ fig : sites1 ] ) , only for the pt@xmath7 , ni@xmath7 and pt@xmath15ni@xmath16 ( to)@xmath17 structures to confirm the an overall preference for edge sites , such as ( 100)/(111 ) and ( 111)/(111 ) trends . for pt - ni clusters , dft+d relaxations predicted the _ along edge _ ( 111)/(100 ) as the strongest chemisorption site ( 1.62 ev ) , closely followed by _ along edge _ ( 111)/(111 ) site ( 1.47 ev ) just as the pbe energetic ordering .
similar gains in energy are calculated for the fcc and hcp sites . for the pt clusters ,
the best adsorption site is now the _ along edge _
( 111)/(100 ) instead of the predicted pbe _ along edge_(111)/(111 ) , at 1.96 ev .
being now the o@xmath0 molecule at close proximity to the oxide surface ( 2.99 ) , dft+d is able to capture an stronger overlap between the electron orbitals of o@xmath0 and those of the oxide surface .
this results in an e@xmath6 increase up to 1.96 ev , from the 1.38 ev single - point ( dft+d ) calculation , an increased also calculated for the _ between edge_(111)/(100)site , from 1.67 to 1.71 ev .
analogous trends are observed for ni clusters , where _ along edge _ ( 111)/(100 ) site remains as the one having the largest e@xmath6 value ( 2.55 ev ) , while marginal gains are calculated for sites _ along edge _
( 111)/(111 ) and _ between edge _ ( 111)/(100 ) ( 2.25 and 1.79 ev , respectively ) . + upon o@xmath0 adsorption , we calculated the projected density of states ( pdos ) for the _ best _ ( strongest ) and _ worst _ ( weakest ) - namely the fcc site - adsorption sites , see fig .
[ fig : dcenters ] . included within the figure , is the calculated _
d_-band center@xcite .
this is done for those pt atoms ( and ni ) involved in the o@xmath0 adsorption ( which can occur , both molecularly and dissociated ) .
comparisons between different pt - ni clusters are difficult due to the size range considered as well as some _ worst _ configurations involve dissociated o@xmath0 .
overall , we observe that the calculated _ d_-band centers of the 2 pt atoms from the _ best _ adsorption sites ( molecularly adsorbed o@xmath0 ) are positioned nearly at the same energy values ( see vertical black lines , fig.[fig : dcenters ] ) . for the all the _ best _ sites , this implies having their corresponding _
d_-centers positioned lower in energy with respect to the fermi energy ( @xmath52 ) , thus accounting for stronger interactions with the molecule o atoms , except for those configurations where o@xmath0 is dissociated , namely pt@xmath15ni@xmath16to(100 ) , pt@xmath43ni@xmath44to(100 ) and pt@xmath48ni@xmath49co(100 ) , fcc sites ) .
for most of these dissociated configurations , we observe that the position of the pt _
d_-centers , of the least coordinated pt atom is highly shifted to low energies thus implying a strong interaction with the o atom ; see pt@xmath43ni@xmath44to(100 ) and pt@xmath48ni@xmath49co(100 ) . most the rest of the _ worst _ configurations , the pt
_ d_-centers are closer to the @xmath52 , thus corresponding to weaker interactions between the o@xmath0 and the cluster .
however , pt@xmath43ni@xmath44to(111 ) is an exception , as the o@xmath0 is adsorbed in a molecular fashion , though the the central pt _ d_-centers - occupying the ( 111 ) facet where adsorption takes - place is shifted to lower energy values as the pt atom is `` lifted '' from its position as it interacts with the o@xmath0 molecule , a geometric effect highlighted in previous work@xcite .
larger clusters , such as pt@xmath48ni@xmath49 ( both co and ih ) and the decahedral pt@xmath21ni@xmath22 , further follow similar _
d_-centers shift trends .
however , close overlaps between calculated _ d_-centers for _ worst _ and _ best _ sites , particularly for pt@xmath21ni@xmath22 , are observed as @xmath31 differences for the two sites become smaller .
thus , to get disentangle electronic from geometric effects , we have calculated the corresponding gcn for both adsorption sites for all pt - ni clusters as done for the _ d_-band centers .
the calculated gcn values , as described in eq .
[ eq3 ] , show for pt - ni clusters an overall linear relationship with respect to calculated @xmath31 values ; see fig . [
fig : gcn ] . as the gcn value increases , a reduction on the @xmath31 values
is seen , particularly above a gcn @xmath10 8 , where most of the calculated values involving fcc and hcp sites resemble the experimental and those calculated values for o@xmath0 adsorption on pt(111 ) surface ( 0.46 up to 0.86 ev)@xcite , and even below .
there was only 1 case where the fcc site reported a lower gcn value ( 6.722 ) , on the the pt@xmath43ni@xmath44(111 ) cluster . here
, the central pt atom of the ( 111 ) facet was lifted with respect to the facert thus separating it from the rest of the core pt atoms and thus lowering its gcn value . on the opposite ,
the strongest @xmath31 values ( @xmath10 0.64 ev and above ) can be found at sites such as _ along edges _ ( 111)/(111 ) ; where gcn between 4 and 5 are calculated . for monometallic pt clusters , a similar trend is observed .
however , the slope of the linear fitting is less pronounced due to the `` soft '' nature of the pt - pt bonds within the cluster , which can stretch their cluster m - m bonds to accommodate the incoming o@xmath0 molecule effecting the overall gcn and @xmath31 values .
ni clusters on the other hand , interact their strongly with o@xmath0 that for gcn values between 5 and 6 ( @xmath53 ) - leaving a narrow window from 6 to 8 for a slightly weaker adsorption - again at gcn above 8 , the o@xmath0 molecule is strongly adsorbed and in some cases fastly dissociated thus affecting the overall linear trend .
interestingly , we have reported similar trends where o@xmath0 adsorption can take place at low @xmath31 ( @xmath41 0.5 ev ) for gcn values above 8 , for the particular case of larger to and co clusters involving 82 and 86 atoms , respectively@xcite .
using a wide variety of pt - ni , pt and ni geometries , sizes and compositions supported over mgo(100 ) our results we shown that molecular have shown that o@xmath0 adsorption on the sub - nanometer range ( from 25 up to 58 atoms ) is a rather complex case . their corresponding electronic structure as analysed via the calculated position of _ d_-band center as well as the local geometry at the adsorption site using the generalised coordination number ( gcn ) .
the calculation of these two quantities show that these two effects are deeply intertwined , directly affecting the calculated e@xmath6 values .
o@xmath0 adsorption is regulated by mainly 2 types of inequivalent sites : ( a ) strongest e@xmath6 values ( above 1 ev ) are calculated at cluster edges , such as the _ along edge_(111)/(111 ) and ( 111)/(100 ) as well as the _ between edge_(111)/(100 ) sites which tend to have a low gcn value ( @xmath96 ) ; and ( b ) the fcc and hcp sites where the _ weakest _ e@xmath6 values ( less than 0.64 ev ) are calculated , involving gcn values @xmath10 8 which resemble those o@xmath0 adsorption energies of the flat pt(111 ) surface . in general , calculated _
d_-band centers positions for those pt atoms at the adsorption are located at lower energies from the fermi level ( e@xmath54 ) , thus indicating a stronger binding between the metal and the o@xmath0 adsorbate .
the opposite is seen for the weakest adsorption sites , where _ d_-band centers positions are closer to the e@xmath54 ; though this trend can be affected if the o@xmath0 is found to be dissociated as pt interactions with atomic o are stronger compared to the molecular case . from our calculations
provide - as far as we are concerned - the first o@xmath0 chemisorption map for across a wide range of supported pt - ni clusters .
clear o@xmath0 adsorption trends provide a deeper understanding on potential adsorption sites for low - barrier o@xmath0 dissociation , following sabatier principle
. this will imply that the best catalysts should be the ones that bind both atoms and molecules not too weakly nor too strongly , in order to activate the reactants and to efficiently desorb the products@xcite .
we show from our results that it is theoretically possible to construct core - shell systems , where the core of the nanoparticle ( cluster ) is made of a cheaper metal ( ni ) compared to a more expensive metal on the shell ( pt ) .
this not only reduces the amount of pt - loading by alloying ; but preserves the pt chemistry at a mere 1 atomic - layer pt - shell where o@xmath0 adsorption process can occur with calculated e@xmath6 values of the order of those seen for a full monometallic pt nanoparticle and bulk pt(111 ) surface .
having systematically analysed the effects of o@xmath0 adsorption over supported over mono ( pt , ni ) and bimetallic ( pt - ni ) clusters ( @xmath551 nm . ) , we foresee that a potential tailoring of chemisorption properties and the rational design of novel nanocatalysts will ultimately depend a synergistic combination of controlled nps preparation methods ( size , geometric and segregation effects ) and on a profound knowledge of the complex electronic properties of both nps and oxide support .
this work has been supported by the uk research council epsrc under the critical mass toucan project , grant number ep / j010812/1 .
fb benefit of the financial support by epsrc , under grant no .
ep / go03146/1 and the royal society grant no .
density functional theory ( dft ) calculations have been performed on archer ( cray xc30 system ) , the uk s national high - performance computing ( hpc ) facility .
lopb acknowledges supercmputo unam ( miztli ) for computational resources under projects sc16 - 1-ig-78 and sc15 - 1-ig-82 as well as financial support from papiit - unam ( project ia102716 ) .
|
we present a systematic analysis of molecular oxygen ( o@xmath0 ) adsorption trends on bimetallic pt - ni clusters and their monometallic counterparts supported on mgo(100 ) , by means of periodic dft calculations for sizes between 25 up to 58 atoms .
o@xmath0 adsorption was studied on a variety of inequivalent sites for different structural motifs , such as truncated octahedral ( to ) , cuboctahedral ( co ) , icosahedral ( ih ) and decahedral ( dh ) geometries .
we found that o@xmath0 prefers to bind on top of two metal atoms , parallel to the cluster , with an average chemisorption energy of 1.09 ev ( pt - ni ) , 1.07 ev ( pt ) and 2.09 ev ( ni ) , respectively .
the largest adsorption energy values are found to be along the edges between two neighbouring ( 111)/(111 ) and ( 111)/(100 ) facets ; while fcc and hcp sites located on the ( 111 ) facets may show a chemisorption value lower 0.3 ev where often fast o@xmath0 dissociation easily occurs .
our results show that , even though it is difficult to disentangle the geometrical and electronic effects on the oxygen molecule adsorption , there is a strong correlation between the calculated general coordination number ( gcn ) and the chemisorption map . finally , the inclusion of dispersion corrections ( dft - d ) leads to an overall increase on the calculated adsorption energy values but with a negligible alteration on the general o@xmath0 adsorption trends .
|
it is known that various instabilities occur in very luminous stars .
most luminous stars called luminous blue variables ( lbvs or s dor variables ) show major events ( sd(s dor)-eruptions and sd - phases ) with which mass loss from the star is thought to be enhanced greatly ( see e.g. , * ? ? ?
* ; * ? ? ?
* for reviews ) .
their distribution on the hr diagram seems to be bounded by the humphreys - davidson ( hd ; * ? ? ?
* ) limit .
@xcite found that the instability boundary for the radial strange modes roughly coincides with the hd limit .
they proposed that the strange mode instability would yield a strong mass loss causing the hd limit .
several nonlinear analyses for radial strange modes have been performed @xcite to generate pulsation driven mass - loss , but the results seem to remain inconclusive .
( in this context , it is interesting to note that @xcite found a luminous b star to change its mass - loss rate on a timescale of the period of photometric and spectroscopic variation , suggesting a correlation between mass - loss rate and pulsation . )
it is fair to mention that various instabilities other than strange mode instability are also proposed for the cause of eruptive mass loss ( see e.g. , * ? ? ?
* for a review ) .
in the present paper ( [ sec : secular ] ) we will discuss the presence of a monotonously unstable mode in models more luminous than the hd limit .
in addition to the major variations ( sd - eruptions and sd - phases ) on timescales of years and decades , lbvs show quasi - periodic microvariations on timescales of weeks to months .
those variations occur also in non - lbv luminous oba supergiants .
they are loosely called @xmath3 cyg variables , and the variations are thought to be caused by stellar oscillations . @xcite and @xcite claimed these microvariations to be caused by radial strange mode pulsations .
@xcite argued , however , that periods of some of the microvariations in lbvs are much longer than the periods of strange modes , and that the positions of some of those variables on the hr diagram contradict the strange mode instability region .
they concluded microvariations in lbvs to be consistent with nonradial g - modes . in the present paper , we find those longer period variations can be interpreted by oscillatory convection ( nonadiabatic g@xmath1 ) modes associated with the envelope convection zone generated by the fe - opacity peak around @xmath6k .
the linear convection modes , which are dynamically ( monotonously ) unstable in the adiabatic analysis ( g@xmath1-modes ) , become overstable ( or oscillatory ) if nonadiabatic effects are included as found by @xcite for high - degree ( @xmath7 ) modes .
these modes have not been paid much attention to before , because they are not expected to be observed due to the high @xmath8 values and their amplitude being confined to the convective zone .
the present paper will show some low - degree ( @xmath9 ) overstable convection modes having periods longer than those of radial strange modes can emerge to the stellar surface and hence be observable . before the emergence of the opal opacity @xcite ,
ob - type stars were considered to have mostly radiative envelopes .
but in reality , the fe - opacity peak at @xmath10k in the opal opacity is strong enough to produce a convective zone of considerable thickness in the envelopes of oba stars .
@xcite argued the importance of the sub - photospheric convection in massive stars for various photospheric velocity fields including pulsations and large microturbulence needed in spectroscopic analyses of ob stars .
furthermore , @xcite found evidence of solar - like oscillations in a o - type star that indicates stochastic excitation of oscillations by turbulent convection to work even in hot stars .
there are also less luminous b supergiants which show periodic microvariations with relatively short periods of one to a few days @xcite .
some of these variations can be interpreted as supergiant extension of spb stars ( spbsg ) , which is possible because g - mode oscillations being reflected at a convection zone above a hydrogen burning shell so that strong dissipation in the core is suppressed @xcite .
the present paper discusses the stability of radial and nonradial oscillations in massive main - sequence and post - main - sequence models , and compares periods of excited modes with observed periods of microvariations in supergiants .
in order to obtain unperturbed models for stability analyses covering the upper part of the hr diagram , evolutionary models in a mass range of @xmath11 were calculated from zams to @xmath12 , using a code based on the henyey method with opal opacity tables @xcite , where @xmath13 means initial mass .
the mixing length is assumed to be 1.5 pressure scale heights ( unless otherwise stated ) .
wind mass loss based on @xcite is included for the models with initial mass of @xmath14 .
stellar rotation and core overshooting are disregarded for simplicity .
two sets of chemical compositions are adopted in systematic calculations ; @xmath15 and @xmath16 .
an additional model sequence was calculated for @xmath17 with the composition @xmath18 to clarify the effect of heavy element abundances on the stability of oscillations . .
parameters for selected models at zams , tams and at @xmath19 [ cols="<,^,^,^,^,^",options="header " , ] table [ modelparam ] gives stellar parameters at zams , tams ( the end of main - sequence ) , and at @xmath19 for selected model sequences .
it is apparent that significant mass is lost during evolution for @xmath20 with @xmath5 . for the low metallicity ( @xmath21 ) mass loss
is considerably weaker .
nonadiabatic radial and nonradial pulsation analyses were performed using the formulae described in @xcite and @xcite , respectively .
the lagrangian perturbation of the divergence of the convective flux is neglected .
the temporal variation of radial and nonradial pulsations is expressed as @xmath22 with @xmath23 being eigenfrequency whose imaginary part determines the stability;i.e , the pulsation is excited if the imaginary part of @xmath23 is negative . in numerical calculations
the frequency is normalized as @xmath24 with the gravitational constant @xmath25 , stellar mass @xmath26 , and radius @xmath27 .
although wind mass loss is included in evolutionary models for @xmath28 , for oscillation analyses a reflective mechanical outer boundary condition ( @xmath29 ) is employed because theory for the boundary condition in the presence of wind has not yet been developed .
[ fig : hrd ] shows various instability boundaries ( visibility boundaries for oscillatory convection modes ) obtained in this paper and selected evolutionary tracks for a standard composition of @xmath30 . for comparison ,
positions of periodic and semi - periodic supergiant variables are plotted from various sources of the observational data .
interestingly , most of those variable supergiants reside in the unstable regions .
the short - dashed line in fig . [ fig : hrd ] shows the instability boundary of low - order radial and nonradial p - mode oscillations . in the long nearly vertical region of @xmath31 ,
the kappa - mechanism at the fe opacity bump excites low - order modes , which correspond to @xmath32 cep variables @xcite .
the instability boundary bends and becomes horizontal due to strange mode instability @xcite which occurs when the luminosity to mass ratio is sufficiently high ( @xmath33 ) and radiation pressure is dominant at least locally in the envelope .
the instability boundary is essentially the same as that obtained by @xcite for @xmath5 . in the hottest part
, the strange modes are associated with the fe opacity peak , while in relatively cooler part the contribution from the opacity peak at the second helium ionization becomes important .
the vertical part around @xmath34 is the well known blue edge of the cepheid instability strip .
a narrow vertical region indicated by a long - dashed line around @xmath35 in fig . [ fig : hrd ] indicates the excitation of relatively high order radial and nonradial modes due the strange mode instability associated with the hydrogen ionization zone .
we will discuss these modes in [ sec : high ] .
monotonously unstable modes exist above the dotted line in the most luminous part of fig .
[ fig : hrd ] , which will be discussed in
[ sec : secular ] . in the regions surrounded by blue and red solid lines , nonradial
g - modes are excited by the fe opacity bump ; red and blue lines for @xmath36 and @xmath37 modes , respectively .
those g - modes can be excited even in the post - main - sequence models if the mass exceeds @xmath38 , because a convection zone above the hydrogen burning shell can block some g - modes from penetrating to the core .
the dash - dotted lines in fig . [ fig : hrd ] are the ranges where oscillatory convection modes would be observable ; red and blue lines are for @xmath36 and @xmath37 modes , respectively .
the observability is determined based on the ratio between the photospheric amplitude and the maximum amplitude in the interior , which is discussed below .
[ conv_amp ] shows distributions of radial displacements ( @xmath39 , bottom panels ) and kinetic energy density ( @xmath40 , top panels ) of oscillatory convection modes of @xmath36 in two ( hot and relatively cool ) models from the @xmath41 ( @xmath5 ) evolutionary sequence .
two convection modes for each model are shown with selected physical quantities .
a convection zone can be recognized as a zone where the ratio of radiative to total energy flux is less than unity;i.e . , @xmath42 .
the fe - convection zone occurring in a temperature range of @xmath43 has a considerable thickness but contains very small mass ( @xmath44 for the hotter model , and @xmath45 for the cooler model ) . because the gas density is very low in the envelope of a massive star , the convective flux is less than 50% of the total energy flux .
in addition , the low density makes the gas pressure much smaller than the radiation pressure ; i.e. , @xmath46 with @xmath32 being the ratio of gas to total pressure .
two modes shown for each model in fig .
[ conv_amp ] are the two shortest period convection modes that are most likely observable ; i.e. , ratios of the photospheric amplitude to the maximum amplitude in the interior are largest ( see also fig .
[ conv_te ] ) .
the numbers in parentheses in the bottom panels of fig .
[ conv_amp ] are normalized ( by @xmath47 ) complex eigenfrequencies ; the real part represents oscillation frequency , and imaginary part growth or damping rate .
when the imaginary part is _ negative _ the oscillation grows .
the eigenfrequencies shown in fig . [ conv_amp ] indicate that the growth times are comparable to the periods .
the periods are comparable to the periods of g - modes .
the kinetic energy of the oscillatory convection modes is confined to and slightly above the convection zone in the hotter model , while it is well confined to the convection zone in the cooler model . for the convection modes shown in this figure ,
the oscillation amplitude at the stellar surface is comparable to that in the convective zone , indicating those modes are very likely observable .
since oscillatory convection modes have short growth - times , we expect that they have substantial amplitudes in the convection zone .
for such a mode to be observable , however , the amplitude at the stellar surface should be , at least , considerable relative to the amplitude in the convection zone .
we assume that the possibility of detecting these convection modes can be measured by the ratio of the photospheric amplitude to the maximum amplitude in the interior .
the ratio differs for different modes in a model and changes as the stellar parameters change .
the top panel of fig .
[ conv_te ] shows variations of the ratio as a function of effective temperature along the evolutionary track of @xmath41 for a few convection modes of @xmath36 ( triangles ) and @xmath37 ( squares ) .
this figure shows that although many convection modes exist in a model , only one or two highest - frequency modes ( for each @xmath8 ) have large values of the amplitude ratio and hence are potentially observable .
generally , the amplitude ratio of each mode has a broad peak as a function of effective temperature , and the peak value tends to be larger for a mode with larger frequency ( bottom panel ) and larger growth rate ( middle panel ) . in other words ,
the visibility is highest for largest frequency modes with highest growth rates in a certain effective temperature range ( @xmath48 for the @xmath41 models ) .
[ conv_cont ] shows contours of the amplitude ratio for the most visible convection modes for @xmath36 ( solid lines ) and @xmath37 ( dashed lines ) . generally , the visibility of the oscillatory convection modes is better in more luminous and hotter models .
we assume in this paper that these modes would be well visible when the ratio exceeds @xmath49 .
the contours for the ratio of @xmath49 are shown in fig .
[ fig : hrd ] . the top panel of fig .
[ conv_cont ] shows the contours obtained for models with @xmath50 . since the opacity is much lower than that for the case of @xmath5 , the convection zone around the fe opacity peak is less extensive .
nonetheless , observable oscillatory convection modes with low degree @xmath51 persist , although the contour for a value of the amplitude ratio shifts upward by @xmath52 as compared to the case of the standard composition .
this indicates that oscillatory convection modes can cause observable variations even in the smc if the stars are luminous enough .
the growth rates of the oscillatory convection modes ( middle panel , fig.[conv_te ] ) are very high , especially in relatively cooler models ; they tend to decrease as @xmath53 increases .
the reason is probably that in hotter models kinetic energy distribution shifts outward and hence a large fraction of the kinetic energy resides above the convective zone boundary losing the driving .
it has been thought that the pulsation modes with the highest growth rates are associated with strange modes ( eg . * ? ? ?
* ; * ? ? ?
the growth rates of the convection modes are even larger than those of radial strange modes in massive stars @xcite .
the oscillation frequencies of convection modes lie in the g - mode range as seen in the bottom panel of fig .
[ conv_te ] , indicating the periods of convection modes are longer than radial modes .
[ conv_q ] shows q - values of convection modes with the amplitude ratio larger than @xmath49 ( filled symbols ) compared with excited radial modes ( open circles ) for some selected mass models , where the q - value is defined as @xmath54 with @xmath55 and @xmath56 being , respectively , pulsation period and mean density ( see e.g. * ? ? ?
the q - value of a convection mode decreases rapidly as the effective temperature decreases in the hottest and the coolest @xmath53 ranges , while it changes little in the intermediate range of effective temperature . at a given effective temperature ,
q - values of oscillatory convection modes tend to be smaller in more massive models . in very massive ( @xmath57 ) models ,
radial pulsations are excited even in cooler ( @xmath58 ) models due to mainly the strange mode effect that works around the second he ionization zone @xcite .
the q - values of these strange modes increases as the effective temperature decreases ; i.e. , as the depth of the he ii ionization zone increases .
[ conv_q ] indicates that periods of oscillatory convection modes are much longer than those of radial pulsations in hotter models , although in cooler and massive models the differences are not very large .
@xcite found that the periods of microvariations of lbvs are orders of magnitudes longer than those predicted for strange modes by @xcite .
this indicates that oscillatory convection modes would be responsible for long - period microvariations in lbvs ( see [ obscomp ] below ) .
red inverted triangles in fig .
[ conv_te ] present the property of oscillatory convection modes of @xmath36 in models calculated with a mixing - length of _ one pressure scale - height _ rather than @xmath59 pressure scale - hight adopted in standard models .
a smaller mixing - length makes the super - adiabatic temperature gradient in the fe - convection zone larger . in the models with the smaller mixing - length , real parts of eigenfrequencies of oscillatory convection modes tend to be smaller , and
the amplitude tends to be slightly more confined to the convection zone .
fortunately , the qualitative property of the oscillatory convection modes is insensitive to the mixing - length parameter . in very massive models ( @xmath60 )
a monotonously unstable radial mode ( with a purely imaginary eigenfrequency ) is found in the range on the hr diagram indicated by a dotted line in fig .
[ fig : hrd ] . fig .
[ growth_sec ] shows growth time as a function of effective temperature for those unstable modes in models with different initial masses .
the growth times tend to be shorter for more massive stars , ranging from a month to a hundred years , which are much faster than the evolutionary change .
it is interesting to note that the range of the growth times includes the timescales of s dor ( sd)-phase @xcite . to the author s knowledge ,
the presence of monotonously unstable modes in very massive stars was not known before .
the presence was unrecognized in the non - linear radial pulsation analyses performed before by various authors , probably because the monotonously unstable mode was concealed by the more fast growing strange mode pulsations .
amplitude and kinetic energy distribution of the monotonously unstable mode in a model with @xmath17 are shown in fig .
[ fig : secmode ] . the amplitude and kinetic energy
have a peak around the bottom of the convective zone generated by the fe - opacity peak ; in this model the convective zone ranges from @xmath61 to @xmath62 .
although , the kinetic energy , as well as amplitude , is confined around the bottom of the convection zone , the amplitude , after attaining a minimum at @xmath63 , gradually increases toward the surface above the convection zone . in a substantial range in this model
the ratio of the gas to radiation pressure is very small ( i.e. , @xmath46 ) .
then , @xmath64 where @xmath65 is radiative luminosity and @xmath66 the local eddington luminosity defined as @xmath67 with @xmath68 being the velocity of light and @xmath69 the mass within the sphere of radius @xmath70 .
this suggests that the monotonously unstable mode arises because the radiative luminosity is very close to the local eddington luminosity in a substantial range of the envelope .
[ fig : beta ] shows the distribution of @xmath32 in selected models , where solid lines are used for models which have a monotonously unstable mode , while dashed lines are for models without it .
there are two minima of @xmath32 at @xmath71 ( fe opacity peak ) and at @xmath72 ( he ii ionization ) .
apparently , @xmath32 must be well below 0.1 in a range including the both opacity peaks for a monotonously unstable mode to appear . note that the hotter model of @xmath73 ( black dashed line ) has a distribution of @xmath32 around @xmath74 similar to that of the cooler model ( black solid line ) , but @xmath32 in the range @xmath75 is significantly higher than that in the cooler model , which inhibits a monotonously unstable mode in the hotter model . with a low - metal abundance of @xmath21 no monotonously unstable mode
is found even in models of @xmath76 .
the reason is obvious from the blue dashed line in fig .
[ fig : beta ] ; @xmath32 at the fe opacity peak is substantially larger , although a further increase in mass seems to cause monotonously unstable modes .
the presence of such a monotonously unstable mode probably corresponds to the presence of an optically thick wind as discussed by @xcite and @xcite for wr stars .
it is known that optically thick winds also occur in nova models . @xcite
have shown that for non - extreme nova cases both static models and models with winds are possible .
the presence of a monotonously unstable mode in a massive evolutionary model might indicate that the static model transits to a model with an optically thick wind .
it is interesting to note that the boundary in the hr diagram for the presence of monotonously unstable modes ( dotted line in fig .
[ fig : hrd ] ) roughly coincides with the humphreys - davidson ( hd ) limit @xcite ; this may suggest the hd limit to be related to the presence of optically thick winds . in the narrow vertical region around @xmath77
seen in fig .
[ fig : hrd ] , relatively high order ( @xmath78 ) radial as well as nonradial modes are excited at the hydrogen ionization zone at @xmath79 . since the normalized frequency of the excited mode increases as the effective temperature decreases ,
the red - edge is formed where the frequency exceeds the critical frequency given as @xmath80 with the pressure scale height @xmath81 and @xmath82 .
( if the frequency exceeds the critical frequency , the pulsation is not reflected at the outer boundary and is expected to be dissipated in the outermost layers . ) fig .
[ radhigh ] shows the amplitude distribution and the work curve for a high - order radial mode excited in a model relatively close to the blue edge .
obviously the excitation occurs at @xmath79 , in the hydrogen ionization zone .
the amplitude of the mode is extremely confined to the surface layers , where the radiation pressure is much larger than the gas pressure . since unstable modes of this type
are present even in the non - adiabatic reversible ( nar ; * ? ? ?
* ) approximation where thermal - time is artificially set to be zero , we can identify such a mode as a strange mode trapped above the opacity peak of the hydrogen ionization .
a similar mode with a similar frequency is excited for @xmath36 and @xmath37 .
these excited nonradial modes probably correspond to the trapped modes discussed in @xcite .
in addition to the trapped modes @xcite found several untrapped modes to be excited around the trapped mode frequency , by using the riccati - shooting method discussed in @xcite . the present study using a finite difference method , however , could not find those untrapped modes . according to @xcite
those modes have many nodes with considerable amplitude in the deep interior around @xmath83 .
the finite difference code even with @xmath84 grid points seems ineffective to find such untrapped modes .
fig . [ te_peri_obs16 ] compares periods of radial and nonradial modes excited in @xmath85 and @xmath86 models with observed periods of relatively less luminous ( @xmath87 ) supergiant variables as function of @xmath53 .
those variables consist of most of the periodic b stars analyzed by @xcite , and relatively cooler @xmath3-cygni variables including @xmath3 cyg itself .
most of the periods of the hotter ( @xmath88 ) less luminous supergiants seem consistent with oscillatory convection modes , spb - type g - modes ( spbsg ) , or p - modes . relatively cool ( @xmath89 ) stars in fig .
[ te_peri_obs16 ] are less luminous @xmath3-cygni variables .
they lie in the @xmath53 range where high - order p - modes are excited ( fig .
[ fig : hrd ] ) , although most of the periods are longer than those of high - order p - modes .
probably they are pulsating in g - modes excited in a similar range of @xmath53 as found by @xcite , who found that g - modes having periods of ranging from about 10 to 20 days are excited at the hydrogen ionization zone , although a large fraction of their kinetic energy resides in the deep interior .
no modes can be assigned to the three stars in a range of @xmath90 in fig .
[ te_peri_obs16 ] , which are the coolest members in the data from @xcite ( fig.[fig : hrd ] ) .
the cause of the oscillations in these stars is not clear .
[ te_peri_obs60 ] compares periods of radial and nonradial modes excited in higher mass models of @xmath91 and @xmath73 with observed ones for microvariations in luminous ( @xmath92 ) supergiants which consist of lbvs and luminous @xmath3 cyg variables . in these massive stars ,
radial and nonradial strange modes are excited ( open symbols ) , and oscillatory convection modes ( filled triangles and squares ) are likely observable ( the amplitude ratio is larger than 0.2 ) in a wide range of @xmath53 , while spb - type g - modes are not excited . fig .
[ te_peri_obs60 ] indicates that most of the periods of microvariations of lbvs and luminous @xmath3 cyg variables are consistent with the periods of oscillatory convection modes and strange modes . the comparisons between observed and theoretical periods in figs .
[ te_peri_obs16 ] and [ te_peri_obs60 ] indicate that @xmath3-cygni variables are inhomogeneous .
luminous ones are similar to the microvariations of lbvs which are identified as oscillatory convection modes or strange modes , while less luminous cooler ones , including @xmath3 cyg itself , seem to be g - modes ( and possibly including high - order strange modes ) excited at the hydrogen ionization zone .
since there are some lbvs and @xmath3 cyg variables in smc , it would be useful to discuss the results for low metal abundances .
[ te_peri_compz ] compares periods of radial and nonradial modes excited in models with @xmath17 having different metal abundances .
the difference between the cases of @xmath5 and @xmath93 is small , while the case of @xmath21 is significantly different from the former cases . because of the fe - opacity peak being reduced , strange mode instability for low - order radial and nonradial modes are reduced significantly for the @xmath21 case , while the excitation of spbsg - type g - modes seems to be enhanced in the lowest metallicity case ( the reason is not clear ) .
it is interesting to note that even for the low metallicity of @xmath21 oscillatory convection modes still seem to be observable in a considerable range of @xmath53 ( see also fig .
[ conv_cont ] ) .
[ fig : hrdz004 ] is the same as fig .
[ fig : hrd ] but for @xmath21 ; i.e. , shows instability ranges for radial and nonradial modes and observable ranges for oscillatory convection modes in the hr diagram .
compared to the case of @xmath5 ( fig .
[ fig : hrd ] ) , the instability region for low - order radial and nonradial p - modes ( i.e. , @xmath32 cep instability strip ) has disappeared completely .
the instability boundary for the low - order strange mode instability has shifted upward considerably , which agrees with the result of @xcite . on the other hand ,
less affected are the spbsg - type g - mode instability regions and the observable region of oscillatory convection modes .
the instability region for high - order modes in @xmath94 is not affected because they are excited at hydrogen ionization zone .
positions of some lbvs ( s dor variables ) and @xmath3 cyg variables in smc are also plotted in fig .
[ fig : hrdz004 ] . fig .
[ te_peri_smc ] compares quasi - periods of the microvariations of those stars with theoretical periods predicted for metal poor models with initial masses ranging from @xmath95 to @xmath96 .
in contrast to the galactic and lmc cases , oscillatory convection modes seem to be responsible for none of the smc stars shown in these figures .
[ te_peri_smc ] seems to indicate that periods of low - order strange modes are comparable with those of the microvariations of r40 and r45 ( and probably r42 ) . for the strange mode interpretation to be true , the initial mass must be greater than @xmath97 . fig . [
fig : hrdz004 ] , however , indicates the initial mass of these stars to be @xmath98 ; i.e. , the luminosities of these stars seem too low at least by a factor of two .
therefore , the mode identification for the microvariations of r40 , r42 and r45 is unclear . for s18 , an @xmath3 cyg variable in the smc , only the effective temperature
is indicated in fig .
[ te_peri_smc ] because the quasi - periods of its variability are uncertain .
@xcite found three types of variations on timescales of a few years , @xmath99 days , and a few days .
[ fig : hrdz004 ] and [ te_peri_smc ] indicate the shortest timescale variation to correspond to the spbsg type g - modes . for the longer - periods variations ,
however , no excited oscillation modes can be assigned .
the stability of radial and nonradial modes in massive stars ( @xmath100 ) was investigated .
the @xmath101-mechanism associated with the fe - opacity bump at @xmath10k excites low - order radial and nonradial p - modes in main - sequence and post - main - sequence models ( @xmath32 cep instability strip ) and low - degree high radial order g - modes in post - main sequence models ( supergiant spb stars ; spbsg ) .
interestingly , it is found that under a low - metal composition of @xmath21 , the spbsg instability region still remains , while the @xmath32 cep instability strip disappears . in luminous models with @xmath102 ( for @xmath103 strange mode instability
occurs associated with opacity peaks due to fe and he ii ionizations . for the metal - poor composition
@xmath21 strange mode instability appears only in more luminous models . in a narrow strip of @xmath104 , high - order radial and nonradial pulsations
are excited at the hydrogen - ionization zone .
the pulsation amplitude is strongly confined to the outermost layers .
low-@xmath8 oscillatory convection modes associated with the fe convection zones in massive stars are found .
they are likely observable in sufficiently luminous models in a considerable range of effective temperature .
the oscillatory convection modes tend to have growth times comparable with the periods . their periods are longer than those of strange modes or spbsg - type g - modes , and are comparable to those of long - period microvariations of lbvs and luminous @xmath3 cyg variables
. rapid growth rates of the oscillatory convection modes are consistent with the presence of irregularities in the microvariations .
in addition to the oscillation modes , a monotonously unstable mode is found in luminous models with @xmath105 and @xmath106 .
the boundary in the hr diagram for the presence of the monotonously unstable mode roughly coincides with the hd limit .
the growth rate is more than hundred times faster than the stellar evolution rate .
it probably suggests the presence of an optically thick wind in such a luminous star .
finally , it should be noted that in the present stability analysis the effect of stellar winds is not included in the outer boundary condition . in the future
, we have to clarify how the stellar wind affects the stability of oscillations of massive stars .
i am grateful to alfred gautschy for making sample calculations with an alternative numerical approach to confirm the presence of overstable low - degree convection modes , and for helpful comments on a draft of this paper .
i am also grateful to the anonymous referee for useful comments , which have improved the paper considerably . aerts c. , lefever k. , baglin a. , degroote p. , oreiro r. , vuckovi m. , smolders k. , acke b. , verhoelst t. , desmet m. , et al . , 2010 , a&a , 513 , l11 cantiello m. , langer n. , brott i. , de koter a. , shore s.n . , vink j.s . , voegler a. , lennon d.j .
, yoon s .- c .
, 2009 , a&a , 499 , 279 chernigovski s. , grott m. , glatzel w. , 2004 , mnras , 348 , 192 cox j.p . , 1980 theory of stellar pulsation ( princeton univ .
press ) degroote p. , briquet m. , auvergne m. , simn - daz s. , aerts c. , noels a. , rainer m. , et al .
, 2010 , a&a , 519 , a38 dorfi e.a .
, gautschy a. , 2000 , apj , 545 , 982 dziembowski w.a . , sawiska j. , 2005 , aca , 55 , 195 fadeyev yu.a . , 2010 ,
ast.let . , 36 ,
362 gautschy a. , 2009 , a&a , 498 , 273 gautschy a. , glatzel 1990a , mnras , 245 , 154 gautschy a. , glatzel 1990b , mnras , 245 , 597 glatzel w. , 1994 , mnras , 271 , 66 glatzel w. , kiriakidis m. , 1993 , mnras , 263 , 375 godart m. , noels a. , dupret m .- a . , lebreton y. 2009 , mnras , 396 , 1833 grott m. , chernigovski s. , glatzel w. , 2005 , mnras , 360 , 1532 humphreys r.m . ,
davidson k. , 1979 , apj , 232 , 409 humphreys r.m . , davidson k. , 1994 , pasp , 106 , 704 iglesias c.a .
, rogers r.j . , 1996 , apj , 464 , 943 kato m. , iben i.jr , 1992 , apj , 394 , 305 kato m. , hachisu i. , 2009 , apj , 699 , 1293 kiriakidis m. , el eid m.f . , glatzel w. , 1992 , mnras , 255 , 1p kiriakidis m. , fricke k.j . ,
glatzel w. , 1993 , mnras , 264 , 50 lamers h.j.g.l.m . ,
bastiaanse m.v . ,
aerts c. , spoon h.w.w . , 1998 ,
a&a , 335 , 605 lefever k. , puls j. , aerts c. , 2007 , a&a , 463 , 1093 lucy l.b . , 1976 , apj , 206 , 499 moskalik p. , dziembowski w.a . , 1992 a&a , 256 , l5 nugis t. , lamers h.j.g.l.m . , 2002 , a&a , 389 , 162 pamyatnykh a.a . , 1999 , acta astron , 49 , 119 rogers , f.j . , iglesias , c.a . , 1992 , apjs , 79 , 507 saio h. , cox j.p . , 1980 , apj , 236 , 549 saio h. , baker n.h . , gautschy a. , 1998 , mnras , 294 , 622 saio h. , kuschnig r. , gautschy a. , cameron c. , walker g.a.h .
, matthews j.m . , et al . , 2006 ,
apj , 650 , 1111 saio h. , winget d.e . , robinson e.l . , 1983 , apj , 265 , 982 shibahashi h. , osaki y. , 1981 , pasj , 33 , 427 schiller f. , przybilla n. , 2008 , a&a , 479 , 849 van genderen a.m. , 2001 , a&a , 366 , 508 van genderen a.m. , sterken , c. , 2002 , a&a , 386 , 926 van genderen a.m. , sterken c. , jones a.f . , 2004 , a&a , 419 , 667 van leeuwen f. , van genderen a.m. , zegelaar , i. , 1998 , a&as , 128 , 117 vink j.s . ,
de koter a. , lamers h.j.g.l.m . , 2001 , a&a , 369 ,
574 waelkens c. , aerts c. , kestens e. , grenon m. , eyer l. , 1998 , a&a , 330 , 215
|
in order to understand the periodic and semi - periodic variations of luminous o- b- a - type stars , linear nonadiabatic stability analyses for radial and nonradial oscillations have been performed for massive evolutionary models ( @xmath0 ) .
in addition to radial and nonradial oscillations excited by the kappa - mechanism and strange - mode instability , we discuss the importance of low - degree oscillatory convection ( nonadiabatic g@xmath1 ) modes .
although their kinetic energy is largely confined to the convection zone generated by the fe opacity peak near @xmath2k , the amplitude can emerge to the photosphere and should be observable in a certain effective temperature range .
they have periods longer than those of the radial strange modes so that they seem to be responsible for some of the long - period microvariations of lbvs ( s dor variables ) and @xmath3 cyg variables .
moreover , monotonously unstable radial modes are found in some models whose initial masses are greater than or equal to @xmath4 with @xmath5 .
the monotonous instability probably corresponds to the presence of an optically thick wind .
the instability boundary roughly coincides with the humphreys - davidson limit .
[ firstpage ] stars : evolution stars : oscillations stars : massive stars : early - type supergiants .
|
it is lyapunov instability which makes statistical mechanics possible@xcite . for stationary boundary conditions , either equilibrium or nonequilibrium , the _ exponential growth _ , @xmath0 , of small phase - space perturbations @xmath1 or @xmath2 , `` sensitive dependence on initial conditions '' , provides longtime averages independent of initial conditions .
in addition to the coordinates and momenta @xmath3 , time - reversible `` thermostat variables '' @xmath4 can be used to impose _ nonequilibrium _ boundary conditions@xcite , such as velocity or temperature gradients .
a prototypical nonequilibrium problem simulates heat flow between two thermal reservoirs maintained at temperatures @xmath5 and @xmath6 . the resulting heat flow through an internal newtonian region , bounded by the two reservoirs ,
can then be studied and characterized@xcite . to describe the lyapunov instability for any of such systems , equilibrium or nonequilibrium , imagine the deformation of a small phase - space hypersphere @xmath7 , comoving with , and centered on , a deterministic `` reference trajectory '' @xmath8 .
as the motion progresses the hypersphere will deform , at first becoming a rotating hyperellipsoid with the long - time - averaged exponential growth and decay rates of the principal axes defining the lyapunov spectrum @xmath9 .
much later the highly - distorted volume element , through the repeated nonlinear bending and folding caricatured by smale s `` horseshoe '' mapping , combined with an overall dissipative shrinking , comes to occupy a multifractal strange attractor .
though the apparent dimensionality of this steady - state attractor varies from point to point , it can be characterized by an overall averaged `` information dimension '' which is necessarily smaller , for stability of the phase - space flow ( with @xmath10 ) than is the phase - space dimension itself@xcite . the time - averaged exponential growth and decay of phase volume
have long been described by a set of orthonormal lyapunov vectors @xmath11 , one for each lyapunov exponent @xmath12 .
lyapunov _ spectra _ , sets of time - averaged local exponents , @xmath13 , for a variety of both small and large systems have been determined based on work pioneered by stoddard and ford@xcite , shimada and nagashima@xcite , and benettin s group@xcite .
the summed - up lyapunov spectrum can have thermodynamic significance , corresponding to the loss rate of gibbs entropy ( the negative of the entropy gain of the thermal reservoir regions ) in nonequilibrium systems interacting with nos - hoover thermostats , @xmath14 .
the relative sizes of the phase - space components of the vector @xmath15 associated with the largest instantaneous lyapunov exponent , @xmath16 , allows instability sources ( `` hot spots '' , or better , `` regions '' ) to be located spatially .
this localization of instability was long studied by lorenz in his efforts to understand the predictability of weather .
recently ideas which had been expressed much earlier by lorenz@xcite and eckmann and ruelle@xcite have been developed into several algorithms describing the phase - space deformation with an alternative set of `` covariant vectors '' , vectors which `` follow the motion '' in a precisely time - reversible but somewhat arbitrary way@xcite .
the literature describing this development is becoming widespread while remaining , for the most part , overly mathematical ( lots of linear matrix algebra ) and accordingly hard to read . in order better to understand this work , we apply both the older gram - schmidt and the newer covariant time - reversibility ideas to three simple harmonic - oscillator problems . to begin
, we describe the three example problems in section ii , along with the usual gram - schmidt method for finding local lyapunov exponents and corresponding vectors .
some of the newer covariant approaches are outlined in section iii .
numerical results for the three problems , followed by our conclusions , make up the last two sections , iv and v.
the models studied here are all harmonic oscillators . all of them incorporate variations on the textbook oscillator problem with coordinate @xmath17 , momentum @xmath18 , and motion equations @xmath19 .
the first and simplest model@xcite can be analyzed analytically .
a @xmath20 phase - space orbit of this oscillator is shown in figure 1 .
notice that the oscillator orbit shown there includes an arbitrary scale factor @xmath21 , here chosen equal to 2 .
the corresponding hamiltonian is @xmath22 : @xmath23 the `` dynamical matrix '' @xmath24 for this oscillator describes the evolution of the set of infinitesimal offset vectors @xmath11 needed to define the lyapunov exponents , @xmath25 in this case the matrix is @xmath26 \ .\ ] ] two local lyapunov exponents , which reflect the short - term tendency of two orthonormal offset vectors to grow or shrink , can then be defined by the two vector relations : @xmath27 @xmath28 we choose the `` infinitesimal '' length of the vectors , arbitrary in this linear problem , equal to unity for convenience .
snapshots of the two orthogonal gram - schmidt vectors are shown at the left in figure 2 .
these local exponents have time - averaged values of zero , but sizable nonzero time - dependent and @xmath21-dependent fluctuations , @xmath29 the ( scalar ) lagrange multipliers @xmath30 and @xmath31 constrain the _ lengths _ of the offset vectors @xmath15 and @xmath32 while @xmath33 constrains the _ angle _ between the vectors .
the length - constraining multipliers define the local gram - schmidt lyapunov exponents .
their time averages give the global two lyapunov exponents , @xmath34 and @xmath35 : @xmath36 this lagrange - multiplier approach is the small timestep limit of the finite - difference gram - schmidt approach to orthonormalization@xcite . for a model with an @xmath37-dimensional phase space @xmath37 lagrange multipliers
@xmath38 constrain the fixed _ lengths _ of the @xmath37 offset vectors while @xmath39 additional multipliers @xmath40 are required to keep the @xmath39 _ pairs _ of vectors orthogonal . in the following two subsections we describe the analytic formulation of the lagrange - multiplier problem for examples with three- and four - dimensional phase spaces . for large @xmath37
the gram - schmidt approach is considerably faster and simpler than the lagrange - multiplier one . the second model , unlike the first , can exhibit long - term chaotic motion , with three - dimensional phase - space perturbations @xmath42 growing or shrinking exponentially in time , @xmath0 . here
@xmath43 is a friction coefficient and controls the instantaneous changing kinetic temperature @xmath44 , so as to match a specified target temperature @xmath45@xcite .
here we allow the temperature to depend upon the oscillator coordinate@xcite , @xmath46 the temperature gradient makes overall _ dissipation _ possible , characterized by a shrinking phase - space volume , @xmath47 , and resulting in a strange attractor , with @xmath48 , or even a one - dimensional limit cycle , @xmath49 .
the @xmath41-dependent temperature gradient opens the irresistable possibility for heat to be absorbed at higher temperatures than those where it is expelled .
the nos - hoover equations of motion for the nonequilibrium oscillator are : @xmath50 the friction coefficient @xmath43 allows the long - time - averaged kinetic temperature , proportional to @xmath51 , to conform to a nonequilibrium steady state , characterized by the imposed temperature profile @xmath45 : @xmath52 because the motion occurs in a three - dimensional phase space the dynamical matrix @xmath24 , which governs the motion of phase - space offset vectors is @xmath53 : @xmath54 = \left [ \begin{array}{c c c } 0 & 1 & 0 \\ -1 & -\zeta & -p \\
-t^\prime & 2p & 0 \end{array } \right ] \ , \ ] ] where @xmath55 is the derivative of temperature with respect to @xmath17 : @xmath56 for simplicity , we choose these vectors to have unit length .
in addition , _ six _ lagrange multipliers are required to maintain the orthonormality of the three offset vectors @xmath57 : @xmath58 \ ; \ ] ] @xmath59 this model can exhibit chaos or regular behavior , depending on the initial conditions as well as the maximum temperature gradient @xmath60 .
chaotic solutions for this oscillator are necessarily numerical , rather than analytical , and are illustrated in section iii .
the nos - hoover oscillator , with a single friction coefficient @xmath43 ( model 2 ) , is not ergodic . for relatively small temperature gradients
it exhibits an infinity of regular solutions bathed in a chaotic sea . for a glimpse of the details
see reference 24 .
this geometric three - dimensional complexity can be reduced at the price of introducing a second thermostat variable .
the resulting _ four - dimensional _ model , as well as a similar extension treated in reference 18 , has _ two _ friction coefficients @xmath61 rather than just one . with a constant temperature @xmath62 the resulting ergodic
canonical - ensemble phase - space probability density is @xmath63 the _ nonequilibrium _ multifractal extension of this model results if ( as in model 2 ) the temperature depends upon the oscillator coordinate , @xmath64 . in this nonequilibrium case
there are four equations of motion : @xmath65 the corresponding four - variable dynamical matrix @xmath24 , which controls the linearized ( `` tangent - space '' ) motion , @xmath66 , is @xmath67 & -p & -p^3 \\ -t^\prime & 2p & 0 & 0 \\
-3p^2 t^\prime & [ 4p^3 - 6pt ] & 0 & 0 \end{array } \right ] \ .\ ] ] a lower - triangular array of constraining lagrange multipliers , can then be defined .
just as before , the diagonal multipliers maintain the lengths of the vectors constant and the off - diagonal multipliers maintain the orthogonality of the vectors .
for this model , with four offset vectors , we have ten lagrange multipliers in all : @xmath27 @xmath68 @xmath69 @xmath70 the gram - schmidt lyapunov exponents are the long - time - averaged diagonal lagrange multipliers@xcite : @xmath71 here , as is usual , we choose the ( arbitrary ) length of the tangent - space offset vectors equal to unity : @xmath72 .
the instantaneous off - diagonal elements , @xmath73 describe the tendency of the offset vectors to rotate relative to one another .
this four - dimensional model is already sufficiently complex to illustrate the distinctions between the gram - schmidt and covariant descriptions of phase - space instabilities .
we include numerical results for this model too , in section iv .
several loosely - related schemes for evaluating _ covariant _ ( as opposed to gram - schmidt ) lyapunov vectors have been described and explored .
their proponents are mostly interested in the mathematics of weather modeling predictability or in better understanding the statistical mechanical sensitivity to phase - space perturbations . lorenz
famous 1972 talk title : `` predictability : does the flap of a butterfly s wings in brazil set off a tornado in texas ? '' indicates the broad scope of this work@xcite .
wolfe s 2006 dissertation@xcite and the kuptsov - parlitz review@xcite are particularly useful guides to this work .
some of the underlying ideas date back to lorenz early studies , in 1965@xcite .
the main idea is not so different to the old gram - schmidt approach and in fact requires information based on that gram - schmidt ( or lagrange multiplier ) approach not only _ forward _ ( as is usual ) , but also _ backward _ ( which can be artificial , violating the second law of thermodynamics ) in time .
the main idea is to seek a representative basis set of comoving and corotating infinitesimal phase - space vectors @xmath74 , ( @xmath75 for _ covariant _ ) guided by the linearized ( treating @xmath76 as constants in the @xmath24 matrix ) motion equations , @xmath77 the covariant basis vectors follow the linearized flow equations , _ without _ lagrange multipliers , and so are generally _ not _ orthogonal . provided that the `` unstable '' manifold , made up of the phase - space directions corresponding to longtime expansion , can be usefully distinguished from the `` stable '' manifold , corresponding to directions associated with longtime contraction , the covariant basis vectors are locally parallel to these two manifolds
. a `` covariant '' set of vectors would seem _ not _ to require gram - schmidt constraints , or lagrange multipliers , because orthogonality is not required .
even so , all the existing computational algorithms which have been developed to find covariant vectors begin by finding the orthonormal gram - schmidt vectors .
_ two sets _ of gram - schmidt vectors are the next requirement , the usual forward - in - time vectors @xmath78 and the not - so - usual backward - in - time set @xmath79 .
this second set of vectors is obtained by post - processing the time - reversed phase - space trajectory . in time - reversible energy - conserving hamiltonian mechanics
the reversed trajectory can be as `` natural '' as the forward trajectory .
in dissipative systems , with phase - space shrinkage , the stored and reversed trajectory is typically `` unnatural '' , and would violate the second law of thermodynamics .
unlike the gram - schmidt forward and backward vectors [ we will denote them by @xmath78 and @xmath79 ] , the covariant vectors @xmath80 are defined so as to be _ identical _ , at least for hamiltonian mechanics , in the two time directions , that is , `` covariant '' .
the gram - schmidt behavior , with the forward and backward vectors generally different , is a symmetry breaking whose source is not at all apparent in the underlying time - symmetric differential equations of motion .
the differential equations for the time development of the covariant vectors account for the simultaneous stretching and rotation of an infinitesimal phase - space hypersphere .
evidently the _ maximum _ growth ( and maximum shrinkage ) rates correspond to the usual largest lyapunov exponents forward and backward in time , @xmath81 and @xmath82 in an @xmath37-dimensional phase space . another covariant offset vector parallels the local direction of the phase - space velocity , @xmath83 . the remaining @xmath84 covariant vectors require more work .
the difference between the gram - schmidt and covariant phase - space vectors is illustrated for the simple harmonic oscillator in figure 2 .
the relatively readable wolfe - samelson approach@xcite begins with the gram - schmidt sets @xmath85 , forward and backward in time .
then covariant unit vectors are associated with the long - time - averaged individual lyapunov exponents , beginning with the second ( or , using time symmetry , beginning with the next - to - last and working backwards ) .
the covariant vectors are then expressed as linear combinations of ( some of ) the forward and backward gram - schmidt vectors .
the problem is overdetermined , in that @xmath86 gram - schmidt vectors are used as bases for the @xmath37 covariant vectors .
second _ covariant vector can be written as a linear combination of the first two gram - schmidt vectors : @xmath87 the constants @xmath88 are then determined by solving a relatively simple eigenvalue problem . likewise , the next - to - last covariant vector can be written in terms of the first two gram - schmidt vectors from the time - reversed trajectory : @xmath89 in general the @xmath90th covariant vector can be expressed as a sum of @xmath90 gram - schmidt vectors with the constants @xmath91 determined by solving a set of linear equations .
all of these vectors are _ unit vectors _ , with length 1 .
the many method variations ( using the first few , the last few , or some of _ both _ sets of gram - schmidt vectors ) are discussed in wolfe s thesis , which is currently available online .
the appendix of romero - bastida , paz , lpez , and rodriguez work@xcite , as well as the kuptsov - parlitz review@xcite are also useful guides .
the main difficulty in putting all of this work into perspective is a result of the fractal / singular nature of the phase - space vectors .
this structure can be traced to bifurcations in the past history and/or in the future evolution of a particular phase point .
this sensitivity to initial conditions means that slightly different differential - equation algorithms can lead to qualitatively different trajectories , making it hard to tell whether or not two computer programs are consistent with one another .
the best approach to code validation is to reproduce properties of the covariant vectors which are insensitive to the integration algorithm .
we will consider some of these properties for our three example oscillator models .
the first of our three illustrative oscillator problems is an equilibrium problem in hamiltonian mechanics , a harmonic oscillator with unit frequency and with a scale factor @xmath21 set equal to 2 : @xmath92 the local lyapunov exponents for such an oscillator depend upon @xmath21 and vanish in the usual case where @xmath93 .
a @xmath20 phase - space plot of a typical orbit for @xmath94 was shown in figure 1 .
the oscillator orbit we choose to analyze is the ellipse shown there : @xmath95 the local lyapunov exponents describe the instantaneous growth rates of infinitesimal `` offset vectors '' comoving with the flow .
these are of three kinds , orthonormal vectors moving forward in time @xmath78 , orthonormal vectors moving backward in time @xmath79 , and special covariant vectors @xmath80 whose forward and backward orientations in phase space are identical .
the initial direction of each of these vectors is arbitrary because a reversed ( antiparallel ) vector @xmath96 satisfies exactly the same equations and definitions . for the simple
scaled harmonic oscillator it is possible to solve analytically for the gram - schmidt vectors forward and backward in time as well as the covariant vectors and all the corresponding local exponents , @xmath97 once the initial conditions are specified . see figure 2 for an illustration of the following choice of initial values , at @xmath98 : @xmath99 in the course of the motion , with period @xmath100 , the first forward gram - schmidt vector @xmath101 , which is generally identical to the first covariant vector , turns out to be _ always _ radial .
this vector is necessarily perpendicular to @xmath102 .
the vectors @xmath103 and @xmath104 generally differ .
second _ covariant vector is a unit vector which remains always parallel or antiparallel ( and hence `` covariant '' ) to the orbit , @xmath105 generally , all the covariant vectors satisfy the
_ linearized _
small-@xmath106 motion equations .
because the oscillator motion equations _ are _ linear , these oscillator results are correct for finite , not just infinitesimal , vectors .
the gram - schmidt lyapunov exponents can be calculated as lagrange multipliers which maintain the orthonormal relations between @xmath107 in both the forward and the backward directions . for this simple problem
the vectors are unchanged by time reversal , @xmath108 .
the first vector is not constrained in orientation , but is restricted to constant length by the lagrange multiplier @xmath30 : @xmath109 using @xmath30 in either the @xmath110 or @xmath111 equation gives a differential equation for the unit vector @xmath112 , which simplifies with the definition of a new variable , the angle @xmath113 : @xmath114 @xmath115 \longrightarrow\ ] ] @xmath116 although the probability density for @xmath117 [ or @xmath118 ] diverges , whenever @xmath119 [ or whenever @xmath120 ] , the probability density for @xmath121 , which describes the nonuniform phase - space rotary motion , is well - behaved , as shown in figure 3 .
the instantaneous lyapunov vectors , both gram - schmidt and covariant , from the orbit shown in figure 2 , are shown as functions of time in figure 4 .
the exponents are : @xmath122 with the second covariant exponent given by the strain rate parallel to the orbit : @xmath123 \ .\ ] ] figure 4 shows that this covariant exponent is @xmath124 out of phase with the exponent associated with @xmath125 . because the oscillator is not chaotic , these detailed results depend upon the choice of initial conditions .
the _ general _ features illustrated by this problem include ( 1 ) the orthogonality of the gram - schmidt vectors , ( 2 ) the possibility of obtaining a `` reversed '' trajectory by using a stored forward trajectory , changing only the sign of the timestep @xmath126 , ( 3 ) the identity of the first gram - schmidt vector with the first covariant vector , and ( 4 ) the identity of the trajectory direction with the covariant vector corresponding to the time - averaged exponent @xmath127 . in the next problem we study the directions of the forward and backward vectors and show that they typically differ , though the _ time - averaged _ exponents ( for a sufficiently long calculation ) agree , apart from their signs .
here we consider the second model oscillator .
its _ nonequilibrium _ temperature profile , @xmath129 provides both chaotic and regular solutions , depending on the strength of the temperature gradient .
trajectory segments for two values of @xmath41 ( the maximum value of the temperature gradient ) are shown in figure 5 .
for the smaller value @xmath130 , the _ second _ lyapunov exponent vanishes , @xmath131 .
its probability density , shown in figure 6 , is quite different to that of the second covariant exponent , @xmath132 , which remains parallel to the trajectory direction at all times . for the larger limit - cycle value , @xmath133 ,
the longtime orbit is a limit cycle , with period 8.650 . on this limit cycle
the largest lyapunov exponent is identical to the largest covariant exponent , and necessarily vanishes .
the time dependence of this exponent , @xmath134 , with @xmath135 is shown in figure 7 .
the three - dimensional system of motion equations , @xmath136 can be `` reversed '' in either of two ways : ( 1 ) replace @xmath137 by @xmath138 in the fourth - order runge - kutta integrator or ( 2 ) leaving @xmath139 unchanged , change the _ signs _ of the momentum @xmath18 and the friction coefficient @xmath43 so that the underlying physical system traces its coordinate history backward in time , @xmath140 . in either case longtime stability with decreasing time
requires that the dissipative forward trajectory be stored and reused .
if the trajectory is _ not _ stored then the reversed motion abruptly leaves the unstable reversed orbit .
figure 8 shows the jump from the illegal reversed motion , violating the second law of thermodynamics , to a stable obedient motion after sixteen circuits of the illegal cycle . for simplicity of notation and description
we have adopted the first reversibility definition throughout the present work .
it is necessary to recognize that the `` motion '' ( forward in time ) we analyze _ is _ dissipative , though time - reversible , while the reversed motion ( backward in time and contrary to the second law of thermodynamics ) is actually so unstable ( because @xmath141 ) as to be unobservable unless the trajectory has been stored in advance .
the instability of the reversed trajectory , leading to symmetry breaking , can be understood by considering the flow of probability density in phase space .
the comoving probability density @xmath142 obeys the analog of liouville s continuity equation . at any instant of time
the changing phase - space probability density responds to the friction coefficient @xmath43 : @xmath143 \equiv ( \partial \dot q/\partial q ) + ( \partial \dot p/\partial p ) + ( \partial \dot \zeta /\partial \zeta ) = 0 + \zeta + 0 = -\dot \otimes/\otimes = -\lambda^f_{11}(t ) - \lambda^f_{22}(t ) - \lambda^f_{33}(t ) \ .\ ] ] phase - volume expands / contracts when @xmath43 is negative / positive .
the _ time - averaged _ value of the friction coefficient @xmath43 is necessarily positive , and reflects the heat transfer ( from larger to smaller values of the oscillator coordinate @xmath17 ) consistent with the second law of thermodynamics . because this nos - hoover problem is chaotic , no analytic solution is available , making it necessary to compute the vectors and exponents numerically .
lyapunov exponents have long been determined numerically .
spotswood stoddard and joseph ford s pioneering implementation was generalized by shimada , nagashima , and by bennetin s group .
these latter authors used gram - schmidt rescaling of phase - space offset vectors .
this continuous limit of this numerical approach was formalized by introducing lagrange multipliers @xmath40 to constrain the offset vectors orthonormality@xcite .
this problem raises a computational question : how best to distinguish the usual gram - schmidt vectors @xmath78 from the covariant ones @xmath80 ?
some quantitative criterion is required , at the least for reproducibility if not for clarity . in the case of the thermostated oscillator
we have computed the mean - squared projections into @xmath128 space for the three gram - schmidt basis vectors , both forward and backward in time .
the results are essentially the same for either time direction : @xmath144 @xmath145 evidently , at least for this problem , there is little geometric dependence of the gram - schmidt vectors on the presence or absence of chaos .
on the other hand , the second covariant vector ( parallel to the flow ) has mean - squared components @xmath146 for @xmath130 and @xmath147 for @xmath133 , and so is significantly different to the gram - schmidt @xmath103 or @xmath148 .
a shortcoming of work on the covariant vectors so far has been the lack of quantitative information concerning them .
this is an undesirable situation , as it makes checking computations problematic .
we turn next to a four - dimensional problem .
although the first and last gram - schmidt vectors and the trajectory direction give three of the four covariant vectors , the fourth requires new ideas , and illustrates the general case of determining a complete set of covariant basis vectors .
problems with four or more phase - space dimensions require the mathematics , mostly linear algebra , of covariant vectors for their solution . here
we choose a doubly - thermostated harmonic oscillator ( two thermostat variables , @xmath43 and @xmath150 ) , again with a nonequilibrium temperature profile , @xmath151 .
the equations of motion ( which give the canonical phase - space distribution characteristic of the temperature @xmath62 when @xmath41 vanishes ) are : @xmath152 again @xmath62 is the kinetic temperature , the time - averaged value of @xmath44 . @xmath43 and @xmath150 are time - reversible thermostat variables which control the second and fourth moments of the velocity distribution . at equilibrium
the solution of these motion equations is the complete canonical phase - space distribution . for @xmath153
this stationary distribution has the form : @xmath154 \ .\ ] ] this oscillator system is ergodic .
when the temperature @xmath62 is made to depend upon the coordinate @xmath17 , a nonequilibrium strange attractor results .
we can give some detailed results for the well - studied@xcite special case @xmath155 .
for this nonequilibrium problem the lyapunov spectrum forward in time is already known , @xmath156 as is also the strange attractor s phase - space information dimension , @xmath157 = 2.56 , reduced by 1.44 from the equilibrium case ( @xmath158 ) , where the spectrum is symmetric : @xmath159 this information dimension result is in definite violation of the kaplan - yorke conjecture@xcite , @xmath160 the gram - schmidt vectors @xmath161 for this problem are easily obtained using the lagrange multiplier approach , and give three of the four covariant vectors : @xmath162 where @xmath163 is the phase - space trajectory velocity , @xmath164 .
wolfe and samelson show that the only missing covariant vector , @xmath165 , can be expressed as a linear combination of the first three forward ( or the last two backward ) @xmath166 vectors .
either choice requires solving an eigenvalue problem numerically and both choices lead to exactly the same covariant vector @xmath165 , even if the gram - schmidt vectors are not perfectly converged .
a comparison of the various local lyapunov exponents is detailed in figures 9 - 12 .
we have ordered the _ thickness _ of the plotted lines throughout , in the order @xmath167 so that the thickest line represents the distribution for @xmath168 ( mean value + 0.073 ) and , in the backward plot , @xmath169 ( mean value + 0.411 ) .
visually , the data for two billion points , shown here , are indistinguishable from data for two hundred million points , shown in version 1 of this work . with a single exception ,
the vanishing gram - schmidt exponent derived from @xmath170 , the data suggest a fractal nature for the probability densities of the various exponents .
figure 9 compares the probability distributions for @xmath171 and the covariant exponent @xmath132 corresponding to motion along the trajectory .
the gram - schmidt histogram is _ much smoother _ than that of its covariant cousin . at the moment
we have no explanation for this interesting qualitative difference . otherwise the local exponent distributions are qualitatively similar large fluctuations compared to the actual exponent values .
the extra work associated with the local covariant spectrum can not be justified based on these data .
it should be noted that the gram - schmidt exponents forward and backward in time are quite different , reflecting the difference between the future and the past trajectories . to make it possible for the reader to coordinate our results with those of wolfe and samelson@xcite
, we write a set of equations which can be solved in order to find the third covariant vector @xmath165 as an expansion in the _ forward - in - time _ gram - schmidt vectors : @xmath172 \ .\ ] ] here the sums include the two values of @xmath173 .
the numbering system , though arbitrary , is crucial . forward in time the correspondence of the vectors with the long - time - averaged lyapunov exponents is @xmath174 _ backward _ in time the ordering is the same with the _ signs changed _ : @xmath175 of course the _ third _ covariant vector going forward could equally well be viewed as the second going backward using an expansion in terms of the _ backward - in - time _ gram - schmidt vectors .
this alternative approach results in a smaller matrix _ without _ the sum over @xmath176 : @xmath177 \ .\ ] ] here the ordering of the gram - schmidt vectors follows this same new ordering of the backward vectors : @xmath178 the vectors forward in time follow the _ same _ ordering with the _ signs changed_. once the ordering of the vectors is correctly negotiated one can find the eight gram - schmidt vectors ( four in each time direction ) and the four covariant vectors ( the same in either time direction ) . for comparison
we show the probability densities for the vectors in figures 10 , 11 , and 12 .
the covariant phase - space vectors , illustrated here for oscillators , are said to have two advantages : [ 1 ] they display the time - symmetry of the underlying equations of motion and [ 2 ] they provide results which are norm - independent " .
the gram - schmidt vectors differ , reflecting the past and steadfastly ignorant of the future .
consider a purely - hamiltonian situation , the sudden inelastic collision of two rapidly - moving blocks , converting ordered kinetic energy to heat .
the gram - schmidt vectors turn out to be more localized in the forward direction of time than in the ( completely unphysical ) reversed direction , where the particle trajectories have been stored@xcite .
the phase - space directions corresponding to maximum growth and decay ( easy to calculate by singular value decomposition of the symmetrized @xmath24 matrix and relatively smooth in phase space , rather than multifractal ) show the same time symmetry as do the covariant vectors .
this time symmetry seems unhelpful for describing the irreversibility of this purely - hamiltonian shockwave process .
we like the time asymmetry of the gram - schmidt vectors , reflecting as they do the past - based nature of the second law of thermodynamics . despite the norm - independence of the lyapunov vectors
it is apparent that `` local '' ( in phase - space or in time ) lyapunov exponents depend on the chosen coordinate system .
even the simple harmonic oscillator considered here shows coordinate - dependent local exponents .
the time symmetry - breaking associated with time - reversible dissipative systems ( like the three - dimensional and four - dimensional oscillator models studied here ) prevents backward analyses unless trajectories are stored .
lyapunov instability helps _ attract _ a repellor state ( violating the second law ) to the law - abiding strange attractor . like the overall trajectory ,
the gram - schmidt vectors are themselves unstable to symmetry breaking .
a reversed trajectory calculation will not also reverse the gram - schmidt vectors unless those vectors too are stored from a forward trajectory .
the covariant vectors are the same going forward and backward in time .
the symmetry breaking which naturally occurs can only be avoided by repeating a trajectory , making it possible to find the covariant solution . despite the double - precision accuracy of runge - kutta solutions
it is difficult to validate computer programs for dissipative chaotic systems precisely due to their lyapunov instability .
for this reason the squared projections of the offset vectors and the exponent histograms are good choices for validation .
the details of specific program results necessarily vary and reflect the fractal nature of the phase - space singularities describing the inevitable past and future bifurcations .
the fractal nature of these singular points frustrates any attempt to gain accuracy through mesh refinement .
both the covariant and the gram - schmidt exponents share this fractal nature , but differently .
the gram - schmidt vectors are analogous to passengers reactions to a curvy highway while the covariant view is that of a stationary pedestrian observer .
we thank mauricio romero - bastida and antonio politi for useful conversations at a workshop organized by thomas gilbert and dave sanders , held in cuernavaca , mexico earlier this year .
rainer klages , pavel kuptsov , roger samelson , franz waldner , and christopher wolfe generously provided some comments and useful literature references .
we specially thank lakshmi narayanan for stimulating this work by requesting a second edition of our book on time reversibility , and stefano ruffo , for expediting publication in communications in nonlinear science and numerical simulation .
w. g. hoover , c. g. hoover , and j. petravic , `` simulation of two- and three - dimensional dense - fluid shear flows _ via _ nonequilibrium molecular dynamics : comparison of time - and - space - averaged stresses from homogeneous doll s and sllod shear algorithms with those from boundary - driven shear '' , physical review e * 78 * , 046701 ( 2008 ) .
h. bosetti , h. a. posch , ch .
dellago , and w. g. hoover , `` time - reversal symmetry and covariant lyapunov vectors for the doubly thermostated harmonic oscillator '' , physical review e * 82 * , 046218 ( 2010 ) .
g. hoover , c. g. hoover , and f. grond , `` phase - space growth rates , local lyapunov spectra , and symmetry breaking for time - reversible dissipative oscillators '' , communications in nonlinear science and numerical simulation * 13 * , 1180 - 1193 ( 2008 ) .
g. hoover , c. g. hoover , and h. a. posch , `` dynamical instabilities , manifolds , and local lyapunov spectra far from equilibrium '' , computational methods in science and technology * 7 * , 55 - 65 ( 2001 ) .
g. hoover , c. g. hoover , h. a. posch , and j. a. codelli , `` the second law of thermodynamics and multifractal distribution functions : bin counting , pair correlations , and the [ definite failure of the ] kaplan - yorke conjecture '' , communications in nonlinear science and numerical simulation * 12 * , 214 - 231 ( 2005 ) .
g. hoover and c. g. hoover , three lectures : nemd , spam , and shockwaves " , presented at the granada seminar on the foundations of nonequilibrium statistical physics , 13 - 17 september , 2010 : arxiv:1008.4947 .
|
we compare the gram - schmidt and covariant phase - space - basis - vector descriptions for three time - reversible harmonic oscillator problems , in two , three , and four phase - space dimensions respectively .
the two - dimensional problem can be solved analytically .
the three - dimensional and four - dimensional problems studied here are simultaneously chaotic , time - reversible , and dissipative .
our treatment is intended to be pedagogical , for july 2011 publication in communications in nonlinear science and numerical simulation , and for use in the second edition of our book on _ time reversibility , computer simulation , and chaos_. comments are very welcome .
|
the x - ray source 4u 2206 + 54 has been studied with numerous ground and space - based observations , but the nature of this source and origins of its variability patten are still unclear .
4u 2206 + 54 was identified with an optical counterpart bd + 53 2790 by steiner et al .
( 1984 ) which was initially classified as a be star .
however , in optical and uv bands the emission spectrum is complex , in particular the behavior of the h@xmath8 emission line , suggesting that this star is an o9.5v star ( negueruela & reig 2001 ; rib et al .
2006 ; blay et al .
2006 ) with a high he abundance ( blay et al .
2006 ) .
x - ray monitoring of 4u 2206 + 54 by rxte suggested a modulation period of 9.6 days ( corbet & peele 2001 ) which may be an orbit period , but this period disappeared with recent swift / bat observations ( corbet et al .
2007 ) which instead found a modulation of 19.25 days consistent with twice the 9.6-day period .
since there is no circumstellar disc around the donor of 4u 2206 + 54 , the material needed for accretion and production of x - rays should come from the stellar wind .
the observed x - ray luminosity of 4u 2206 + 54 varies from 10@xmath9 erg s@xmath2 from the rxte , bepposax , swift and integral light curves between 1996 and 2005 ( torrejon et al . 2004 ; masetti et al . 2004 ; blay et al . 2005 ; corbet et al .
rib et al .
( 2006 ) found a low wind terminal velocity of @xmath10 km s@xmath2 in 4u 2206 + 54 .
with such a low wind terminal velocity , and assuming an eccentric orbit , one could reproduce the x - ray luminosity and orbit variability of the system ( rib et al .
2006 ) .
the nature of the compact object in 4u 2206 + 54 has been in dispute for a long time ( negueruela & reig 2001 ; corbet & peele 2001 ) .
broad band x - ray observations and radio studies on 4u 2206 + 54 favored the presence of a neutron star ( torrejon et al .
2004 ; blay et al .
recent reports on the possible detection of electron cyclotron resonant absorption line at @xmath11 kev suggested a magnetic field of @xmath12 g by different observations of rxte , bepposax and integral ( torrejon et al .
2004 ; masetti et al . 2004 ; blay et al . 2005 ) . non - detection of pulsation in 4u 2206 + 54 was always used as a doubt on the neutron star scenario .
recently , a possible 5500-s pulsation period in the light curve of 4u 2206 + 54 was discovered with the rxte observations ( reig et al .
if it is true , the compact object in 4u 2206 + 54 will be a neutron star with a spin - period of 5500 seconds . in this work , we will study hard x - ray characteristics of 4u 2206 + 54 with the integral / ibis observations from 2003 may to 2005 december .
then we can show the hard x - ray spectral properties both in quiescent states ( average x - ray luminosity @xmath13 erg s@xmath2 ) and active states ( @xmath14 erg s@xmath2 ) .
two cyclotron resonant absorption lines ( fundamental and first harmonic ) were found in the spectrum of the active states during 2005 dec , while in the quiescent states , no cyclotron line features are detected . the existence of cyclotron absorption lines suggested a strong magnetic field in 4u 2206 + 54 , and the high energy emission in active states would mainly come from magnetosphere near the neutron star .
therefore , it is possible that we could search for a pulsation period from the light curve of 4u 2206 + 54 in active states during 2005 dec with integral / ibis observations . finally , in order to confirm the existence of the orbit period in x - ray binary 4u 2206 + 54 , we also re - analyzed the archival data of the all sky monitor ( asm ) aboard rxte from 1997 to 2008 to search for the orbit period . [ cols="<,^,^,^,<",options="header " , ] ) ; 2003 dec 13 to 31 ( @xmath15 ) ; 2005 dec 5 to 12 ( @xmath16 ) .
false color representation of significance is displayed on a logarithmic scale .
, width=321 ]
the international gamma - ray astrophysics laboratory ( integral , winkler et al .
2003 ) is esa s currently operational space - based hard x - ray / soft gamma - ray telescope .
there are two main instruments aboard integral , the imager ibis ( ubertini et al .
2003 ) and the spectrometer spi ( vedrenne et al . 2003 ) , supplemented by two x - ray monitors jem - x ( lund et al . 2003 ) and an optical monitor omc ( mas - hesse et al .
all four instruments are co - aligned , allowing simultaneous observations in a wide energy range .
4u 2206 + 54 was observed during the integral surveys of the cygnus and cassiopeia regions .
4u 2206 + 54 can not be detected by jem - x which has a smaller field of view ( fov ) , and spi has the largest fov but larger uncertainties compared with ibis , so we only use ibis data for the analysis in this work .
the data were collected with the low - energy array called isgri ( integral soft gamma - ray imager ) which consists of a pixellated 128@xmath17 cdte solid - state detector that views the sky through a coded aperture mask ( lebrun et al .
ibis / isgri has a 12 ( fwhm ) angular resolution and arcmin source location accuracy in the energy band of 15 200 kev . in table 1
, we summarize the integral revolutions in our analysis .
we used the archival data which are available from the integral science data center ( isdc ) .
the analysis was done with the standard integral off - line scientific analysis ( osa , goldwurn et al .
2003 ) software , ver .
7.0 . in this latest version , the energy correction has changed and new calibration laws have been used for the ibis / isgri analysis .
the drift of the energy calibration gain and offset with activation and time are much better calibrated at intermediate energies ( around 50 kev ) .
this version results in more constant crab light curve .
individual pointings processed with osa 7.0 were mosaicked to create sky images according to the methods and processes described in bird et al ( 2007 ) .
and we have used the 20 60 kev band for source detection and to quote fluxes .
we displayed the significance mosaic maps around 4u 2206 + 64 as seen with integral / ibis in the energy range of 20 60 kev during three observational time intervals in figure 1 and the light curve of 4u 2206 + 54 from 2003 may to 2005 december obtained by both ibis observations in figure 2 ( top ) .
we also presented the light curve of 4u 2206 + 54 in the energy band of 1.5 12 kev from the rxte / asm long - term monitoring for a comparison ( figure 2 bottom ) .
the count rate from 4u 2206 + 54 in the energy range of 20 60 kev varied in more than two years : @xmath18 cts s@xmath2 on 2003 may 2 and 2003 july 2 ; and undergoing a quiescent state , @xmath19 cts s@xmath2 around 2003 dec , 2004 feb and 2004 dec ; again an active state , @xmath20 cts s@xmath2 around 2005 dec 5 13 ( also see table 1 ) .
the spectral results on the source 4u 2206 + 54 during the revolutions 67 and 87 with integral / ibis data have been presented in blay et al .
they reported a possible cyclotron absorption line around 32 kev . here
we will not show the spectral results during these time intervals again , but carry out spectral analyses on the data during the quiescent state and the active state around dec 2005 respectively .
the spectral analysis software package used is xspec 12.4.0x ( arnaud 1996 ) . for the quiescent state
, we summed up images from the integral resolutions 142 148 , 161 162 and 262 264 ( see table 1 ) to improve the detection significance level , and then derived the spectrum which was shown in figure 3 .
the 20 150 kev spectrum can be fitted with a power - law model of a photon index @xmath21 .
the obtained flux in the quiescent state from 20 100 kev is @xmath22 erg @xmath23 s@xmath2 , corresponding to a hard x - ray luminosity of @xmath24 erg s@xmath2 assuming a distance of @xmath25 kpc ( blay et al .
no significant cyclotron absorption features were found in quiescence . if assuming the presence of the cyclotron absorption lines at @xmath11 kev ( blay et al . 2005 ) and @xmath26 kev ( see below ) , we derived the upper limits on the equivalent width ( ew ) of the absorption lines : @xmath27 kev at 30 kev ( 2@xmath28 ) and @xmath29 kev at 60 kev ( 2@xmath28 ) .
the same spectral analysis was carried out during the active state of 4u 2206 + 54 from dec 5 to 12 , 2005 .
the derived spectrum was displayed in figure 4 . in the active state
, a simple power - law model can not fit the spectrum , we used a thermal bremsstrahlung model to fit it .
however , possible absorption features around 30 and 60 kev can not be fitted ( also seen in the residuals in figure 4 top panel ) .
therefore , we used the thermal bremsstrahlung model to fit the continuum , and added the cyclotron resonant absorption model by using the xspec cyclabs to the continuum fit ( see figure 3 ) . in this case
, we found a thermal bremsstrahlung model of @xmath30 kev plus two electron cyclotron resonant absorption lines at @xmath31 kev ( f - test probability : 2.9@xmath32 ) with a fwhm of @xmath33 kev and @xmath34 kev ( f - test probability : 4.2@xmath35 ) with a fwhm of @xmath36 kev ( reduced @xmath37 , 6 _ d.o.f _ ) .
the derived ew is @xmath38 kev at @xmath39 kev and @xmath40 kev at @xmath26 kev separately .
the obtained flux from the 20 100 kev continuum during the active state in 4u 2206 + 54 is @xmath41 erg @xmath23 s@xmath2 , corresponding to a hard x - ray luminosity of @xmath42 erg s@xmath2 .
it is the first time that we detected both the fundamental and second harmonic of the cyclotron resonant absorption lines in the x - ray binary 4u 2206 + 54 .
the detection of the line feature at @xmath11 kev also confirmed the previous reports from different measurements ( torrejon et al .
2004 ; massetti et al .
2004 ; blay et al .
discovery of the cyclotron resonant absorption lines at @xmath11 and 60 kev strongly suggested a magnetic neutron star located in the binary 4u 2206 + 54 .
we can calculated the value of the magnetic field of the neutron star in 4u 2206 + 54 by using the formula = [ e_cycl/11.6kev](1+z ) , where @xmath43 is the energy of the fundamental line , here @xmath44 kev , and @xmath45 is the gravitational redshift near the surface of the neutron star . for a canonical neutron star of 1.4 with a radius of 10 km , we can take @xmath46 ( kreykenbohm et al .
so we obtain a magnetic field of @xmath47 g for the neutron star in 4u 2206 + 54 , and this value is still in agreement with those in the previous studies ( torrejon et al . 2004 ; massetti et al . 2004 ; blay et al .
the high mass x - ray binary 4u 2206 + 54 shows a variable x - ray light curve .
intensive searches for the pulsation period of 4u 2206 + 54 have been performed with exosat ( corbet & peele 2001 ) , rxte ( negueruela & reig 2001 ; torrejon et al .
2004 ; corbet et al .
2007 ) , bepposax ( torrejon et al . 2004 ; massetti et al . 2004 ) and integral ( blay et al . 2005 ) . however , these studies suggested the lack of the x - ray pulsation on timescales from @xmath48 ms to @xmath49 hr .
recently , reig et al . (
2009 ) use the new rxte observational data to search for the pulsation period longer than 1 hr , and discovered a possible 5500-s pulsation period in the light curve of 4u 2206 + 54 .
in addition , the early rxte / asm data suggested a modulation period of @xmath50 days ( corbet & peele 2001 ) , but the swift / bat data reported a period of @xmath51 days ( corbet et al .
2007 ) , so that corbet et al . ( 2007 ) suggested the orbit period should be 19.25 days instead of 9.6 days . here
we will check these period reports with recent data of integral / ibis and rxte / asm .
we have reported the cyclotron resonant absorption line features during the active state in 4u 2206 + 54 , suggesting that during the active state , the x - ray emission mainly coming from the surface of the neutron star .
so we try to search for a pulsation period using the light curve data during the active state .
we applied the fft to the observational intervals from dec 5 to 12 , 2005 .
the power spectrum was shown in figure 5 , binned at 100 s intervals .
a significant period signal was found at @xmath52 s ( figure 5 right ) .
this period is a little lower than that reported by reig et al .
( 2009 ) , but still consistent considering the error bar range .
thus , 4u 2206 + 54 should be a x - ray pulsar with a very slow pulsation .
the folded light curve of 4u 2206 + 54 at a pulsation period ( 5400 s ) is also shown in figure 6 .
a possible double main peak feature appears in the pulse profile , one at @xmath53 and the other around 0.6 0.9 . since the available database during the active state
is limited to less than 7 days , we can not search for the longer modulation period ( like orbit period , possible @xmath5 9.6 d or 19.2 d ) with the integral database during the active state of 4u 2206 + 54 .
we used archival rxte / asm data with observations from 1997 2008 to search for the orbit period in 4u 2206 + 54 .
firstly , we averaged the asm 1.5 12 kev light curves into 1 hour bins . after subtracting the mean value of all inhabited bins from each inhabited bin , we take an fft to the light curve .
the power spectrum is shown in figure 7 . in the top panel of fig .
7 , we have shown a power spectrum in a wider frequency range , and no significant signal was found in the high frequency range ( @xmath54 hours ) , while two significant peaks were detected in the lower frequency range .
then in the bottom panel , from the zoom - in of the low frequency band , we determined two modulation periods at 9.56@xmath55 days and @xmath56 days ( twice the former one ) .
so we confirm the existence of the possible orbit periods reported by both the early rxte / asm data ( corbet & peele 2001 ) and the recent swift / bat observations ( corbet et al .
2007 ) . in figure 8 , the folded light curves of 4u 2206 + 54 using the rxte / asm data at two modulation periods at 9.56 day and 19.11 day were also displayed separately .
a possible peak at @xmath57 days also appeared in the power spectrum though not significantly yet , which may be just a noise signal .
in this paper , we studied the hard x - ray emission of the high mass x - ray binary 4u 2206 + 54 with integral / ibis observations to reveal the nature of the compact object .
in hard x - ray bands , 4u 2206 + 54 undergone both the quiescent and active states . in the quiescent state ,
the 20 150 kev spectrum of 4u 2206 + 54 can be fitted by a power - law model with @xmath58 , and no cyclotron absorption line features are found . while , in the active state , two cyclotron absorption lines at @xmath11 kev and 60 kev are discovered in the hard x - ray spectrum . with known the cyclotron resonant absorption energy , we determined the magnetic field of @xmath59 g for the compact object in 4u 2206 + 54 .
analysis of the x - ray light curve during the active state , we also found a modulation period of @xmath60 s which would be a pulsation period for a neutron star .
thus we identified the compact object in 4u 2206 + 54 as a magnetic neutron star with a very slow pulsation . from the asm light curve of 4u 2206 + 54 , we found two modulation periods of @xmath5 9.56 days and the twice , 19.11 days .
then which one is the real orbit period in 4u 2206 + 54 ?
one possibility is that the 19.11-day period is just the first harmonic of the 9.6-day period .
but the swift / bat data only found one period at 19.2 days , and recent rxte / asm data from 2004 2006 showed the modulation period at 19.1 days , not significantly at 9.6 days yet ( corbet et al .
2007 ; we also folded the light curve at 9.6 day using the rxte / asm data after 2003 , no significant features were detected in phase profiles ) .
then the 9.6 day period was not a permanent signal in x - ray light curve .
then appearance of 19.2-day period may not be the first harmonic of 9.6 day .
in addition , optical radial velocity measurements ( blay et al .
2006 ) have not shown the period at 9.6 days but the power spectrum of the complete set of radial velocity observations suggested a peak at @xmath61 days , which was interpreted as an artifact .
modulation of x - ray emission on the orbit period in a high mass x - ray binary may occur in different ways as follows .
firstly , if the orbit is eccentric , the peak emission would occur at periastron passage ( @xmath62 assumed by rib et al . 2006 ) . in this case , it is difficult to understand why two periods appeared in the x - ray light curve .
in addition , an eccentricity of @xmath63 was suggested from the x - ray light curve analysis ( reig et al .
2009 ) , so 4u 2206 + 54 has a modestly eccentric orbit .
in the second case , if the orbit plane is inclined to the plane of an equational circumstellar disc , two possible outbursts could occur when the neutron star passes through the disc .
this scenario has been proposed to explain two modulation periods detected in two x - ray binaries : a be / neutron star binary gro j2058 + 42 ( corbet et al .
1997 ) and a supergiant system 4u 1907 + 09 ( mashall & ricketts 1980 ) .
though 4u 2206 + 54 has the difference in the nature of the mass donor , the orbit behavior may be similar to the case in gro j2058 + 42 and 4u 1907 + 09 .
therefore , we suggested that the orbit period of 4u 2206 + 54 should be @xmath5 19.11 days , which was consistent with the conclusion by corbet et al .
( 2007 ) . in hard x - ray band
, only the orbit period at @xmath64 days could be found with swift / bat observations ( corbet et al .
2007 ) , while in soft x - ray band of 1.5 12 kev , we found two modulation periods .
if the 9.6-day period is due to the neutron star passing through the equational plane of the mass donor , soft x - ray emission may be affected by the enhanced wind around the disc , but hard x - ray emission has no significant dependence on it .
therefore , the hard x - ray emission above 20 kev and its variation pattern should have a different origin from those in the soft x - ray band , which requires more studies .
in the be star / neutron star binary systems , there exists a possible correlation between orbit period and pulse period ( corbet 1986 ) .
though 4u 2206 + 54 have a different mass donor but may be still similar to a be star system , it is predicted to have a pulse period of a few second for a orbit period of 19.11 days according to the correlation , which however is unconsistent with the pulsation period of @xmath5 5400 s reported in this work ( maybe @xmath5 5560 s , also see reig et al .
two high mass x - ray binary systems also show the similar behavior away from the correlation : a0538 - 66 with an orbit period of 16.7 days has an short pulse period of 69 ms ( skinner et al . 1982 ) and sax j2103.5 + 4545 with an orbit period of 12.7 days has a longer pulse period of 358.6 s ( baykal et al .
generally , the 19 day orbit period is relatively short for a be star system .
therefore , the orbit - pulse period correlation would be unvalid at the short orbit period . from pulse period versus orbit period diagram ( see reig et al .
2009 ) , 4u 2206 + 54 fall into the wind - fed supergiant system region which was consistent with the proposed wind - accretion scenario from the x - ray emission properties ( corbet & peele 2001 ; rib et al .
2006 ) .
though we have now confirmed the nature of the compact object in 4u 2206 + 54 , many questions on this high mass x - ray binary are still unclear , requiring further studies in both theories and observations .
origin of the low - pulsation neutron star in 4u 2206 + 54 is a mystery according to the present scenario of the neutron star s spin evolution in a close binary system ( davies & pringle 1981 ; li & van den heuvel 1999 ) .
so reig et al .
( 2009 ) suggested that the neutron star in 4u 2206 + 54 was born as a magnetar with @xmath65 g which was used to explain the long pulsation period of 2s 0114 + 65 ( li & van den heuvel 1999 ) .
since the donor star of 4u 2206 + 54 is a main sequence massive star , the magnetic field of the neutron star decaying from @xmath66 g to 10@xmath67 g during wind accretion processes within several million years is also questionable .
there exist other possible ways to resolve the problems , like x - ray pulsar in 4u 2206 + 54 may not follow the present standard evolution models in close binaries .
some numerical simulations suggested that no significant angular momentum transfers onto the neutron star from the wind of supergiant systems ( ruffert 1999 ) . in the case of 4u 2206 + 54 with a magnetic neutron star ,
it is possible that the calculations ( ruffert 1999 ) were inapplicable , so that the 5000-s pulsation period was formed during the wind - fed accretion phase .
we discovered cyclotron resonant absorption lines at @xmath11 kev 60 kev in the active state of 4u 2206 + 54 , but no detection of cyclotron lines in the quiescent state .
in previous studies , the possible detection of electron cyclotron absorption line at @xmath39 kev was reported by observations of rxte , beppossax and integral(torrejon et al .
2004 ; masetti et al . 2004 ; blay et al . 2005 ) .
but recent rxte observations by reig et al . ( 2009 ) did not find the cyclotron line at @xmath11 kev similar to the quiescent state reported in this paper .
then from the present data , cyclotron lines could be detected when the hard x - ray luminosity ( 20 100 kev ) is higher than @xmath68 erg s@xmath2 ( also see blay et al .
we suggested that in the active state ( like @xmath69 erg s@xmath2 ) , hard x - ray emissions will mainly come from the surface ( e.g. , polar cap region ) of the neutron star , while in the quiescent state , the x - ray emission would mainly come from the standing point region where the magnetic pressure is equal to the accretion pressure . among the wind - fed massive binary system , 4u 2206 + 54
is a unique one which was identified as a magnetic neutron star with a very low pulsation .
thus , detailed studies and long - term monitoring of the peculiar neutron star binary 4u 2206 + 54 in x - rays will help to understand accretion physics and evolution of wind - fed systems .
the author is grateful to the referee for the fruitful suggestions to improve the manuscript and also to jiang , y. y. for the help to prepare some figures .
this paper is based on observations of integral , an esa project with instrument and science data centre funded by esa member states ( principle investigator countries : demark , france , germany , italy , switzerland and spain ) , the czech republic and poland , and with participation of russia and us .
we used the archival data from asm light curve webpage developed by the asm team at the kavli institute for astrophysics and space research at the massachusetts institute of technology .
w. wang is supported by the national natural science foundation of china under grants 10803009 , 10833003 .
arnaud , k.a . , 1996 , in astronomical data analysis software and systems
v , asp conf .
ser . , 101 , 17 baykal , a. , stark , m.j . & swank , j.h . , 2000 , apj , 544 , l129 bird , a.j .
et al . , 2007 , apjs , 170 , 175 blay , p. et al . , 2005 ,
a&a , 438 , 963 blay , p. et al . , 2006 , a&a , 446 , 1095 corbet , r.h.d . , 1986 , mnras , 220 , 1047 corbet , r.h.d . , peele , a.g . & remillard , r. , 1997 , iau circ . , 6556 , 3 corbet , r.h.d .
& peele , a.g . , 2001 , apj , 562 , 936 corbet , r.h.d .
, markwardt , c.b . & tueller , j. , 2007 , apj , 655 , 458 davies , r.e . & pringle , j.e .
, 1981 , mnras , 196 , 209 goldwurm , a. et al . , 2003 ,
a&a , 411 , l223 kreykenbohm , i. et al . , 2004 ,
a&a , 427 , 975 lebrun , f. et al . , 2003 ,
a&a , 411 , l141 li , x.d . &
van den heuvel , e.p.j .
, 1999 , apj , 513 , l45 lund , n. et al . , 2003 ,
a&a , 411 , l231 marshall , n. & ricketts , m.j . , 1980 , mnras , 193 , 7p masetti , n. et al . , 2004 , a&a , 423 , 311 mas - hesse , j.m .
et al . , 2003 , a&a , 411 , l261 negueruela , i. & reig , p. , 2001
, a&a , 371 , 1056 reig , p. et al . , 2009 ,
a&a , 494 , 1073 rib , m. et al . , 2006 , a&a , 449 , 687 ruffert , m. , 1999 , a&a , 346 , 861 skinner , g.k .
et al . , 1982 ,
nature , 297 , 568 steiner , j.e .
et al . , 1984 ,
apj , 280 , 688 torrejn , j.m .
et al . , 2004 ,
a&a , 423 , 301 ubertini , p. et al . , 2003 , a&a , 411 , l131 vedrenne , g. et al . , 2003 ,
a&a , 411 , l63 winkler , c. et al . , 2003 , a&a , 411 , l1
|
the hard x - ray source 4u 2206 + 54 is a peculiar high mass x - ray binary with a main - sequence donor star .
recent x - ray observations suggested that the compact object in 4u 2206 + 54 may be a neutron star .
the x - ray emission comes from the accretion of stellar winds from the massive donor stars , and variability of luminosity may be due to the changes of its orbit phase . to further reveal the nature of compact object , we studied 4u 2206 + 54 with integral / ibis observations in two years , and found that in most time , 4u 2206 + 54 undergone a quiescent state and sometime an active state . in the quiescent state the spectrum can be fitted by a power - law model of @xmath0 with a hard x - ray luminosity of @xmath1 erg s@xmath2 ( 20 100 kev ) . while in the active state
, the 20 100 kev hard x - ray luminosity reaches @xmath3 erg s@xmath2 and the spectrum is fitted by a thermal bremmstrahlung model of @xmath4 kev plus two cyclotron absorption lines at @xmath5 30 and 60 kev .
then we derived a magnetic field of 3.3@xmath6 g for the compact object in 4u 2206 + 54 . during the active state
, we found a pulsation period of @xmath5 5400 s in the light curve of 4u 2206 + 54 .
so the compact object in 4u 2206 + 54 should be a magnetic neutron star with a slow pulsation .
cyclotron absorption lines detected in the active state and non - detection in the quiescent state suggested that two different accretion states have possible different hard x - ray emission regions : surface of neutron star in the active state ; the magnetic - accretion pressure equivalent point in the quiescent state .
the re - analysis of the rxte / asm light curve found the modulation periods at @xmath7 days and 19.11 days , and the orbit period of 4u 2206 + 54 should be 19.11 days .
stars : individual ( 4u 2206 + 54 ) stars : neutron magnetic fields stars : binaries : close x - rays : binaries .
|
for integers @xmath5 fixed , we are considering the `` non - standard '' non - self - adjoint eigenvalue problems @xmath9u(z,\lambda)=\lambda u(z,\lambda),\quad\text{for some $ \lambda\in{\mathbb c}$},\ ] ] with the boundary condition that @xmath10 where @xmath8 is a polynomial of degree at most @xmath11 of the form @xmath12 we let @xmath13 be the coefficient vector of @xmath14 .
we are mainly interested in the case when @xmath8 is real , that is , when @xmath15 .
however , some interesting facts in this paper hold also for @xmath16 .
so except for theorem [ main_theorem ] below , we will use @xmath16 .
if a nonconstant function @xmath17 satisfies with some @xmath18 and the boundary condition , then we call @xmath19 an _ eigenvalue _ of @xmath20 and @xmath17 an _ eigenfunction of @xmath20 associated with the eigenvalue @xmath19_. also , the _ geometric multiplicity of an eigenvalue @xmath19 _ is the number of linearly independent eigenfunctions associated with the eigenvalue @xmath19 . the operator @xmath20 in with potential @xmath21 is called _
@xmath22-symmetric _ if @xmath23 , @xmath24 .
note that @xmath21 is a @xmath22-symmetric potential if and only if @xmath15 .
before we state our main theorems , we first introduce some known facts by sibuya @xcite about the eigenvalues @xmath19 of @xmath20 .
[ main2 ] the eigenvalues @xmath6 of @xmath20 have the following properties
the set of all eigenvalues is a discrete set in @xmath25 .
the geometric multiplicity of every eigenvalue is one .
infinitely many eigenvalues , accumulating at infinity , exist .
the eigenvalues have the following asymptotic expansion @xmath26\quad\text{as $ n$ tends to infinity},\quad n \in { \mathbb n},\ ] ] where the error term @xmath27 could be complex - valued .
this paper is organized as follows . in section [ prop_sect ]
, we will introduce work of hille @xcite and sibuya @xcite , regarding properties of solutions of .
we then improve on the asymptotics of a certain function in @xcite . in section [ entire_sect ]
, we introduce an entire function @xmath28 whose zeros are the eigenvalues of @xmath20 , due to sibuya @xcite . in section [ asymp_sect ] , we then provide asymptotics of @xmath28 as @xmath29 in the complex plane , improving the asymptotics of @xmath28 in @xcite . in section [ asymp_eigen ] , we will improve the asymptotic expansion of the eigenvalues . in particular , we will prove the following . throughout this paper
, we use that @xmath30 is the largest integer that is less than or equal to @xmath31 .
[ main_thm1 ] let @xmath16 be fixed .
then there exist @xmath32 such that the eigenvalues @xmath6 of @xmath20 have the asymptotic expansion @xmath33 where @xmath34 one can compute @xmath35 directly ( or see equation ( 2.22 ) in @xcite with the identity @xmath36 ) and obtains @xmath37 in the last section , we prove the following theorem , regarding monotonicity of @xmath38 . [ ineq_eq ] for each @xmath16 there exists @xmath39 such that @xmath40 if @xmath41 . this is a consequence of .
finally , when @xmath20 is @xmath22-symmetric ( i.e. , @xmath15 ) , @xmath42 is an eigenfunction associated with an eigenvalue @xmath19 if and only if @xmath43 is an eigenfunction associated with the eigenvalue @xmath44 .
thus , the eigenvalues either appear in complex conjugate pairs , or else are real .
so theorem [ ineq_eq ] implies the following .
[ main_theorem ] suppose that @xmath15 . then the eigenvalues @xmath19 of @xmath20 are all real and positive , with only finitely many exceptions .
for the rest of the introduction , we will mention a brief history of problem . in recent years
, these @xmath22-symmetric operators have gathered considerable attention , because ample numerical and asymptotic studies suggest that many of such operators have real eigenvalues only even though they are not self - adjoint . in particular , the differential operators @xmath20 with some polynomial potential @xmath45 and with the boundary condition have been considered by bessis and zinn - justin ( not in print ) , bender and boettcher @xcite and many other physicists @xcite . around 1992 bessis and zinn - justin ( not in print ) conjectured that when @xmath46 , @xmath47 , the eigenvalues are all real and positive , and in 1998 , bender and boettcher @xcite conjectured that when @xmath48 , @xmath47 , the eigenvalues are all real and positive .
many numerical , asymptotic and analytic studies support these conjectures ( see , e. g. , @xcite and references therein and below ) .
the first rigorous proof of reality and positivity of the eigenvalues of some non - self - adjoint @xmath20 in was given by dorey , dunning and tateo @xcite in 2001 .
they proved that the eigenvalues of @xmath20 with the potential @xmath49 , @xmath50 , are all real if @xmath51 and @xmath52 , and positive if @xmath51 and @xmath53 . then in 2002
the present author @xcite extended the polynomial potential results of dorey , dunning and tateo to more general polynomial cases , by adapting the method in @xcite .
namely , when @xmath21 , the eigenvalues are all real and positive , provided that for some @xmath54 the coefficients of the real polynomial @xmath8 satisfy @xmath55 for all @xmath56 . however , there are some @xmath22-symmetric polynomial potentials that produce non - real eigenvalues .
delabaere and pham @xcite , and delabaere and trinh @xcite studied the potential @xmath57 and showed that a pair of non - real eigenvalues develops for large negative @xmath58 . moreover , handy @xcite , and handy , khan , wang and tymczak @xcite showed that the same potential admits a pair of non - real eigenvalues for small negative values of @xmath59 .
also , bender , berry , meisinger , savage and simsek @xcite considered the problem with the potential @xmath60 , @xmath61 , under decaying boundary conditions at both ends of the real axis , and their numerical study showed that more and more non - real eigenvalues develop as @xmath62 .
so without any restrictions on the coefficients @xmath63 , theorem [ main_theorem ] is the most general result one can expect about reality of eigenvalues .
also , the method used to prove theorem [ main_theorem ] in this paper is new .
the method used in @xcite is useful in proving reality of _ all _ eigenvalues , but i think that some critical arguments in proving reality of eigenvalues in @xcite can not be applied to the cases when some non - real eigenvalues exist .
the asymptotic expansion itself is interesting , and also implies theorem [ ineq_eq ] .
note that is not enough to conclude theorem [ ineq_eq ] .
finally , theorem [ ineq_eq ] and @xmath22-symmetry of @xmath20 explained right before theorem [ main_theorem ] above imply the partial reality of the eigenvalues in theorem [ main_theorem ] .
in this section , we introduce work of hille @xcite and sibuya @xcite about properties of the solutions of .
first , we scale equation because many facts that we need later are stated for the scaled equation .
let @xmath17 be a solution of ( [ ptsym ] ) and let @xmath64 .
then @xmath65 solves @xmath66v(z,\lambda)=0,\ ] ] where @xmath5 and @xmath8 is a polynomial ( possibly , @xmath67 ) of the form .
since we scaled the argument of @xmath17 , we must rotate the boundary conditions .
we state them in a more general context by using the following definition . _
the stokes sectors _ @xmath68 of the equation ( [ rotated ] ) are @xmath69 see figure [ f : graph1 ] . .
the dashed rays represent @xmath70 it is known from hille @xcite that every nonconstant solution of ( [ rotated ] ) either decays to zero or blows up exponentially , in each stokes sector @xmath68 .
that is , one has the following result .
[ gen_pro ] @xmath71 * for each @xmath72 , every solution @xmath65 of ( [ rotated ] ) ( with no boundary conditions imposed ) is asymptotic to @xmath73^{\frac{1}{2}}\,d\xi\right]\ ] ] as @xmath74 in every closed subsector of @xmath68 . *
if a nonconstant solution @xmath65 of decays in @xmath68 , it must blow up in @xmath75 .
however , when @xmath65 blows up in @xmath68 , @xmath65 need not be decaying in @xmath76 or in @xmath77 .
lemma [ gen_pro ] ( i ) implies that if @xmath65 decays along one ray in @xmath68 , then it decays along all rays in @xmath68 .
also , if @xmath65 blows up along one ray in @xmath68 , then it blows up along all rays in @xmath68 .
thus , since the rotation @xmath78 maps the two rays in onto the center rays of @xmath79 and @xmath80 , @xmath81 next we will introduce sibuya s results , but first we define a sequence of complex numbers @xmath82 in terms of the @xmath63 and @xmath19 , as follows . for @xmath83
fixed , we expand @xmath84 note that @xmath85 do not depend on @xmath19 , so we write @xmath86 for @xmath87 .
so the above expansion without the @xmath88 term still gives @xmath82 for @xmath89 .
we further define @xmath90 if @xmath91 is odd , and @xmath92 if @xmath91 is even .
the following theorem is a special case of theorems 6.1 , 7.2 , 19.1 and 20.1 of sibuya @xcite that is the main ingredient of the proofs of the main results in this paper .
[ prop ] equation ( [ rotated ] ) , with @xmath93 , admits a solution @xmath94 with the following properties . 1 .
@xmath94 is an entire function of @xmath95 and @xmath19 .
2 . @xmath94 and @xmath96 admit the following asymptotic expansions .
let @xmath97 .
then @xmath98,\nonumber\\ f^{\prime}(z , a,\lambda)=&-z^{r_m+\frac{m}{2}}(1+o(z^{-1/2}))\exp\left[-f(z , a,\lambda ) \right],\nonumber\end{aligned}\ ] ] as @xmath2 tends to infinity in the sector @xmath99 , uniformly on each compact set of @xmath100-values . here
@xmath101 3 . properties and uniquely determine the solution @xmath94 of ( [ rotated ] ) .
4 . for each fixed
@xmath16 and @xmath102 , @xmath103 and @xmath104 also admit the asymptotic expansions , @xmath105\lambda^{-1/4}\exp\left[l(a,\lambda)\right],\label{eq1}\\ f^{\prime}(0,a,\lambda)=&-[1+o(1)]\lambda^{1/4}\exp\left[l(a,\lambda)\right],\label{eq2}\end{aligned}\ ] ] as @xmath29 in the sector @xmath106 , where @xmath107 5 .
the entire functions @xmath108 and @xmath109 have orders @xmath110 . in sibuya
s book @xcite , see theorem 6.1 for a proof of ( i ) and ( ii ) ; theorem 7.2 for a proof of ( iii ) ; and theorem 19.1 for a proof of ( iv ) .
moreover , ( v ) is a consequence of ( iv ) along with theorem 20.1 .
note that properties ( i ) , ( ii ) and ( iv ) are summarized on pages 112113 of sibuya @xcite . using this theorem , sibuya (
* theorem 19.1 ) also showed the following corollary that will be useful later on .
[ lemma_decay ] let @xmath16 be fixed .
then @xmath111 as @xmath19 tends to infinity in the sector @xmath112 , and hence @xmath113 as @xmath29 in the sector @xmath114 .
in particular , @xmath115 as @xmath29 in any closed subsector of the sector @xmath116 .
in addition , @xmath117 as @xmath29 in any closed subsector of the sectors @xmath118
. this asymptotic expansion will be clear from lemma [ asy_lemma ] below , or alternatively , see ( * ? ? ?
* theorem 19.1 ) for a proof . based on the above corollary , sibuya ( * ? ? ?
* theorem 29.1 ) also proved the following asymptotic expansion of the eigenvalues .
@xmath119,\quad\text{as}\quad n\rightarrow \infty,\ ] ] where @xmath120.\ ] ] notice that in this paper we consider the boundary conditions of the scaled equation ( [ rotated ] ) where @xmath65 decays in @xmath121 , while sibuya studies equation ( [ rotated ] ) with boundary conditions such that @xmath65 decays in @xmath122 .
the factor @xmath123 in our formula ( [ ak ] ) is due to this scaling of the problem . throughout this paper
, we will deal with numbers like @xmath124 for some @xmath125 , and @xmath126 . as usual
, we will use @xmath127\ ] ] and if @xmath128 is specified , then @xmath129=s\left[{{\rm re}\,}(\nu)\frac{2\pi}{m+2}+\arg(\lambda)\right],\quad s\in{\mathbb r}.\ ] ] if @xmath130 then the branch of @xmath131 is chosen to be the negative real axis .
next , we provide an improved asymptotic expansion of @xmath132 .
we will use this new asymptotic expansion of @xmath132 to improve the asymptotic expansion of the eigenvalues .
[ asy_lemma ] let @xmath5 and @xmath16 be fixed .
then there exist constants @xmath133 , @xmath134 , such that @xmath135 as @xmath29 in the sector @xmath136 .
the function @xmath137 is defined as an integral over @xmath138 in theorem [ prop ] .
we will rotate the contour of integration using cauchy s integral formula . in doing so
, we need to justify that the integrand in the definition of @xmath137 is analytic in some domain in the complex plane .
let @xmath139 be a fixed number .
suppose that @xmath140 .
then if @xmath141 , there exists @xmath142 such that @xmath143 provided that @xmath144 . since @xmath145 lies in a large disk centered at the origin for @xmath146 , we see that for all @xmath19 with @xmath147 large , we have that @xmath148 and @xmath149 for all @xmath150 in the sector @xmath141 , and hence @xmath151 is analytic in the sector @xmath141 if @xmath19 lies outside a large disk and in the sector @xmath140 .
let @xmath152 then , since @xmath153 as @xmath150 tends to infinity in the sector @xmath141 , we have by cauchy s integral formula , upon substituting @xmath154 for all @xmath19 with @xmath147 large enough , @xmath155 where @xmath156 similarly , holds for @xmath157 .
next , we examine the following square root in @xmath158 : @xmath159 where @xmath160 are functions such that @xmath160 are all integrable on @xmath161 $ ] for any @xmath162 . moreover , by the definition of @xmath82 in , we see that for @xmath89 , @xmath163 thus , @xmath164 so @xmath165 for all @xmath166 .
next , when @xmath91 is even and @xmath167 , we write @xmath168 where we take @xmath169 .
thus , we have that @xmath135 as @xmath29 in the sector @xmath106 , where @xmath170 this completes the proof .
in this section , we will prove that the eigenvalues are zeros of an entire function .
first , we let @xmath171 then recall that the function @xmath94 in theorem [ prop ] solves ( [ rotated ] ) and decays to zero exponentially as @xmath172 in @xmath173 , and blows up in @xmath121 .
next , one can check that the function @xmath174 which is obtained by scaling @xmath175 in the @xmath2-variable , also solves ( [ rotated ] ) .
it is clear that @xmath176 .
also , @xmath177 decays in @xmath68 and blows up in @xmath75 since @xmath175 decays in @xmath173 .
since no nonconstant solution decays in two consecutive stokes sectors ( see lemma [ gen_pro ] ( ii ) ) , @xmath178 and @xmath179 are linearly independent and hence any solution of ( [ rotated ] ) can be expressed as a linear combination of these two .
especially , there exist some coefficients @xmath28 and @xmath180 such that @xmath181 we then see that @xmath182 where @xmath183 is the wronskian of @xmath184 and @xmath185 . since both @xmath184 and @xmath185 are solutions of the same linear equation ( [ rotated ] ) , we know that the wronskians are constant functions of @xmath2 . also , since @xmath185 and @xmath179 are linearly independent , @xmath186 for all @xmath187
. moreover , we have the following lemma that is useful later on .
suppose @xmath188 .
then @xmath189 and @xmath190 , where @xmath191 moreover , @xmath192 where @xmath193 see sibuya @xcite for proof . here
, we mention that by , we have @xmath194 and hence @xmath195 .
now we can identify the eigenvalues of @xmath20 as the zeros of the entire function @xmath196 .
[ iden_thm ] for each fixed @xmath16 , the function @xmath196 is entire .
moreover , @xmath19 is an eigenvalue of @xmath20 if and only if @xmath197 .
since @xmath198 and since @xmath199 is a wronskian of two entire functions , it is clear from that @xmath28 is an entire function of @xmath19 for each fixed @xmath16 . next , suppose that @xmath19 is an eigenvalue of @xmath20 with a corresponding eigenfunction @xmath17 , then the scaled eigenfunction @xmath64 solves and decays in @xmath121 . hence , @xmath65 is a ( nonzero ) constant multiple of @xmath200 since both decays in @xmath80 .
similarly , @xmath65 is also a constant multiple of @xmath201 .
thus , @xmath201 is a constant multiple of @xmath200 , implying @xmath197 .
conversely , if @xmath197 , then @xmath201 is a constant multiple of @xmath200 , and hence @xmath200 also decays in @xmath79 .
thus , @xmath200 decays in @xmath202 and is a scaled eigenfunction with the eigenvalue @xmath19 .
moreover , the following is an easy consequence of : for each @xmath72 we have @xmath203 where we use @xmath204 for @xmath180 since it is independent of @xmath19 .
in this section , we provide asymptotic expansions of the entire function @xmath28 as @xmath29 along all possible rays to infinity in the complex plane .
first , we provide an asymptotic expansion of the wronskian of @xmath205 and @xmath184 in preparation for providing an asymptotic expansion of @xmath28 .
[ lemma7 ] suppose that @xmath206 .
then for each @xmath16 , @xmath207\exp\left[l(g^{j}(a),\omega^{2j - m-2}\lambda)+l(a,\lambda)\right],\ ] ] as @xmath29 in the sector @xmath208 we fix @xmath206 .
then , @xmath209\exp\left[l(g^{j}(a),\omega^{2j - m-2}\lambda)+l(a,\lambda)\right]\nonumber\\ & \underset{\lambda\to\infty}{=}[2i\omega^{-\frac{j}{2}}+o(1)]\exp\left[l(g^{j}(a),\omega^{2j - m-2}\lambda)+l(a,\lambda)\right],\nonumber\end{aligned}\ ] ] where we used and with @xmath210 .
next , we provide an asymptotic expansion of @xmath211 as @xmath29 along the rays near the negative real axis .
notice from that @xmath212 . also , @xmath213 is a nonzero constant function of @xmath19 .
so from these one gets an asymptotic expansion of @xmath28 . for each fixed @xmath16 and @xmath139 , @xmath214\exp\left[l(g^{-1}(a),\omega^{-2}\lambda)+l(g(a),\omega^{-m}\lambda)\right],\ ] ] as @xmath215 along the rays in the sector
@xmath216 moreover , there exists a constant @xmath217 such that @xmath218 for all @xmath19 in the sector if @xmath219 .
this is an easy consequence of lemma [ lemma7 ] and equation .
the last assertion of the theorem is a consequence of the asymptotic expansion .
the asymptotic expansion of @xmath28 in a sector near the positive real axis is obtained in the following theorem .
[ zero_thm ] suppose that @xmath220 .
then for each fixed @xmath16 and @xmath221 @xmath222\exp\left[l(g^{-1}(a),\omega^{-2}\lambda)-l(a,\lambda)\right]\nonumber\\ & + [ \omega^{\frac{1}{2}+2\nu(a)}+o(1)]\exp\left[l(g(a),\omega^{2}\lambda)-l(a,\lambda)\right],\nonumber\end{aligned}\ ] ] as @xmath29 in the sector @xmath223 suppose @xmath224 .
then from , and lemma [ lemma7 ] , @xmath225\frac{\exp\left[l(g^{k}(a),\omega^{2k - m-2}\lambda)+l(g^{-1}(a),\omega^{-2}\lambda)\right]}{\exp\left[l(g^{k}(a),\omega^{2k - m-2}\lambda)+l(a,\lambda)\right]}\nonumber\\ & -[\omega^{-\frac{1}{2}}+o(1)]\widetilde{c}(a)\frac{\exp\left[l(g^{k}(a),\omega^{2k - m-2}\lambda)+l(g(a),\omega^{2}\lambda)\right]}{\exp\left[l(g^{k}(a),\omega^{2k - m-2}\lambda)+l(a,\lambda)\right]}\nonumber\\ = & [ \omega^{\frac{1}{2}}+o(1)]\exp\left[l(g^{-1}(a),\omega^{-2}\lambda)-l(a,\lambda)\right]\nonumber\\ & + [ \omega^{-\frac{1}{2}}+o(1)]\omega^{1 + 2\nu(a)}\exp\left[l(g(a),\omega^{2}\lambda)-l(a,\lambda)\right],\nonumber\end{aligned}\ ] ] as @xmath29 such that @xmath226 that is , @xmath227 provided that @xmath228 so in order to complete the proof , we choose @xmath229 the sectors and do not cover the entire complex plane near infinity .
the next theorem covers a sector in the upper half plane , connecting the sectors and in the upper half plane .
[ thm_sector2 ] suppose that @xmath16 and @xmath139 . if @xmath220 then @xmath230\exp\left[l(g^{-1}(a),\omega^{-2}\lambda)-l(a,\lambda)\right]\nonumber\\ & -[i\omega^{1+\mu(a)+4\nu(a)}+o(1)]\exp\left[-l(g^2(a),\omega^{2-m}\lambda)-l(a,\lambda)\right],\label{asy_asy}\end{aligned}\ ] ] as @xmath29 in the sector @xmath231 if @xmath232 then @xmath233\exp\left[l(g^{4}(a),\omega^{-2}\lambda)-l(a,\lambda)\right]\nonumber\\ & -[i\omega^{\frac{7}{4}}+o(1)]\exp\left[-l(g^2(a),\omega^{-1}\lambda)-l(a,\lambda)\right],\nonumber\end{aligned}\ ] ] as @xmath29 in the sector @xmath234 moreover , if @xmath235 then there exists a constant @xmath236 such that @xmath237 for all @xmath19 in the sector if @xmath238 .
suppose that @xmath220 .
then from , and lemma [ lemma7 ] , @xmath239\exp\left[l(g^{2}(a),\omega^{2-m}\lambda)+l(g^{-1}(a),\omega^{-2}\lambda)\right]}{[2i\omega^{-\frac{2}{2}}+o(1)]\exp\left[l(g^{2}(a),\omega^{2-m}\lambda)+l(a,\lambda)\right]}\nonumber\\ & -\widetilde{c}(a)\frac{\omega^{-1}w_{0,1}(g(a),\omega^2\lambda)}{[2i\omega^{-1}+o(1)]\exp\left[l(g^{2}(a),\omega^{2-m}\lambda)+l(a,\lambda)\right]}\nonumber\\ = & [ \omega^{\frac{1}{2}}+o(1)]\exp\left[l(g^{-1}(a),\omega^{-2}\lambda)-l(a,\lambda)\right]\nonumber\\ & -\frac{-\omega^{1 + 2\nu(a)}2\omega^{\mu(g(a))}}{[2i+o(1)]\exp\left[l(g^{2}(a),\omega^{2-m}\lambda)+l(a,\lambda)\right]}\nonumber\end{aligned}\ ] ] as @xmath29 such that @xmath240 that is , @xmath241 next , we use @xmath242 to get .
suppose @xmath232 .
then @xmath243 .
also , @xmath244 since @xmath245 .
thus , we have that @xmath246\exp\left[l(g^{4}(a),\omega^{-2}\lambda)+l(g^{2}(a),\omega^{-1}\lambda)\right]}{[2i\omega^{-\frac{2}{2}}+o(1)]\exp\left[l(g^{2}(a),\omega^{-1}\lambda)+l(a,\lambda)\right]}\nonumber\\ & -\widetilde{c}(a)\frac{\omega^{-1}w_{0,1}(g(a),\omega^2\lambda)}{[2i\omega^{-1}+o(1)]\exp\left[l(g^{2}(a),\omega^{-1}\lambda)+l(a,\lambda)\right]}\nonumber\\ = & [ -\omega^{-2}+o(1)]\exp\left[l(g^{4}(a),\omega^{-2}\lambda)-l(a,\lambda)\right]\nonumber\\ & -\frac{-\omega^{1 + 2\nu(a)}2\omega^{\mu(g(a))}}{[2i+o(1)]\exp\left[l(g^{2}(a),\omega^{2-m}\lambda)+l(a,\lambda)\right]},\nonumber\end{aligned}\ ] ] as @xmath29 such that @xmath247 that is , @xmath248 in order to show the last assertion , we suppose that @xmath197 for some @xmath19 in with large @xmath147 . then from the asymptotic expansion
, we have @xmath249=[i\omega^{-\frac{1}{2}+\mu(a)+4\nu(a)}+o(1)].\ ] ] by corollary [ lemma_decay ] , @xmath250 note that if @xmath235 , then @xmath251 in .
since @xmath252 we see that @xmath253 as @xmath29 in , and hence the left hand side of blows up .
thus , @xmath28 can not have infinitely many zeros in .
this completes the proof .
the next theorem covers a sector in the lower half plane , connecting sectors and .
suppose that @xmath16 and @xmath139 . if @xmath220 then @xmath254\exp\left[-l(a,\omega^{-m-2}\lambda)-l(g^{-2}(a),\omega^{-4}\lambda)\right]\nonumber\\ & + [ \omega^{\frac{1}{2}+2\nu(a)}+o(1)]\exp\left[l(g(a),\omega^{-m}\lambda)-l(a,\omega^{-m-2}\lambda)\right],\nonumber\end{aligned}\ ] ] as @xmath29 in the sector @xmath255 if @xmath232 then @xmath256\exp\left[-l(a,\omega^{-5}\lambda)-l(g^{-2}(a),\omega^{-4}\lambda)\right]\nonumber\\ & + [ \omega^3+o(1)]\exp\left[l(g(a),\omega^{-3}\lambda)-l(a,\omega^{-5}\lambda)\right],\nonumber\end{aligned}\ ] ] as @xmath29 in the sector @xmath257 moreover , if @xmath235 then there exists a constant @xmath258 such that @xmath237 for all @xmath19 in the sector if @xmath259 .
suppose that @xmath220 . then from , and lemma [ lemma7 ] , @xmath260\exp\left[l(a,\omega^{-m-2}\lambda)+l(g^{-2}(a),\omega^{-4}\lambda)\right]}\nonumber\\ & -\widetilde{c}(a)\frac{[2i\omega^{-\frac{3}{2}}+o(1)]\exp\left[l(g(a),\omega^{-m}\lambda)+l(g^{-2}(a),\omega^{-4}\lambda)\right]}{[2i\omega^{-\frac{2}{2}}+o(1)]\exp\left[l(a,\omega^{-m-2}\lambda)+l(g^{-2}(a),\omega^{-4}\lambda)\right]}\nonumber\\ = & \frac{2\omega^{\mu(g^{-2}(a))}}{[2i\omega^{-1}+o(1)]\exp\left[l(a,\omega^{-m-2}\lambda)+l(g^{-2}(a),\omega^{-4}\lambda)\right]}\nonumber\\ & + [ \omega^{-\frac{1}{2}}+o(1)]\omega^{1 + 2\nu(a)}\frac{\exp\left[l(g(a),\omega^{-m}\lambda)+l(g^{-2}(a),\omega^{-4}\lambda)\right]}{\exp\left[l(a,\omega^{-m-2}\lambda)+l(g^{-2}(a),\omega^{-4}\lambda)\right]},\nonumber\end{aligned}\ ] ] as @xmath215 such that @xmath261 that is , @xmath262 which is .
suppose that @xmath232 .
then , @xmath263\exp\left[l(a,\omega^{-5}\lambda)+l(g^{-2}(a),\omega^{-4}\lambda)\right]}\nonumber\\ & -\omega^2\widetilde{c}(a)\frac{[2i\omega^{-\frac{2}{2}}+o(1)]\exp\left[l(g^3(a),\omega^{-4}\lambda)+l(g(a),\omega^{-3}\lambda)\right]}{[2i\omega^{-\frac{2}{2}}+o(1)]\exp\left[l(a,\omega^{-5}\lambda)+l(g^{-2}(a),\omega^{-4}\lambda)\right]}\nonumber\\ = & \frac{2\omega^{\mu(g^{-2}(a))}}{[2i\omega^{-1}+o(1)]\exp\left[l(a,\omega^{-5}\lambda)+l(g^{-2}(a),\omega^{-4}\lambda)\right]}\nonumber\\ & + [ \omega^{2}+o(1)]\omega^{1 + 2\nu(a)}\frac{\exp\left[l(g^{-2}(a),\omega^{-4}\lambda)+l(g(a),\omega^{-3}\lambda)\right]}{\exp\left[l(a,\omega^{-5}\lambda)+l(g^{-2}(a),\omega^{-4}\lambda)\right]},\nonumber\end{aligned}\ ] ] as @xmath29 such that @xmath264 that is , @xmath265 finally , the proof of the last assertion of this theorem follows as in the proof of theorem [ thm_sector2 ] . from the asymptotic expansions in the previous four theorems , one obtains the order of the entire function @xmath196 .
the order of an entire function @xmath266 is defined by @xmath267 where @xmath268 for @xmath269 . if for some positive real numbers @xmath270 , we have @xmath271\leq m(r ,
g)\leq \exp[c_2 r^{\sigma}]$ ] for all large @xmath272 , then the order of @xmath266 is @xmath273 .
the entire function @xmath196 is of order @xmath110 .
the sectors in , , , , cover a neighborhood of infinity in the complex plane .
so the nonconstant entire function @xmath274 is bounded above by @xmath275 $ ] for some constant @xmath276 .
also , along the ray @xmath277 , one can see from and that @xmath274 is bounded below by @xmath278 $ ] for some constant @xmath279 .
hence , the order of @xmath280 is @xmath110
. since the eigenvalues are the zeros of the entire function @xmath196 of order @xmath281 , there are infinitely many discrete eigenvalues as was already mentioned in theorem [ main2 ] .
in this section , we prove theorem [ main_thm1 ] by using the asymptotic expansions of @xmath28 and @xmath137 .
recall that by theorem [ iden_thm ] , @xmath19 is an eigenvalue of @xmath20 if and only if @xmath197 . for @xmath220 and @xmath16 fixed ,
suppose that @xmath197 for some @xmath19 with @xmath147 large .
then from the asymptotic expansion of @xmath28 in theorem [ zero_thm ] we have @xmath282\exp\left[l(g(a),\omega^2\lambda)-l(g^{-1}(a),\omega^{-2}\lambda)\right ] = -\omega^{-2\nu(a)},\ ] ] and absorbing @xmath283 $ ] into the exponential function then yields @xmath284 = -\omega^{-2\nu(a)}.\ ] ] thus , from lemma [ asy_lemma ] if @xmath91 is odd , we infer @xmath285+o(1)\label{region_def}\\ & = 2ik_{m,0}\sin\left(\frac{2\pi}{m}\right)\lambda^{\frac{1}{2}+\frac{1}{m}}+\sum_{j=1}^{\lfloor\frac{m}{2}+1\rfloor}c_{m , j}(a)\lambda^{\frac{1}{2}+\frac{1-j}{m}}+o(1),\nonumber\end{aligned}\ ] ] where @xmath286 similarly , if @xmath91 is even , then from lemma [ asy_lemma ] we have with @xmath287 in for @xmath54 , and @xmath288 where we used @xmath289 note that there exist constants @xmath39 and @xmath97 such that the function @xmath290 is continuous in the region @xmath291 and @xmath292 . from
we then see that the function maps the region @xmath291 and @xmath292 onto a region that contains the entire positive imaginary axis near infinity .
thus , from we get that for every sufficiently large @xmath293 there exists @xmath6 such that @xmath294 thus , @xmath295 let @xmath296 then @xmath297 introduce the decomposition @xmath298 where @xmath299 then from we have @xmath300 thus , @xmath301 and hence @xmath302 thus , one concludes @xmath303 where @xmath304 next , from along with we have @xmath305 thus , @xmath306 and hence @xmath307 so we choose @xmath308 where @xmath309 next , we replace @xmath310 in by . upon iterating this process
we get @xmath311 as @xmath7 , that is , .
suppose that @xmath232 .
for this case we will use the asymptotic expansion in theorem [ thm_sector2 ] that is valid in . similarly to what we did for the case @xmath220 , if @xmath197 then from the asymptotic expansion in theorem [ thm_sector2 ] we have @xmath282\exp\left[l(g^{4}(a),\omega^{-2}\lambda)+l(g^{2}(a),\omega^{-1}\lambda)\right]=-i\omega^{\frac{15}{4}}.\ ] ] thus , since @xmath312 we have @xmath313 so the continuous function @xmath314 maps a neighborhood of the positive real axis near infinity onto a neighborhood of the negative imaginary axis near infinity .
hence , there exist a sequence of @xmath6 near the positive real axis such that for all large enough positive integers @xmath315 , @xmath316 from this result one concludes that the asymptotic expansion holds for @xmath232 as well similarly to the proof for the case @xmath220 .
first , note from that @xmath317 as @xmath7 .
|
we study the eigenvalue problem @xmath0u(z)=\lambda u(z)$ ] with the boundary conditions that @xmath1 decays to zero as @xmath2 tends to infinity along the rays @xmath3 , where @xmath4 is a polynomial and integers @xmath5 .
we provide an asymptotic expansion of the eigenvalues @xmath6 as @xmath7 , and prove that for each _ real _ polynomial @xmath8 , the eigenvalues are all real and positive , with only finitely many exceptions .
_ 2000 _ mathematics subject classification _
: 34l40 , 34l20
_ = 18pt
|
flows of gas play a crucial role in galaxy formation both the infalling primordial gas from which galaxies ultimately form , and outflows which enrich the intergalactic medium and regulate galaxy formation .
much attention has recently been placed on one possible way of directly observing this gas : ly@xmath0 fluorescence .
the idea , originally proposed by @xcite is that the plentiful ultraviolet ( uv ) radiation in the high redshift universe will be absorbed by any neutral gas , and some fraction of the incident ionising photons will be re - radiated as ly@xmath0 photons @xcite
. this fluorescent ly@xmath0 emission will be brightest where the uv radiation is strongest : close to luminous qsos . @xcite and @xcite predicted that qsos with sufficiently high redshifts should be surrounded by halos of ly@xmath0 emission , as the uv flux from the qso causes the still infalling primordial gas to fluoresce .
the predicted intensity of this flux is , however , very dependent upon assumptions on how the gas is clumped
. the observational picture is currently confusing .
it has long been established that many high redshift radio sources , both qsos and radio galaxies , have bright ly@xmath0 emission coming from around them ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , but this may be caused by the radio jets , winds or cooling flows , rather than fluorescence .
luminous extended ly@xmath0 nebulae ( `` blobs '' ) are fairly common at high redshifts , but not obviously associated with qsos : they may be caused by cooling flows , superwinds or photoionisation by concealed agns ( e.g .. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and refs therein ) .
there are less data available for radio - quiet qsos .
damped ly@xmath0 absorption - line systems that lie close to the qso redshift have been detected in possible fluorescent emission in a few cases @xcite .
more recently , probable fluorescent emission was seen from another absorption system , but this time the absorber was 380 kpc away , in front of a different qso @xcite .
@xcite failed to detect any fluorescent emission around a qso , and indeed claimed that the presence of the qso was suppressing even the normal background population of ly@xmath0 emitting galaxies . on the other hand , @xcite detected ly@xmath0 fuzz close to a @xmath1 qso which they modeled as a spherical infalling halo of primordial gas .
@xcite found spectacularly bright ly@xmath0 fuzz , with a high velocity dispersion , around a qso at redshift 4.46 .
@xcite suggest that evidence for infalling primordial gas around high redshift qsos can be seen in the detailed profiles of the qso emission line .
it is thus becoming clear that fluorescent ly@xmath0 can be detected with current telescope facilities , but on the basis of the extremely small number statistics available to date , its properties appear to be heterogeneous
. a larger sample size would clearly be useful . in this paper
we contribute to this effort , searching for fluorescent emission near the brightest accessible qso with a redshift above four : pss 2155 + 1358 , at redshift 4.28 .
previous detections of fluorescent ly@xmath0 have used either long - slit spectroscopy ( e.g. * ? ? ? * ) or narrow - band imaging ( eg . * ? ? ?
long - slit spectroscopy has the disadvantage that the slit may not be at the right orientation to catch any emission .
narrow - band imaging is only possible if the qso redshift is known with high precision , which is not usually the case . in this paper
we try a different approach : integral field unit ( ifu ) spectroscopy .
we assume a cosmology with @xmath2 , @xmath3 and @xmath4 throughout this paper .
the qso pss 2155 + 1358 was observed with the integral field unit of the gemini multi - object spectrograph ( gmos ) on the gemini north telescope .
the b600 grating was used in two - slit mode , giving a spectral resolution of @xmath5 , over the 5@xmath6 ifu field of view .
the @xmath7 filter was used to limit the wavelength range to 562 698 nm and hence prevent spectra from the two slits from overlapping . a central wavelength of 625 nm was used , with some dithering in the spectral direction between frames to eliminate chip gaps .
observations were obtained in queue mode .
observations were taken on 2004 august 14 , september 11 and september 14 , in clear weather conditions and with seeing varying between 0.7 and 0.9 arcsec .
a total exposure time of 6 hours was obtained .
the data were reduced using the gemini iraf ( image reduction and analysis ) package .
standard settings were used to flat field and bias subtract the frames , extract the spectra , wavelength calibrate and sky subtract the spectra , and assemble them into a data cube .
there was a problem with the wavelength calibration , due to a ( now rectified ) issue with the grating mounts .
as a result , the absolute wavelength calibration of each frame , based on arc lines , was out by a small amount : different frames had to be aligned using absorption - lines in the spectra before co - addition .
we used sky lines to make sure that the wavelength calibration was at least self - consistent between all the fibres in a give slit .
the final co - added data cube was put on a correct wavelength scale by comparing the absorption line centroids with those measured in an independent higher resolution spectrum ( [ absorb ] ) . the product was a data cube with 0.2 arcsec spatial pixels and 0.092 nm spectral pixels .
sky subtraction used a separate 5@xmath8 sky region , one arcmin from the qso .
this subtraction ( as performed by the gemini iraf package ) left a flat residual , probably due to scattered light : this residual was weakly wavelength dependent but differed considerably between the two slits .
this led to a wavelength dependent step in the background level between the left and right hand sides of any image slice through the data cube .
as the qso sat exactly on the boundary between the two halves , this would have been disastrous for point spread function subtraction if it had not been corrected .
an iraf script was written to perform this correction .
for each spatial slice through the data cube , the top and bottom three pixels in each column were medianed , and this median subtracted from the whole column .
we first searched for diffuse line emission which might fill the whole ifu field of view , as would be seen if the qso was embedded , for example , in a giant ly@xmath0 blob . for this measurement ,
the scattered light correction mentioned in the previous section was not used . in each @xmath9 spatial slice
, we averaged the two ends : the first and last @xmath10 regions , which were free of qso light .
the average of these two regions was plotted , looking for narrow spikes .
the largest spikes had an amplitude ( average , over both @xmath10 regions ) of @xmath11 over a three spectral pixel ( 1300 @xmath12 ) velocity range .
this is a conservative upper limit : over the majority of the spectrum ( away from the rare strong sky lines ) the limit is a factor of three better .
we next searched the data cube for any compact emission - line sources .
this was done both with and without subtraction of the qso point spread function .
point spread function ( psf ) subtraction was done using an iraf script as follows . for each spectral slice through the data cube , a separate point spread function was calculated .
it was formed from the median of all the spectra slice images in the velocity ranges @xmath13:@xmath14 and @xmath15:@xmath16 .
images in which the qso flux was very low due to an absorption line were excluded from this median .
the psf image thus constructed was scaled to have the correct flux in the central @xmath17 region of the qso , and subtracted from the spectral slice image .
this procedure worked very well in general , giving psf subtraction errors typically smaller than the background noise .
several other psf subtraction techniques were used , but gave poorer results .
these other techniques were , however , used as a check on any objects discovered .
unfortunately , a sub - set of image slices had errors considerably above the background noise , for a variety of reasons : * wavelength shifts between the two halves of the image .
it turned out that the wavelength solution for the two halves of the image ( deriving from the two gmos slits ) was different , albeit only at the sub - pixel level .
this introduced artifacts in the psf subtracted image where the gradient of the qso spectrum was steep , typically at the edges of narrow absorption lines .
unfortunately our spectral sampling was not sufficient to fully remove it .
it produced , however , an easily recognised characteristic symmetrical left - right pattern , and only occurred at the edge of strong narrow absorption lines .
* chip gaps .
the gmos camera uses a three - ccd mosaic , which produced gaps in the wavelength coverage of any individual spectrum .
we observed at four different central wavelengths , ensuring that we had at least three quarters of our data at any given wavelength .
as the seeing differed between the different nights data , however , the seeing width in the co - added data - cube varied depending which nights data were used at a given wavelength .
this led to strong residuals in the psf subtraction near the edges of any of the regions affected by a chip gap . once again
, this produced a very characteristic pattern ( a positive or negative ring in the subtracted image ) and it was easy to check that such rings were indeed produced by a chip gap .
* other glitches .
there were occasional glitches away from the qso position , due to cosmic ray hits , defects in the ccds and problems in the spectrum extraction software .
these were mercifully rare , and typically easy to recognise because they affected either a region smaller than the psf , or a whole row or column in the data .
a disadvantage of fibre ifu designs such as gmos is that the mapping between pixels in the raw data and those in the final data cube is very complex and not one - to - one , making it extremely time - consuming checking the origin of glitches such as these .
for all these reasons , it did not prove possible to do an automated search for line emission .
instead , the whole cube was eyeballed independently by the two authors , and lists of candidate emission - line objects assembled .
these candidates were then checked in detail to eliminate the various glitches .
errors were estimated from the size of the largest candidate objects for which it was not possible to confidently decide whether they were glitches or not .
this error limit typically a factor of @xmath18 greater than the ( @xmath19 ) sky noise limit .
our detection limit was estimated by inserting artificial point - sources into various image slices , carrying out the point spread function subtraction and seeing if they were obviously seen , and easily distinguished from the various artifacts . as shown in fig
[ limits ] , our sensitivity to point sources is a strong function of how far they are in projection from the qso .
the qso spectrum is shown in fig [ qsospec ] , and shows a strong ly@xmath0 line peaking at around 642 nm , and many absorption lines .
the observed qso flux is 181 @xmath20jy at 660 nm .
only one significant ly@xmath0 emitter was discovered , at 641.97 nm ( we defer discussion of a second , marginally significant emitter to [ marginal ] ) .
this emission - feature can be clearly seen in each night s data independently , so we regard the detection as secure .
the 641.97 nm line emission comes from @xmath21 east of the qso .
it is spectrally unresolved ( fig [ profile ] ) with a line width ( full width at half maximum height , fwhm ) of @xmath22 nm ( @xmath23 ) .
a flux of @xmath24 was measured by point - spread function fitting of the qso psf subtracted image .
note that this is a lower limit on the true flux , as any flux overlapping the centre of the qso image will have been removed by the qso psf subtraction .
no continuum emission is detected at this location : this allows us to place a lower limit of @xmath25 nm ( observed frame ) on the equivalent width of the line .
the line lies at the peak of the broad qso ly@xmath0 emission - line , at almost the same wavelength as a strong narrow ly@xmath0 absorption - line system ( fig [ qsospec_close ] ) .
the emission peak is offset by @xmath26 ( @xmath27 uncertainty ) to the blue of the absorption - line centroid .
the emitter is not an artefact of the wavelength calibration problem at the edge of strong absorption - lines , noted in
[ compact ] : it lies on the flat part of the qso spectrum near the base of the absorber , where this artefact is not important .
what are the properties of this absorption - line , which lies at the same redshift as the ly@xmath0 emitter ?
celine proux kindly made available to us a spectrum of qso pss 2155 + 1358 obtained with the ultra - violet echelle spectrograph ( uves ) on the very large telescope ( vlt ) .
this spectrum is described in @xcite .
this spectrum was used to investigate the properties of the absorption - line at the same wavelength as the z=4.2793 emitter .
the line is clearly seen in multiple lyman - series lines , allowing an accurate neutral hydrogen column density of @xmath28 to be measured . the line width parameter @xmath29 .
the line is also detected in n v and c iv , where it breaks up into at least three sub - components each with a velocity width of no greater than 10@xmath12 ( fig [ plotc4 ] ) , spread over a 40@xmath12 velocity range .
we fit the absorption with a three - sub - component model : component @xmath30 is at a vacuum heliocentric redshift of z=4.2806 , component @xmath31 at z=4.2809 and component @xmath32 at z=4.2812 .
column densities were measured interactively using the xvoigt program @xcite the c iv column densities are 6.3 , 9.1 and @xmath33 for components @xmath30 , @xmath31 and @xmath32 respectively . for n
v , component @xmath30 has a column density @xmath34 , while the combined column density of components @xmath31 and @xmath32 is @xmath35 .
upper limits on the column density of each individual component in other transitions are given in table [ ulims ] .
.upper limits on the absorption column density of each absorbing component.[ulims ] [ cols="<,^",options="header " , ]
is this emission line due to ly@xmath0 emission at redshift 4.2793 , or could it be emission from a foreground galaxy , in a different line ?
the detected line is too narrow to be [ o ii ] ( 372.7 nm ) , as we would have easily resolved the doublet .
the main possibilities are thus h@xmath36 and the [ o iii ] doublet .
if it were any of these lines , we might have expected to see one of the others .
the expected wavelengths for all the various possibilities were checked and nothing was seen .
we can put a conservative lower limit on the equivalent width of the line of 10 nm .
this is quite normal for ly@xmath0 emitting galaxies , but would be extremely high for a foreground galaxy ( eg . * ? ? ?
we therefore conclude that the line is most likely ly@xmath0 .
further evidence for this hypothesis comes from the probability argument in [ redshift ] . to work out whether the ly@xmath0 emitter we saw is associated with the qso
, we need to know the qso redshift . @xcite
estimated the qso redshift by fitting gaussians to the peak of the ly@xmath0 , si iv / o iv ( 140 nm ) and c iv ( 154.8 nm ) lines . their data consisted of a spectrum with a relatively low resolution and signal - to - noise ratio .
the results were discrepant : z=4.285 from ly@xmath0 , 4.269 from si iv / o iv and a much lower value of 4.216 from c iv .
the discrepancy was unsurprising , as all three lines were strongly affected by absorption - line systems .
we used the uves spectrum to better determine the redshift .
its higher resolution and signal - to - noise ratio allowed us to clearly see the continuum between the absorption lines . from this spectrum , we see that the ly@xmath0 emission line really does have quite a sharp peak , and that the downturn below 639 nm is real and not due to absorption . using this peak wavelength ,
we get a redshift estimate of @xmath37 . measuring a redshift from the metal lines proved to be harder , as both have extremely broad , flat profiles , without a well defined peak .
our best estimate is @xmath38 for si iv / o iv , and @xmath39 for c iv .
the results are thus still discrepant , albeit all overlapping within their respective error bars .
@xcite showed that civ is systematically blue - shifted in radio - quiet qsos , and that this blueshift is strongest when the line is broadest and has a flat top .
the narrow core of ly@xmath0 , if present , seems to be a more reliable redshift indicator , provided you have enough spectral resolution to measure the true continuum shape through the ly@xmath0 forest absorption .
another clue to the redshift comes from a series of associated c iv absorption lines , covering a range of redshift , with the highest at z=4.2793 .
unless this gas is infalling rapidly into the qso , this places a lower limit on the true redshift .
taken together , we thus prefer the higher redshift measured in ly@xmath0 : @xmath40 . in this section
we argue that the odds of detecting a ly@xmath0 emitter by chance close in both redshift and projected distance to the qso are small .
we therefore conclude that the emitter we saw is indeed physically close to the qso .
@xcite measure the surface density of ly@xmath0 emitters at z=4.5 down to a flux limit close to ours .
using their surface density , we would only have expected to find @xmath41 emitters in our whole data cube .
the odds of seeing one by chance within one projected arcsec of the qso is @xmath42% . and the odds of finding one that is also within the redshift range @xmath37 is @xmath43% .
we can also do the probability calculation internally to our data , by noting that no ly@xmath0 emitters were found in the outer regions of our data cube ( 20 square arcsec , @xmath44 ) while one was seen within one arcsec of the qso in the redshift range @xmath37 .
the ratio of volumes is thus 128:1 .
it is thus quite unlikely that we would have found our one source so close to the qso by chance .
we therefore conclude that the probability of finding a ly@xmath0 emitter is enhanced close to the qso , and that the qso and emitter really lie close to each other .
the wavelength of our ly@xmath0 emitter lies within @xmath45 ( @xmath46 nm ) of the wavelength of an associated absorption - line system in the qso spectrum .
is this coincidence , or could the two be physically connected ?
how likely is such a coincidence ? consider a random ly@xmath0 emitter , which lay at a redshift within our uncertainty on the redshift of the qso ( @xmath37 ) , and which we would thus consider to be associated with the qso .
there are two strong absorption - lines within this redshift interval ( fig [ qsospec_close ] ) .
the odds of a random emitter lying within @xmath46 nm of one of these absorbers is 8% .
this assumes no wavelength dependence in our sensitivity limit . in practice , we are more sensitive where the qso spectrum is weaker , but our emitter is strong enough to have been detected regardless of the qso flux at that wavelength .
this number is small , but not small enough to rule out the possibility that the redshift match between absorber and emitter is coincidence .
we conclude that while it is quite likely that the absorber and emitter are connected , this has not been proved .
we have concluded that the ly@xmath0 emitter at @xmath47 lies close to the qso , and may well be connected with the gas producing the absorption line @xmath45 further to the red .
what can we learn about the physical properties of this absorber ? for our adopted cosmology ,
the ly@xmath0 emitting cloud is spectrally unresolved ( fwhm @xmath48 ) , spatially either marginally resolved or unresolved ( size @xmath49kpc ) , and has a luminosity of @xmath50 : more if we ve over - done the psf subtraction .
it lies at a projected distance of 5kpc from the qso let us first assume that the cloud is @xmath51 kpc in size , and optically thick at the lyman - limit .
following the method in @xcite , we can estimate the fluorescent ly@xmath0 emission that the incident qso flux will induce in it , as a function of its ( proper ) distance from the qso . if it were @xmath52kpc from the qso , its predicted luminosity would be @xmath53 times greater than that observed .
there are several ways to explain its faintness : * the cloud , while seen close in projection to the qso , is actually @xmath54kpc in - front of or behind the qso , and thus exposed to less ionizing radiation . *
the cloud is much smaller than the seeing disk : less than 1 kpc on a side . * the cloud , while large , is mostly optically thin , with only a small filling factor of optically thick gas reprocessing the qso flux . *
dust or optical depth effects suppress the ly@xmath0 flux . *
the qso could have been much fainter a few thousand years ago : due to light travel time , the cloud luminosity reflects the qso luminosity in the past .
if , however , we assume that the emitting gas is connected with the absorption - line system , we can deduce much more about its physical conditions , and rule out several of these possibilities .
firstly , if the emitting and absorbing gas come from different parts of the same cloud , that gives us a cloud size of at least 5kpc . secondly , the absorption consists of three sub - clumps , spread over @xmath55 .
if the cloud really had a very small filling factor of these absorbing sub - clumps , the odds of finding three along the qso sight - line would have been small .
it is therefore likely that the filling factor of sub - clumps is such that most sight - lines would intersect at least one .
thirdly , the measured neutral hydrogen column is too small to make the clouds optically thick in ly@xmath0 , and the fact that we see the background qso through the clouds implies that little dust is present .
we modeled the gas using version 06.02.09 of gary ferland s cloudy photoionisation code @xcite .
the absorbing sub - clumps were modeled as plane - parallel slabs of constant density gas , exposed to continuum emission with a typical qso spectrum , normalised to the observed luminosity of pss 2155 + 1358 .
we place a lower limit on the ionisation parameter @xmath56 .
this limit comes from our observed lower limit on the ratio of the c iv to c iii column density , and is , to first order , independent of the density .
no upper limit on @xmath57 can be placed .
the gas is therefore highly ionised : the fraction of hydrogen that is in the neutral state is @xmath58 .
while the measured neutral column density is only @xmath59 , the total ionised hydrogen column is more like that of a damped lyman-@xmath0 absorption - line system . for @xmath60
, we get a good fit to the observed column densities of both nitrogen and carbon if the metallicity is 6% ( -1.2 dex ) of solar .
if the ionisation parameter is larger , we need carbon to be overabundant with respect to nitrogen - a trend which has been seen before in qso spectra ( eg . * ? ? ?
each sub - clump reprocesses around 15% of incident ionising photons into ly@xmath0 photons .
if most sight - lines through the cloud intersect a sub - clump ( as suggested by the presence of three along the qso sight - line ) , this means that the cloud must be around 50 kpc from the qso to have the luminosity observed . if it is further , an additional ionisation source would be required .
how big are the sub - clumps ? if they are 50kpc from the qso , and @xmath61 , the inferred gas density needed to give this ionisation parameter is @xmath62 .
the total gas column is @xmath63 , giving a sub - clump thickness of @xmath64 .
if @xmath65 , the inferred density drops by a factor of ten and the ionised column increases by an order of magnitude , giving sub - clumps a hundred times as thick .
thus if the absorbing gas is associated with the ly@xmath0 emitter , we re not looking at the host galaxy of the qso .
instead , we are looking at an isolated cloud of metal - enriched gas many tens of kpc away , perhaps a satellite galaxy of the qso host .
if , however , the absorber is not connected with the emission , we have fewer constraints , and could be looking at gas closer in , the emission from which is suppressed by one of the mechanisms listed in the bullet points above .
if the emitter really is @xmath66 kpc from the qso , as required if it is connected with the absorption line , then it is a little surprising that we see it at a projected separation of only @xmath21 from the qso .
50kpc in the plane of the sky would correspond to a separation of 8 arcsec .
if emitters are distributed randomly within 50 kpc of the qso , 28% would lie within our data cube , and only 2.5% would lie within one projected arcsec of the qso .
thus if we saw anything , the odds of it being within @xmath67 are @xmath25% .
this is not unlikely enough to rule anything out , but is curious and perhaps provides weak evidence that the emitter is closer to the qso and hence not connected with the absorber .
in addition to the main emission - line at z=4.2793 , a second , more marginal line is seen at z=4.2071 ( fig [ grid565_label ] ) .
this is located @xmath68 west of the qso , and has a flux of about @xmath69 .
it too is spectrally unresolved .
formally , it is significant at about the @xmath70 level , but it is too faint to see clearly in the data from individual nights , and hence could be some sort of glitch in the data .
this emitter lies in the blue wing of a sub - damped - lyman@xmath0 absorption - line system in the qso spectrum ( fig [ qsospec_close ] ) .
this absorption system was studied by @xcite and consists of two components , at redshifts of 4.212229 and 4.212628 . if the existence of this emitter is confirmed , and if it is indeed associated with this nearby absorber , it would join the small number of such absorption - line systems seen in emission ( eg . * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and refs therein ) .
perhaps our most striking conclusion is how little neutral gas lies close to qso pss 2155 + 1358 .
we detect one isolated cloud , which is probably @xmath66kpc from the qso .
but there is no extended ly@xmath0 nebula , as seen around radio galaxies , radio - loud qsos and in ly@xmath0 blobs .
the emission we do detect is much fainter than that seen by @xcite and @xcite .
clearly the properties of gas around high redshift qsos are diverse .
based on observations obtained at the gemini observatory , which is operated by the association of universities for research in astronomy , inc .
, under a cooperative agreement with the nsf on behalf of the gemini partnership : the national science foundation ( united states ) , the particle physics and astronomy research council ( united kingdom ) , the national research council ( canada ) , conicyt ( chile ) , the australian research council ( australia ) , cnpq ( brazil ) and conicet ( argentina ) .
the program number was gn-2004b - q-21 .
data from the european southern observatory s very large telescope were also used .
we d like to thank cline peroux for making available the reduced uves spectrum of pss 2155 + 1358.0
|
we use integral field spectroscopy with the gemini north telescope to detect probable fluorescent ly@xmath0 emission from gas lying close to the luminous qso pss 2155 + 1358 at redshift 4.28 .
the emission is most likely coming not from primordial gas , but from a multi - phase , chemically enriched cloud of gas lying about 50 kpc from the qso .
it appears to be associated with a highly ionised associated absorber seen in the qso spectrum . with the exception of this gas cloud ,
the environment of the qso is remarkably free of neutral hydrogen .
we also marginally detect ly@xmath0 emission from a foreground sub - damped - ly@xmath0 absorption - line system .
[ firstpage ] intergalactic medium quasars : absorption - lines quasars : individual : pss 2155 + 1358 galaxies : high - redshift
|
in early 80 s , chinese government decided to build an @xmath0 collider running at the tau - charm energy region , called bepc , which is completed in 1989 .
the only detector at the machine is called beijing spectrometer(bes ) . in mid 90 s ,
there has been a minor upgrade of the detector , which is then called bes - ii . since then
, hundreds of papers have been published on the international journals , some with significant impacts to the community .
the upgrade of bepc was decided at the beginning of this century , called bepcii , which has a designed luminosity of @xmath1 , an increase of a factor of 100 .
the corresponding detector , called bes - iii , adopted latest detector technology to minimize systematic errors in order to match the unprecedented statistics .
the physics program of the bes - iii experiment includes light hadron spectroscopy , charmonium , electroweak physics from charmed mesons , qcd and hadron physics , tau physics and search for new physics . due to its huge luminosity and small energy spread ,
the expected event rate per year is historical , as listed in table [ tab : lum ] .
.@xmath2-charm productions at bepcii in one year s running(@xmath3s ) .
[ cols= " < , < , < , < " , ] [ tab : lum ] it is well known that j/@xmath4 and @xmath4 decays is an ideal laboratory for light hadron studies since it has a huge production cross section with a very clean and gluon reach environment .
we plan to study the meson and baryon spectroscopy , search for glueballs and other exotics such as hybrids and multi - quark states .
recently , bes - ii found several new structures and threshold enhancements in various decays channels @xcite , which leads to a number of speculations .
it is clear that more data is needed to understand these results , study their decay properties and establish a theoretical framework to accommodate them .
[ fig : x1835 ] shows a comparison of the x(1835 ) signal at bes - ii and the corresponding expectation at bes - iii .
the improvement comes mainly from the increase of luminosity by a factor of 100 , and the decrease of the energy resolution of the electromagnetic calorimeter by a factor of 10 .
similar results are expected for other new structures and threshold enhancements .
recently , a lot of new @xmath5 resonance have been found above or below the open charm threshold from b decays .
bes - iii should be able to study the direct production of 1@xmath6 states and some of the low mass states , such as y(2175 ) .
we will systematically study charmonium transitions and their decays , search for rare decays and new phenomena , and calibrate lattice qcd calculations . for charm physics
, the precision of ckm matrix elements can be significantly improved by measuring the leptonic and semi - leptonic decays of charmed mesons , and test the unitarity of ckm matrix .
the d@xmath7-mixing can be measured at the level of 10@xmath8 and the cp violation will be searched for at the level of 10@xmath9 .
rare and forbidden decays can be searched for at a typical level of 10@xmath10 .
the tau mass measurement can be improved by a factor of two over the bes - ii results with a new beam energy calibration based on the compton scattering technique .
a summary of bes - iii physics program , called yellow book , is under preparation .
it will be published at the beginning of next year .
the new bepcii has two storage rings with a circumstance of 224 m , one for electron and one for positron , each with 93 bunches spaced by 8 ns @xcite .
the total current of the beam is designed to be 0.93 a , and the crossing angle of two beams 22 mrad .
the peak luminosity is expected to be @xmath11 at the beam energy of 1.89 gev , the bunch length is estimated to be 1.5 cm and the energy spread @xmath12 . at this moment , all the linac equipments have been installed and successfully tested .
parameters such as the beam current , emittance and energy spread etc . for both electron and positron beams , have been measured .
all the design specifications have been satisfied .
the storage rings have been installed in two phases . in the first phase ,
conventional magnets at the interaction region were installed and tested .
both electron and positron beams have been stored up to 500 ma with a reasonable life time .
synchrotron radiation beams have been delivered to user for about three months in total .
tests of @xmath13 collision have been performed with an estimated peak luminosity of about @xmath14 . in the second phase , super - conducting quadrapoles for the final beam focusing at the interaction region
have been installed and beams have successfully stored .
collision of @xmath13 beams have been observed and synchrotron radiation run will start soon .
we plan to move the bes - iii detector into the interaction region early next year after the luminosity is more than @xmath15 and the beam background is under control .
the bes - iii detector @xcite , as shown in fig .
[ fig : besiii ] , consists of a drift chamber in a small cell structure filled with a helium - based gas , an electromagnetic calorimeter made of csi(tl ) crystals , time - of - flight(tof ) counters for particle identification made of plastic scintillators , a muon system made of resistive plate chambers(rpc ) , and a super - conducting magnet providing a field of 1 t
. current status of the construction is summarized in the following .
the drift chamber has a cylindrical shape with two chambers jointed at the end flange : an inner chamber without outer wall and an outer chamber without inner wall .
there are a total of 6 stepped end flanges made of 18 mm al plates in order to give space for the focusing magnets .
the inner radius of the chamber is 63 mm and the outer radius is 810 mm , with a length of 2400 mm .
both the inner and outer cylinder of the chamber are made of carbon fiber with a thickness of 1 mm and 10 mm respectively .
a total of 6300 gold - plated tungsten wires(3% rhenium ) with a diameter of 25 um are arranged in 43 layers , together with a total of 22000 gold - plated al wires for field shaping .
the small drift cell structure of the inner chamber has a dimension of @xmath16 and the outer chamber of @xmath17 , filled with a gas mixture of 60% helium and 40% propane .
the designed single wire spatial resolution and de / dx resolution are 130 @xmath18 m and 6% , respectively .
all the wiring have been completed with a very high quality , the wire tension and the leakage current are well under control .
the assembly of the chamber has been completed together with all preamplifiers and related electronics .
the whole chamber has been tested using cosmic - rays for three months .
the obtained single wire resolution is about 120 @xmath19 , as shown in fig .
[ fig : mdc - cosmic ] , well satisfying our design goal .
the chamber has now been installed successfully into the bes - iii detector .
the csi(tl ) crystal electromagnetic calorimeter consists of 6240 crystals , 5280 in the barrel , and 960 in two endcaps .
each crystal is 28 cm long , with a front face of about @xmath20 , and a rear face of about @xmath21 .
all crystals are tiled by 1.5@xmath22 in the azimuth angle and 1 - 3@xmath22 in the polar angle , respectively , and point to a position off from the interaction point by a few centimeters .
the designed energy and position resolution are 2.5% and 6 mm at 1 gev , respectively .
all the crystals have been produced and shipped , been tested , and assembled .
[ fig : csi_performance ] shows the test results of the light yield , uniformity and radiation hardness .
all the barrel crystals have been installed into the mechanical structure , which has been installed into the bes - iii detector as well .
the readout electronics of crystals , including preamplifiers , main amplifiers and charge measurement modules are tested at the ihep e3 beam line together with a crystal array and photodiodes .
results from the beam test shows that the energy resolution of the crystal array reached the design goal of 2.5% at 1 gev and the total noise achieved the level of less than 1000 equivalent electrons , corresponding to an energy of 220 kev .
the particle identification at bes - iii is based on the momentum and de / dx measurements by the drift chamber , and the tof measurement by plastic scintillators .
the barrel scintillator bar is 2.4 m long , 5 cm thick and 6 cm wide .
a total of 176 such scintillator bars constitute two cylinders , to have a good efficiency and time resolution . for the endcap ,
a total of 48 fan - shaped scintillators form a single layer .
a 2-inch fine - mesh phototube is directly attached to each end of the scintillator to collect the light .
the intrinsic time resolution is designed to be 90 ps including contributions from electronics and the common start / stop time .
such a time resolution , together with contributions from the beam size , momentum uncertainty , etc . can distinguish charged @xmath23 from k mesons for a momentum up to 0.9 gev at the 2@xmath24 level .
beam tests show that the intrinsic time resolution can be better than 90 ps and 75 ps for the barrel and the endcap tof counters , respectively .
currently , all the pmts and scintillators have been delivered and tested .
the average attenuation length of all barrel scintillators is 4.8 m and the relative light yield exceeds our specification .
all barrel scintillator have been assembled outside of the mdc and installed into the bes - iii detector successfully .
the bes - iii muon chamber is made of resistive plate chambers(rpc ) interleaved in the magnet yoke .
there are a total of 9 layers in the barrel and 8 layers in the endcap , with a total area of about @xmath25 m@xmath26 .
the readout strip is 4 cm wide , alternated between layers in x and y directions .
the rpc is made of bakelite with a special surface treatment without linseed oil @xcite .
such a simple technique for the rpc production shows a good quality and stability at a low cost .
all rpcs have been manufactured , tested , assembled and installed with satisfaction .
the bes - iii super - conducting magnet has a radius of 1.48 m and a length of 3.52 m. it use the al stabilized nbti / cu conductor with a total of 920 turns , making a 1.0 t magnetic field at a current of 3400 amp .
the total cold mass is 3.6 t with a material thickness of about 1.92 x@xmath27 . in collaboration with wang nmr of california ,
the magnet is designed and manufactured at ihep .
the magnet was successfully installed into the iron yoke of the bes - iii , together with the valve box . a stable magnetic field of 1.0 t at a current of 3368 a was achieved .
the field mapping together with super - conducting quadrapole magnets for final focusing of the beam has been completed , and results are shown in fig . [
fig : field ] .
the uniformity of the magnetic field is satisfactory .
the bes - iii offline software consists of a framework based on gaudi , a monte carlo simulation based on geant4 , an event reconstruction package , a calibration package and a database package using mysql .
currently all codes are working as a complete system , and tests using cosmic - ray data and beam test data are underway .
analysis tools such as the particle identification , secondary vertex finding , kinematic fitting , event generator and partial wave analysis are still under development .
data challenge of the whole system is planed .
in summary , the bepcii and bes - iii construction went on smoothly .
currently all the mass production of detector components have completed , most of the assembly and installation of the detector are finished .
we plan to take data in 2008 .
9 see for example , m. ablikim _
et al . _ ,
* 91 * ( 2003)022001 ; phys .
lett . * 93*(2004)112002 ; phys .
lett . * 95*(2005)262001 ; phys .
lett . * 96*(2006)162002 ; phys .
lett . * 97*(2006)142002 .
|
interesting results from bes - ii and other experiments raised actually many new questions which shall be answered by its upgrade program , bepcii and bes - iii . the design and current status of bepcii and bes - iii are reported .
|
the compas group was supported in part by the russian foundation for basic research grants rfbr-98 - 07 - 90381 and rfbr-01 - 07 - 90392 .
k.k . is in part supported by the u.s .
contract de - fg-02 - 91er40688-task a. we thank professor jean - eudes augustin for the hospitality at lpnhe - universit paris 6 , where part of this work was done .
we thank o. selyugin for his corrections to the database , a. sobol for discussions on the soft physics program of the totem & cms , and yu.kharlov for discussions on possibilities for @xmath22 physics at rhic and at the lhc .
99 computerised models , parameter evaluation for theory and experiment .
unit de recherche des universits paris 6 et paris 7 , associe au cnrs .
j. r. cudell , v. v. ezhela , p. gauron , k. kang , yu .
v. kuyanov , s. b. lugovsky , b. nicolescu , and n. p. tkachenko , phys . rev .
d * 65 * , 074024 ( 2002 ) .
j. r. cudell et al . , phys .
d * 61 * , 034019 ( 2000 ) ; phys .
d * 63*. 059901 ( e ) ( 2001 ) .
k. hagiwara et al .
( particle data group ) , review of particle physics , to appear in phys .
d. see _ e.g. _ http://www.rhic.bnl.gov/pp2pp/ and http://totem.web.cern.ch / totem/. s. i. dutta , m. h. reno , i. sarcevic and d. seckel , phys . rev .
d * 63 * , 094020 ( 2001 ) [ arxiv : hep - ph/0012350 ] . g. baur , k. hencken , d. trautmann , s. sadovsky and y. kharlov , phys . rept . * 364 * , 359 ( 2002 ) [ arxiv : hep - ph/0112211 ] .
j. r. cudell , k. kang and s. k. kim , phys .
b395 * , 311 ( 1997 ) [ hep - ph/9601336 ] .
p. desgrolard , m. giffon , e. martynov and e. predazzi , eur .
j. c * 18 * , 555 ( 2001 ) [ arxiv : hep - ph/0006244 ] .
a. donnachie and p. v. landshoff , phys
b296 * , 227 ( 1992 ) [ hep - ph/9209205 ] .
l. d. soloviev , pisma v zhetf * 18 * , 455 ( 1973 ) ; pisma v zhetf * 19 * , 185 ( 1974 ) ( in russian ) ; jetp lett .
* 19 * , 116 ( 1974 ) .
p. gauron and b. nicolescu , phys
b486 * , 71 ( 2000 ) [ hep - ph/0004066 ] .
e. m. levin and l. l. frankfurt , pisma v zhetf * 3 * , 652 ( 1965 ) ( in russian ) , e. m. levin and l. l. frankfurt , jetp lett . * 2 * , 65 ( 1965 ) .
k. johnson and s.b .
treiman , phys .
* 14 * , 189 ( 1965 ) ; p. g. o. freund , phys . rev
. lett . * 15 * , 929 ( 1965 ) .
k. igi and m. ishida , arxiv : hep - ph/0204162 .
s. chekanov _ et al . _
[ zeus collaboration ] , nucl .
b * 627 * , 3 ( 2002 ) [ arxiv : hep - ex/0202034 ] . c. avila
_ et al . _
[ e-811 collaboration ] , phys .
* b537 * , 41 ( 2002 ) .
p. gauron and b. nicolescu , phys . lett . *
b333 * , 142 ( 1994 ) ; m. g. albrow , a. beretvas , l. nodulman and p. giromini , `` comparison of the total cross sections measurements of cdf and e811 , '' preprint fermilab - tm-2071 , 1999 .
m. froissart , phys .
rev . * 123 * , 1053 ( 1961 ) ; a. martin , nuovo cimento , * a42 * , 930 ( 1966 ) ; see also w. heisenberg , z. phys . *
133 * , 65 ( 1952 ) .
the latest zeus point @xcite on @xmath75 , that supersedes the previous two zeus points , is included in the present study for rrp@xmath49l2@xmath50 .
also the whole database was changed slightly since the last session of the cross assessment @xcite due to misprints and errata corrections .
this improves slightly the global fit and central values are shifted noticeably within the error bars when one compares with results calculated from the parametrizations of @xcite .
note that we checked @xcite that the predictions of simple - pole models nevertheless fall into our allowed region .
note that , apart from the tiny statistical error , none of the errors or bands given in this work has a probabilistic interpretation .
we have constructed a preliminary version of a web - predictor , allowing one to obtain the central values and fit errors without systematics from our highest rank models at any collision energies @xmath76 5 gev , see http://www.ihep.su/@xmath77tka4ehko / cs / models/.
|
we present predictions on the total cross sections and on the ratio of the real part to the imaginary part of the elastic amplitude ( @xmath0 parameter ) for present and future @xmath1 and @xmath2 colliders , and on total cross sections for @xmath3 hadrons at cosmic - ray energies and for @xmath4 hadrons up to @xmath5 tev .
these predictions are based on an extensive study of possible analytic parametrisations invoking the biggest hadronic dataset available at @xmath6 .
the uncertainties on total cross sections , including the systematic errors due to contradictory data points from fnal , can reach @xmath7 at rhic , @xmath8 at the tevatron , and @xmath9 at the lhc , whereas those on the @xmath0 parameter are respectively @xmath10 , @xmath11 , and @xmath10 . in recent works
@xcite , we have performed an exhaustive study of the analytic parametrisations of soft data at @xmath6 . for this purpose
, we gathered the largest available set of data at @xmath12 , which includes all measured total cross sections and ratios of the real part to the imaginary part of the elastic amplitude ( @xmath13 parameter ) for the scattering of @xmath14 @xmath15 , @xmath16 , @xmath17 , and total cross sections for @xmath18 , @xmath19 and @xmath20 @xcite .
several experiments are under way @xcite , or being planned , to measure the hadronic amplitudes at @xmath6 .
some authors @xcite also presented what they feel are reference values for the total @xmath21 and @xmath4 hadrons cross sections .
thus it is timely and appropriate to present independently our predictions for the forward observables at rhic , the tevatron - run ii and the lhc as well as for @xmath21 total cross section at cosmic - ray energies and for @xmath22 total cross sections up to 1 tev .
we can summarize the general form of the parametrisations by quoting the form of total cross sections , from which the @xmath0 parameter is obtained via analyticity .
the ingredients are the contribution @xmath23 of the highest meson trajectories ( @xmath0 , @xmath24 , @xmath25 and @xmath26 ) and the rising @xmath27 term @xmath28 from the pomeron contribution to the total cross section , which can be written for the scattering of @xmath25 on @xmath29 : @xmath30 the first term is parametrised via regge theory , and we allow the lower trajectories to be partially degenerate , _
i.e. _ our experience shows that it is enough to introduce one intercept for the @xmath27 trajectories , and another one for the @xmath31 @xcite .
a further lifting of the degeneracy is certainly possible , but does not seem to modify significantly the results @xcite . hence we use @xmath32 with @xmath33 gev@xmath34 .
the contribution of these trajectories is represented by rr in the model abbreviations .
as for the part rising with energy , we consider here two main options : it can rise as a @xmath35 , or as a @xmath36 , with in each case the possibility to add a constant term .
we shall not consider the simple - pole parametrisation @xcite , not only because it is disfavored by our ranking procedure ( see below ) , but also because we want to make predictions at very high energies , where unitarisation must set in @xcite . in the following
, we shall only refer explicitly to our preferred parametrisation of @xmath28 , which we note as pl2 : @xmath37 where @xmath38 is a universal scale parameter ( to be determined by the fits ) identical for all collisions .
we have considered several possible constraints on the parameters of eqs .
( [ lower]-[pom3 ] ) : degeneracy of the reggeon trajectories ( @xmath39 ) ; universality of rising terms ( @xmath40 independent of the hadrons ) @xcite ; factorization for the residues in the case of the @xmath19 and @xmath18 cross sections ( @xmath41 ) ; quark counting rules @xcite ( predicting the @xmath42 cross section from @xmath1 , @xmath43 and @xmath44 ) ; and finally the johnson - treiman - freund @xcite relation for the cross section differences . out of the 256 possible variants
, we showed that 24 met our criteria for applicability ( an overall @xmath45 and a non - negative pomeron contribution at all energies ) if one fitted only to @xmath46 for @xmath47 gev , and 5 did for @xmath48 gev ( see table xi from @xcite ) , whereas 20 ( resp .
4 ) variants obeyed this criterion when a fit to both @xmath46 and @xmath0 was performed , for @xmath47 ( resp .
5 ) gev ( see table xiv from @xcite ) .
we shall neither give here the list of models , nor spell out ranking criteria based on new indicators that quantify certain qualities of the fits , but simply mention that the triple - pole parametrisation rrp@xmath49l2@xmath50 @xcite was determined to be the highest - ranking model leading to the most satisfactory description of the data ( see similar conclusions in @xcite ) .
this parameterization has a universal ( u ) @xmath51 term , a non - factorizing ( nf ) constant term and non - degenerate lower trajectories .
we start by giving the predictions of this model , adjusted for ( @xmath52 gev ) , with updated data points from zeus @xcite .
these predictions include statistical errors calculated from the full error matrix @xmath53 .
we define @xmath54 with @xmath55 or @xmath0 and @xmath56 the parameters of the model .
these errors are shown in figs . 1 and 2 by a filled band , and in tables [ table2 ] , [ table3 ] , and [ table4 ] .
parameter .
the curves and band are as in fig . 1 . ] in these figures and tables , we also give our estimate of the systematic uncertainty coming from the discrepancy between different fnal measurements of @xmath57 : we fit rrp@xmath49l2@xmath50 either to the high data ( cdf ) or to the low ones ( e710/e811 ) , and get two error bands .
the distances from the central value of the combined fit to the upper ( resp .
lower ) border of these bands give us the positive ( resp .
negative ) systematic errors .
we estimate the total errors as the sum of the systematic and of the statistical uncertainties @xcite .
one can see that the total errors on total cross sections are of the order of 1.9@xmath58 at rhic , of the order of @xmath8 at the tevatron and as large as @xmath9 at the lhc and dominated by the systematic errors .
the errors on the @xmath0 parameter are much larger , reaching 5.4@xmath58 at rhic , 5.2@xmath58 at the tevatron and 5.4@xmath58 at the lhc .
this is due to the fact that experimental errors are bigger , hence less constraining , but this also stems from the incompatibility of some low - energy determinations of @xmath0 @xcite .
this means that the systematic error is always bigger than the statistical one . concerning the contradictory data , we are forced to use them in our fits until the discrepancy is resolved by further experiments . in the case of the tevatron data
, one can see that the discrepancy results in a big shift ( of more than @xmath59 ) in the central value of the coefficient @xmath60 of the @xmath36 term , which controls the asymptotic behavior , and hence that asymptotic predictions are appreciably weakened by the present situation .
the opportunities to measure @xmath57 and @xmath0 will be scarce in the future , hence any new measurement at rhic , the tevatron run ii and the lhc should not be missed .
unfortunately , the recent publication of e-811 @xcite does not clear the problem as their value for @xmath0 is fully compatible with our preferred model , whereas their number for @xmath57 ( which is highly correlated with @xmath0 ) has hardly changed .
it is interesting to note that the choice of one fnal result or the others leads to a variation of the overall fit quality , as shown in table [ table1 ] ( last two columns ) @xcite : the variant with cdf data has slightly better overall @xmath61 and better @xmath62 distribution over sub - samples .
we can consider this as an indication that the global picture emerging from fits to all data on forward observables supports the cdf data and disfavors the e710/e811 data at @xmath63 tev ( see also an analogous conclusion based on the other arguments in @xcite ) .
lcccc & db02 & db02z & db02z & db02z + sample & & & @xmath64cdf & @xmath64e710/e811 + total & 0.968 & 0.966 & 0.964 & 0.951 + + @xmath65 & 1.15 & 1.15 & 1.12 & 1.05 + @xmath66 & 0.84 & 0.84 & 0.84 & 0.84 + @xmath67 & 0.96 & 0.96 & 0.96 & 0.96 + @xmath68 & 0.71 & 0.71 & 0.71 & 0.71 + @xmath69 & 0.62 & 0.62 & 0.62 & 0.61 + @xmath70 & 0.71 & 0.71 & 0.71 & 0.71 + @xmath71 & 0.38 & 0.38 & 0.38 & 0.38 + @xmath21 & 0.61 & 0.58 & 0.58 & 0.58 + @xmath22 & 0.65 & 0.64 & 0.64 & 0.63 + + @xmath65 & 0.52 & 0.52 & 0.52 & 0.53 + @xmath66 & 1.83 & 1.83 & 1.83 & 1.80 + @xmath67 & 1.10 & 1.10 & 1.09 & 1.14 + @xmath68 & 1.50 & 1.50 & 1.52 & 1.46 + @xmath69 & 1.00 & 0.99 & 1.01 & 0.96 + @xmath70 & 1.07 & 1.07 & 1.10 & 0.98 + + & 0.307(10)&0.307(10)&0.301(10)&0.327(10 ) + [ table1 ] finally , we also present in figs
. 1 and 2 our estimate of the region where new physics would be discovered .
for each of the 20 parametrisations which satisfy our criteria for applicability @xcite for @xmath47 gev , and which obey the froissart - martin bound @xcite , we construct errors bands according to eq .
( [ staterr ] ) .
this gives us 20 @xmath59-error bands .
their union represents the allowed region " where analytic models built according to ( [ mods ] ) can reproduce the data . a measurement outside of this region would imply that new physics ingredients are needed .
to conclude , we believe that we have given here the best possible estimates for present and future @xmath1 and @xmath2 facilities .
although one might be tempted to use only data in an energy range close to the one measured , one must realize that analytic parametrisations are constrained both by lower - energy data , and by their asymptotic regime . because the pomeron mixes ( physically and numerically ) with the @xmath26 trajectory , fits to all data help to disentangle the two contributions .
a sharpening of our error bars would enable one to decide if the unitarisation plays an essential role and what form it takes .
this in turn can have an impact on the determination of the survival of probability gaps in hard scattering , and on the usefulness of pomeron exchange as a detection tool .
any significant deviation from the predictions based on model rrp@xmath49l2@xmath50 will lead to a re - evaluation of the hierarchy of models and presumably change the preferred parametrisation to another one . a deviation from the allowed region " would be an indication that strong interactions demand a generalization of the analytic models discussed so far , _
e.g. _ by adding odderon terms , or new pomeron terms , as suggested by qcd .
.predictions for @xmath57 and @xmath0 , for @xmath72 ( at @xmath73 gev ) and for @xmath1 ( all other energies ) .
the central values and statistical errors correspond to the preferred model rrp@xmath49l2@xmath50 , and the systematic errors come from the consideration of two choices between cdf and e-710/e-811 @xmath74 data in the simultaneous global fits . [ cols="^,^,^",options="header " , ] [ table4 ]
|
massive stars ( @xmath0@xmath18 m@xmath2 , @xmath3@xmath110@xmath4 l@xmath2 ) play a fundamental role in many areas of astrophysics .
they are the principal source of uv radiation and heavy elements in galaxies , and are responsible for injecting huge amounts of kinetic energy into the ism through powerful molecular outflows , strong stellar winds and supernova explosions .
the momentum imparted through these processes provides an important source of mixing and turbulence within the ism .
they are also thought to play a key role in regulating star formation , either by disrupting molecular clouds before stars have been able to form , or constructively through the expansion of their hii regions ( e.g. , collect and collapse ) or the propagation of strong shocks into their surroundings ( e.g. , radiatively driven implosion ) .
massive stars have an enormous influence not only on the physical structure and chemistry of their local environment , but also on the structure and evolution of their host galaxies . despite the importance of massive stars
the processes involved in their formation and the early stages of their evolution are still poorly understood .
massive stars reach the main sequence while still deeply embedded within their natal molecular clouds , and therefore their entire pre - main sequence formation and evolution is hidden behind high levels of extinction .
they are extremely rare and as a consequence are generally located farther away than regions of low - mass star formation .
the large distances are compounded by the fact that massive stars are through to form exclusively in clusters and limited spatial resolution makes it difficult to identify and attribute derived quantities to individual sources .
furthermore , the evolution of massive stars is extremely rapid , which means that key stages in their early evolution are short lived . due to these observational difficulties , until relatively recently , the only catalogue of massive young stellar objects ( mysos ) had been limited to 30 or so serendipitously detected sources ( @xcite ) most of which are nearby .
as massive stars reach the main sequence while still deeply embedded , the obvious place to begin to search for mysos is at infrared wavelengths where the majority of the bolometric luminosity ( @xmath110@xmath4 l@xmath2 ; @xcite ) is emitted after reprocessing by dust .
nuclear fusion has almost certainly begun due to the very short kelvin - helmholtz contraction time scale compared to their free - fall time scale ( @xcite ) .
accretion is likely to be taking place as the majority of mysos are associated with massive outflows ( @xcite ) , generally thought to be powered by accretion .
although fusion is taking place , the ongoing accretion is thought to impede the formation of an hii region either by : 1 ) significantly increasing the star s radius , leading to a lower effective stellar temperature and characteristic accretion disk temperature ( @xcite ) , or 2 ) the hii region being quenched by high infall rates ( @xcite ) .
mysos are also known to possess ionised stellar winds that are weak thermal radio sources ( @xmath51 mjy at 1 kpc ; @xcite ) .
mysos can be roughly parametrised as mid - infrared bright , and thus slightly later than the hot core phase , and radio quiet , and thus earlier than the uchii region phase , with luminosities consistent with young o and early b - type stars . to date
there have been several attempts to search for mysos using colour - selection criteria and the iras point source catalogue ( e.g. , @xcite ) . however , due to confusion in the large iras beam ( @xmath53 - 5 at 100 @xmath6 m ) these samples tended to avoid dense clustered environments and the galactic mid - plane where the majority of mysos are expected to be located since the scale height of massive stars is @xmath530(@xcite ) .
although these iras selected samples have identified many genuine mysos , they tend to be biased towards bright , isolated sources that are probably not representative of the general population of mysos . there is a need for an unbiased sample of mysos that is large enough to have a sufficient number of sources in each luminosity bin to allow the processes involved in massive star formation to be investigated in a statistically robust way . the midcourse space experiment ( msx ; @xcite ) surveyed the whole galactic plane ( @xmath7@xmath85@xmath9 ) in four mid - infrared bands centred at 8 , 12 , 14 and 21 @xmath6 m at a resolution of 18 in all bands .
although the sensitivity of msx is similar to that of iras , its beam size is @xmath550 times smaller than that of the iras 12 and 25 @xmath6 m bands and thus avoids most of the confusion problem found with iras .
the msx survey therefore offers us the opportunity , for the first time , to identify a truly representative sample of mysos located throughout the galaxy .
more recently the spitzer satellite has surveyed a large region of the inner galaxy ( glimpse ; @xcite ) with unprecedented spatial resolution and sensitivity , offering a great improvement on msx data .
however , most mysos are found to be highly saturated and would be missed in any glimpse colour - selected sample . in order to select potential mysos we compared the colours of known mysos with sources identified in the msx and 2mass point source catalogues ( @xcite and
@xcite respectively ) and developed colour selection criteria ( @xcite ) .
a lower limit of 2.7 jy was placed on the 21 @xmath6 m flux .
the detection of a source in 2mass was not necessary as mysos can be totally obscured in the near - infrared ; these data were just used to exclude sources that are too blue .
the msx images of all colour selected sources were visually inspected to eliminate sources that are not point - like ( see below ) .
sources toward the galactic centre ( @xmath10@xmath9 ) were excluded due to confusion and difficulties in calculating kinematic distances . in total
we identified approximately 2000 myso candidates .
the shape of the spectral energy distribution from an optically thick cloud is insensitive to the type of heating source .
this is a problem with a colour - selected sample such as ours , as it results in contamination of our sample by several other types of embedded , or dust enshrouded objects that have similar near- and mid - infrared colours to mysos ( e.g. , ultra compact ( uc ) hii regions , evolved stars and planetary nebulae ( pne ) ) .
the core of the rms survey is a multi - wavelength programme of follow - up observations designed to distinguish between genuine mysos and these other types of embedded or dusty objects ( @xcite ) and to compile a database of complementary multi - wavelength data with which to study their properties .
the main source of contamination of our sample is likely to be by uchii regions .
these can easily be identified through the free - free emission emitted by their ionised nebular gas .
radio continuum observations are thus an essential part of our follow - up programme .
we have completed a programme of 5 ghz observations using the atca ( @xcite ) and the vla ( urquhart et al . in prep . ) of all mysos candidates .
these observations have a spatial resolution of 1 and a sensitivity of 1 mjy ( 3@xmath11 ) , sufficient to detect a b0.5 or earlier type star on the far side of the galaxy ( @xcite ) .
about 25% of rms sources have detectable radio emission , the majority of which are classified as uchii regions , due to their morphologies and tight correlation around the galactic mid - plane .
mid - infrared imaging has been used to identify genuine point sources from more extended emission . given the typical size of mysos at 10 @xmath6 m is 0.005 pc ( @xcite ) and a spatial resolution of 1 we would not expect to be able to resolve a myso at distances @xmath1 1 kpc .
however , uchii regions with a typical size of @xmath50.1 pc would be resolved at a distance of 15 kpc ( @xmath51.4 ) as would most pne .
therefore , mid - infrared imaging can prove extremely useful in identifying , and removing extended mid - infrared objects .
moreover , mid - infrared imaging is complementary to the radio continuum observations and can be used to obtain accurate astrometry and avoid excluding mysos that are located in close proximity to uchii regions . for the majority of our sources
the glimpse data have been used for this purpose , however , the @xmath5700 sources located outside the glimpse region have been observed using ukirt , gemini and the eso 3.6 m telescope at la silla ( @xcite ) . to obtain kinematic distances
we have made molecular line observations towards every source using @xmath12co ( @xmath13=21 ) and ( @xmath13=10 ) transitions and mopra , onsala and purple mountain observatory telescopes , and the jcmt ( @xcite ; urquhart et al .
submitted to a&a ) .
we have complemented these observations with archival data extracted from the galactic ring survey ( grs ; see @xcite for details ) . in cases
where multiple co components are detected towards an rms source we have used a combination of nearby methanol and water masers , or cs ( @xmath13=21 ) observations to identify the component associated with the myso candidate .
we have calculated kinematic distances using the source v@xmath14 and the rotation curve of @xcite ; these are crucial for calculating luminosities and thus distinguishing between nearby low- and intermediate - mass ysos from genuine mysos .
the majority ( @xmath580% ) of our sample is located within the solar circle and therefore the rotation curve produces two possible kinematic distances , referred to as the ` near ' and ` far ' distances ; this is known as the kinematic distance ambiguity .
we have used the hi self - absorption technique ( @xcite ) and hi data from the international galactic plane survey ( igps ; for details see http://www.ras.ucalgary.ca/igps/ ) to resolve these distance ambiguities ( e.g. , @xcite )
. while the observations just discussed are useful in eliminating the vast majority of uchii regions , pne and low- and intermediate - mass ysos , they are not so effective at eliminating evolved stars .
we have eliminated most of these using near infrared images in combination with other available data since they tend to appear as isolated point sources at all wavelengths , and are not generally associated with strong co emission ( @xmath15co ( @xmath13=10 and @xmath13=21 ) typically less than 1 k ; @xcite ) .
in contrast , near - ir images of mysos often show associated shocked emission from outflows , reflection nebulosity , extinction lanes , as well as clusters of lower mass ysos .
the combination of near- and mid - infrared imaging , radio continuum and molecular line observations allows us to eliminate most of the contaminating sources . however , there are a few weak uchii regions , or chance alignments of evolved stars with molecular material , that are not identified .
to weed out any remaining contaminating sources , and to confirm the identifications of our final sample of mysos , we are obtaining near - infrared spectroscopy from which a definitive classification can be made ( e.g. , @xcite ) .
we have obtained spectra towards approximately 350 rms sources so far .
classification and distribution of rms sources identified so far using data obtained from our multi - wavelength observations.,scaledwidth=59.0% ]
as the observational data have become available they have been used to classify the sources in our sample .
so far we have classified all but a few of the @xmath52000 sources and have identified @xmath5450 ysos and a further @xmath5500 uchii regions ( see fig .
[ fig : classification ] for a detailed breakdown ) . along with the ysos and uchii regions
so far identified , a further @xmath5200 sources have been identified as either ysos and/or uchii regions .
we are waiting on additional data before making a final classification .
potentially , taking into account these sources the final number of ysos and uchii regions identified so far will be significantly larger .
left panel : histogram of the luminosities for sources identified as either ysos or uchii regions and for which good iras data was available .
these luminosities have been calculated following @xcite .
right panel : distribution of source luminosities as a function of heliocentric distance.,title="fig:",scaledwidth=49.0% ] left panel : histogram of the luminosities for sources identified as either ysos or uchii regions and for which good iras data was available .
these luminosities have been calculated following @xcite .
right panel : distribution of source luminosities as a function of heliocentric distance.,title="fig:",scaledwidth=49.0% ] in the left panel of fig .
[ fig : luminosity ] we present a histogram of the luminosities of sources that have been classified as either ysos or uchii regions and for which good quality iras data are available ( @xmath525% ) , and for which any distance ambiguities have been resolved . in the right panel of fig . [
fig : luminosity ] we plot the distribution of luminosities as a function of heliocentric distance .
these plots reveal that a number of the sources identified as ysos are actually nearby low- and intermediate - mass ysos . however , the vast majority have luminosities consistent with late o and early b - type stars .
galactic distribution of rms sources identified as either uchii regions or ysos .
the location of the spiral arms taken from model by @xcite and updated by @xcite are over - plotted in grey .
the positions of the galactic centre and sun are shown by black filled and unfilled circles respectively .
the roman numerals in the corners refer to the galactic quadrants and the two thick black lines which originate from the location of the sun enclose the region excluded from our survey .
the two dashed circles indicate the solar circle and tangent points.,title="fig:",scaledwidth=99.0% ] + we have determined kinematic distances for all ysos and uchii regions , for which an unambiguous v@xmath14 could be attributed , using the rotation curve of @xcite . using these distances and source positions in fig .
[ fig : gal_rms_distribution ] we plot the distribution of ysos and uchii regions on a face - on view of the galactic plane . on this plot
we have indicated the positions of the galactic centre , the sun , the solar circle and the spiral arms ( see figure caption for details ) .
massive stars are thought to be primarily associated with the spiral arms , and thus could potentially be used as a probe of galactic structure .
comparing the distribution of our sources and the position of the spiral arms ( from the model by @xcite and updated by @xcite ) we find them to be reasonably well correlated with each other , especially taking into account the errors in the kinematic distances ( typically @xmath161 kpc ) .
the correlation is particularly apparent towards the inner galactic arms ( i.e. , 1st and 4th quadrants ) , but quite poor for sources in the outer galaxy .
the rms survey aims to produce a large , unbiased sample of massive young stellar objects ( mysos ) located throughout the galaxy .
we have colour - selected approximately 2000 myso candidates from the msx and 2mass point source catalogues that have colours similar to known mysos .
we have almost completed a multi - wavelength programme of follow - up observations to distinguish between genuine mysos and types of embedded or dust enshrouded objects that have similar mid - infrared colours that contaminate our sample .
so far using these observations we have unambiguously classified @xmath590% of our sample , approximately half of which are identified as either ysos and uchii regions .
the luminosities of a sub - sample of these sources , for which good iras data are available , confirms that we are detecting a significant number of genuine mysos .
when complete , we estimate our database will contain a well selected sample of @xmath5500 bona fide mysos and a further 500600 uchii regions , along with complementary multi - wavelength observations . now
the classification is almost complete we have begun a number of observational programmes to exploit our sample and to address many open questions concerned with massive star formation .
these include : high resolution observations of potentially triggered regions to investigate different triggering mechanisms ; chemical surveys using millimetre lines and ir absorption features to investigate the chemistry and possible chemical evolution of mysos ; investigation of molecular outflows associated with mysos ; and high resolution spectroscopy of accretion disks .
all of these programmes are being conducted on sub - samples selected as a function of luminosity , distance and location .
the whole sample and all of the observational results of our multi - wavelength campaign are available at www.ast.leeds.ac.uk/rms
. we would like to thank the directors and staff of the various telescopes mentioned in this article for their help and support during our observations .
jsu is supported by a postdoctoral grant from the science and technology facilities council ( stfc ) .
churchwell , e. , wolfire , m. g. , & wood , d. o. s. 1990 , , 354 , 247 clarke , a. j. , lumsden , s. l. , oudmaijer , r. d. , busfield , a. l. , hoare , m. g. , moore , t. j. t. , sheret , t. l. , & urquhart , j. s. 2006 , , 457 , 183 cordes , j. m. 2004 , milky way surveys : the structure and evolution of our galaxy , 317 , 211 molinari , s. , brand , j. , cesaroni , r. , & palla , f. 1996 , , 308 , 573 mottram , j. c. , hoare , m. g. , lumsden , s. l. , oudmaijer , r. d. , urquhart , j. s. , sheret , t. l. , clarke , a. j. , & allsopp , j. 2007 , arxiv e - prints , 709 , arxiv:0709.2040 , a&a in press .
urquhart , j. s. , busfield , a. l. , hoare , m. g. , lumsden , s. l. , clarke , a. j. , moore , t. j. t. , mottram , j. c. , & oudmaijer , r. d. 2007 , , 461 , 11 urquhart , j. s. , et al .
2007 , , 474 , 891 walmsley , m. 1995 , revista mexicana de astronomia y astrofisica conference series , 1 , 137 walsh , a. j. , hyland , a. r. , robinson , g. , & burton , m. g. 1997 , , 291 , 261
|
here we describe the red msx source ( rms ) survey which is the largest , systematic , galaxy - wide search for massive young stellar objects ( mysos ) yet undertaken .
mid - ir bright point sources from the msx satellite survey have been followed - up with ground - based radio , millimetre , and infrared observations to identify the contaminating sources and characterise the mysos and uchii regions . with the initial classification
now complete the distribution of sources in the galaxy will be discussed , as well as some programmes being developed to exploit our sample .
|
considerable progress has been made in recent years in our understanding of the mechanism of prestin - based somatic motility , or `` electromotility , '' of outer hair cells ( ohcs ) in the cochlea , both on cellular- @xcite and the molecular levels @xcite , as well as clarifying its physical basis that it is based on electromechanical coupling @xcite .
for example , experiments on isolated ohcs have determined load - free displacement @xcite and isometric force production @xcite .
these experimental observations can be described by static models @xcite . nonetheless , these are the conditions under which those cells do not provide energy .
some theoretical works have addressed energy production by ohcs @xcite by extrapolating from these _ in vitro _ conditions .
however , these analyses do not provide an equation of motion or the dependence of nonlinear capacitance on external mechanical load , the essential features to describe the production of mechanical energy for amplifying acoustic signal .
there are a number of issues to be addressed for describing ohcs in a dynamic environment .
one such issue is the frequency dependence of the motile response .
the amplitude of displacement in response to voltage changes rolls off at about 15 khz under load - free condition , while force production near isometric condition remains flat up to 60 khz @xcite .
this difference likely indicates that the frequency response depends on the mechanical load , more specifically viscoelastic drag .
another issue is attenuation of the receptor potential by the membrane capacitance at operating frequencies @xcite .
the membrane capacitance consists of two main components .
one is structural and is proportional to the membrane area .
another is nonlinear component associated with the mobile charge of prestin , which flips in the electric field on conformational changes .
nonlinear capacitance has been described under load - free condition , at which it is expected to be the largest .
since constraint on the membrane area almost eliminates the nonlinear component @xcite , it could be expected that external load reduces this component and thus reduces capacitive current , which attenuates the receptor potential at higher frequencies .
this feedback would improve the performance of ohcs , particularly at high frequencies .
this effect still needs to be described quantitatively .
previous treatments used either load - free capacitance @xcite or the linear capacitance alone @xcite . here a model is presented for describing the motion of a single ohc under mechanical loads . in the following ,
the basic equations are derived .
that is followed by derivations of quantities that characterize the motile element .
these quantities are determined using experimental values for ohc that operates at 4 khz .
then the balance of energy input and output is examined for those cells under mass - free condition .
the implications are discussed on those cells , which operate at higher frequencies .
here we consider a simple system , which consists of an ohc , elastic load , drag , and a mass ( fig . [ fig : system ] ) .
this is not to approximate the organ of corti , but to describe an ohc , which is subjected to mechanical load .
this system with a single degree of freedom is described by writing down the equations for the motile mechanism , the receptor potential , and the equation of motion .
these equations are interrelated and they constitute a set of simultaneous equations , which are examined in the subsequent sections .
outer hair cells ( ohcs ) have a motile mechanism driven by the membrane potential based on mechanoelectric coupling . here a one - dimensional model is used instead of a membrane model @xcite for simplicity .
we assume that the cell has @xmath0 motile elements , which has two discrete states , compact and extended , and during a transition from the compact state to the extended state , the cell length increases by @xmath1 and the electric charge @xmath2 flips across the plasma membrane .
let @xmath3 be the fraction of the motile units in the extended state .
its equilibrium value @xmath4 follows the boltzmann distribution , @xmath5}{1+\exp[-\beta\delta g]},\end{aligned}\ ] ] with @xmath6 , where @xmath7 is boltzmann s constant and @xmath8 the temperature , and @xmath9 represents the difference in the free energy in the two states , referenced to the compact state . here
@xmath10 is the force applied in the direction of extending the cell and @xmath11 the membrane potential .
the quantity @xmath12 is a constant that determines the operating point . here
both @xmath2 and @xmath1 are positive because rising membrane potential and decreasing extensive force increases @xmath13 , and thus reduces the fraction @xmath4 of the extended state .
now consider a case , in which the motile mechanism is connected to an external elastic element ( fig . [
fig : system ] without the mass or the dashpot ) .
the force @xmath10 applied to the cell depends on the elastic elements as well as the conformational change of the motile element elicited to the voltage change .
the elastic elements include the material stiffness @xmath14 of the cell as well as an external elastic load @xmath15 .
assume that at the membrane potential @xmath11 changed from its resting value @xmath16 .
the resulting displacement @xmath17 of the cell produces force @xmath10 on the external spring .
the same force is applied to the cell reciprocally , producing a displacement @xmath18 .
thus the displacement @xmath17 is determined from @xmath19 , where the fraction @xmath3 of the extended state , is changed from its resting value @xmath20 .
this change is expressed by , @xmath21 with effective spring constant @xmath22 .
notice that an increase in @xmath3 , which increases @xmath17 , generates force in the contracting direction and that the maximal value of @xmath23 is @xmath14 .
this force @xmath10 is applied to the motile element in the cell .
this leads to the expression for the difference @xmath13 in the free energy of the two states , @xmath24 notice here that an increase in @xmath3 increases the energy @xmath13 , making a further increase less favorable .
if the system is in equilibrium , @xmath25 and the equilibrium value of @xmath4 is determined by combining eq .
[ eq : distr ] with eq .
[ eq : energy ] , in which @xmath3 is substituted with @xmath4 . under this condition , all forces are balanced and no movement takes place . here
we notice that @xmath13 , and therefore @xmath4 , can be determined for any set of @xmath11 and @xmath3 by combining eqs .
[ eq : distr ] and [ eq : energy ] .
since @xmath26 in equilibrium , the difference @xmath27 quantifies the deviation from equilibrium that moves the cell .
now we examine how to describe the movement of the cell , which has mechanical loads ( fig .
[ fig : system ] ) .
the system has a single degree of freedom and is described by using length displacement @xmath17 of the cell as the variable .
what is the force that drives the cell ? let the displacement be fixed at @xmath17 and the state of the motile element be @xmath4 in the beginning .
suppose the membrane potential suddenly changes at a certain moment .
the cell generates force that moves the cell .
the force that drive the cell is @xmath28 if the stiffness of the cell is @xmath14 and if @xmath29 is small enough so that the linear term is dominant . even though the motile element contributes to reduce the stiffness from the intrinsic stiffness @xmath14 , in section [ sec : param ] we confirm that this is indeed a good approximation for a set of parameter values that we use .
then , the equation of motion should be , @xmath30 where @xmath31 is the drag coefficient , and @xmath32 the mass . for a given displacement @xmath17 , @xmath3 can be given by eq .
[ eq : x ] .
since @xmath20 is a time - independent constant , its time derivatives does not contribute .
the equation of motion can then be expressed , @xmath33 notice that a factor @xmath34 drops out from the equation because it is shared by all terms . in the special case of @xmath35 ,
[ eq : motion ] turns into a relaxation equation with a time constant @xmath36 , which is intuitive .
the motile response of the cell is driven by the receptor potential , generated by the receptor current , which is , in turn , elicited by changes in the hair bundle resistance by mechanical stimulation .
this current is driven by the sum of two electromotive forces .
one is @xmath37 , which is primarily determined by the k@xmath38 conductance of the basolateral membrane of the cell .
the other is the endocochlear potential @xmath39 , which is generated by the stria vascularis , a tissue that lines a part of the scala media ( fig .
[ fig : system]b ) .
the magnitude of the receptor potential is determined not only by changes in the hair bundle conductance but also by the basolateral conductance because the electric current through the apical membrane must be equal to the current through the basolateral membrane .
changes in the membrane potential elicit not only ionic currents but also a capacitive current proportional to the regular capacitance in the basolateral membrane .
in addition , they flip the charge of the motile units , produces an additional current , similar to the capacitive current .
thus , kirchhoff s law leads to , @xmath40 here @xmath41 is the apical membrane resistance , which is dominated by mechanotransducer channels in the hair bundle .
the basolateral membrane has the resistance @xmath42 and the linear capacitance @xmath43 , which is determined by the membrane area and the specific membrane capacitance of @xmath44 f / m@xmath45 for biological membranes @xcite .
the apical membrane capacitance is ignored for simplicity .
the last term on the right - hand - side of eq .
[ eq : potential ] is an additional displacement current due to the charge movement in the motile elements .
it has the negative sign because a voltage increase results in a decrease in @xmath3 as mentioned earlier .
as we will see later , nonlinear capacitance appears from this term .
in the following , experimentally observable quantities are derived from the model so that values for the cellular parameters can be determined from experimental data .
length changes of ohcs have been quantified by changing the membrane potential gradually or stepwise while measuring cell length without load .
that corresponds to describing @xmath4 as a function of @xmath11 under the condition of @xmath46 .
thus , eq . [ eq : distr ] is accompanied by @xmath47 instead of eq .
[ eq : energy ] .
this equation has the same form as the one that has been used for fit experimental data @xcite .
isometric force generation per voltage changes can be obtained by evaluating @xmath48 from the equation @xmath49 with eq .
[ eq : distr ] and @xmath50 $ ] for a given displacement @xmath17 .
this leads to , @xmath51 with @xmath52 .
the dependence on length displacement @xmath17 enters through the value of @xmath4 .
the maximum value is @xmath53 at @xmath54 .
the effective compliance of the cell can be determined by @xmath55 for a given voltage @xmath11 .
if we introduce the effective stiffness @xmath56 , this leads to , @xmath57 thus , the minimal value @xmath58 of the effective stiffness is @xmath59 at @xmath54 .
another characteristic quantity that described the motile element is a contribution of the motile element to the membrane capacitance , which is often referred to as nonlinear capacitance .
let us consider small periodic changes with amplitude @xmath60 in the membrane potential on top of a constant value @xmath61 , @xmath62.\end{aligned}\ ] ] then the response can be described by , @xmath63 , \\
\label{eq : p - period } p(t ) = & \overline{p}+p \exp[i\omega t],\end{aligned}\ ] ] where the variables expressed in lower case letters are small and those marked with bars on top are time - independent . under time - independent condition ,
@xmath64 and @xmath65 is expressed by eq .
[ eq : distr ] with @xmath13 , in which @xmath3 is replaced by @xmath65 .
if the amplitude @xmath60 is small , we can ignore second - order terms , eqs .
[ eq : distr ] and [ eq : motion ] respectively lead to , @xmath66 these equations lead to , @xmath67 with constants @xmath68 , @xmath69 , and @xmath70 , which respectively characterizes viscoelasticity , resonance , and the operating point : @xmath71 the capacitive current due to voltage changes is @xmath72 $ ] , which should be also expressed as @xmath73 $ ] using a component @xmath74 of the membrane capacitance .
hence , @xmath75 $ ] because conformational change @xmath76 of each motile element carries charge @xmath76 and the cell has @xmath0 such elements .
the membrane capacitance @xmath77 is the sum of @xmath74 and the structural capacitance @xmath43 , due primarily to lipid bilayer of the plasma membrane . in the following ,
the contribution to the capacitance is examined .
the order is from the most restricted case , where this quantity is better studied , to more general cases .
let us start with the simplest case , where @xmath79 and @xmath80 .
that leads to the total membrane capacitance @xmath77 , expressed by @xmath81 letting @xmath82 or @xmath83 , we recover the familiar expression for the membrane capacitance of outer hair cells @xmath84 under load - free condition @xcite .
in addition , this expression shows that nonlinear capacitance decreases with increasing elastic load .
that is consistent with intuition that constraints on the cell length reduce nonlinear capacitance . in an extreme limit ,
in which a rigid load does not allow transitions of the motor elements , nonlinear component diminishes .
this expectation is consistent with greatly diminished nonlinear capacitance observed in rounded ohcs with constrained membrane area @xcite . now let us proceed to a more general case , in which the viscous term does not disappear .
the total membrane capacitance @xmath77 is expressed , @xmath85 as expected , in the limit of low frequency , this expression turns into eq .
[ eq : capk ] . in the limit of @xmath83
, it leads to the expression of the frequency dependence of nonlinear capacitance that was previously derived based on an assumption that transition rates between the states are intrinsic @xcite .
the present interpretation supports the interpretation that the experimentally observed frequency roll - off @xcite of the membrane capacitance is indeed the result of viscoelastic relaxation @xcite .
it should be noted that the imaginary part of @xmath76 contributes to the conductance .
however , it diminishes in both low frequency limit and high frequency limit .
a similar evaluation of nonlinear capacitance can be performed for a system with non - zero mass , which has mechanical resonance .
the quantity @xmath76 is expressed by , @xmath86}{[1+\gamma a^2 n \tilde{k}-(\omega/\omega_{r})^2]^2+(\omega/\omega_{\eta})^2 } , \label{eq : cm - mres}\ ] ] under the condition @xmath87 , eq . [ eq : cm@m=0 ] can be obtained .
it should be noted that nonlinear capacitance disappears and @xmath88 at @xmath89 .
in addition , the capacitance becomes quite singular near this resonance frequency as the characteristic frequency @xmath68 of viscoelasticity exceeds the resonance frequency @xmath90 .
a reduction in the capacitance @xmath77 near resonance leads to an increase in the receptor potential because it reduces the attenuation due to the resistance - capacitance ( rc ) circuit in the cell .
that is analogous to piezoelectric resonance .
.a set of parameter values for outer hair cells with @xmath91 pf linear capacitance @xmath43 .
these values corresponds to ohcs from the region with best frequency of 4 khz for rats and gerbils @xcite .
the quantities in the upper rows are electric properties and lower rows characterize the motile element .
the values for resistances are from rats and values of motility - related parameters are from guinea pigs . @xmath92
the sum of the endocochlear potential ( @xmath9390 mv ) @xcite and the reversal potential ( @xmath37 ) of the basolateral membrane ( @xmath94 mv ) .
@xmath95 @xcite .
@xmath96 @xcite . @xmath97
the unit mobile charge @xmath2 ( in the electronic charge @xmath98 ) , and its number @xmath0 have been determined by nonlinear component of the membrane capacitance . @xmath99
the amplitude of load free displacement @xmath100 is between 4 and 5% of the total length . @xmath101 both force generation and the elastic modulus do not show length dependence @xcite .
the stiffness value corresponds the elastic modulus of 0.51 @xmath102n per unit strain for a 30 @xmath102m - long cell . [ cols="^,>,>,<",options="header " , ] now we determine the parameters using experimental data .
those quantities important for determining the parameters are : the amplitude @xmath100 of the load - free mechanical displacement , which is between 4 and 5 % of the cell length , the steepness of load - free mechanical displacement and that of nonlinear capacitance , both of which are characterized by @xmath2 , and the number @xmath0 of the motile elements in the cell , which is obtained by dividing the total charge movement @xmath103 divided by @xmath2 . in addition , experimental values are available for the maximal value of nonlinear capacitance @xmath104 , the axial stiffness of the cell @xmath56 , and isometric force production @xmath48 .
the parameter values are listed in table [ table : paramvals ] . for numerical examination
, we assume that the cell length of the ohc is 30 @xmath102 m and that the motor is 40 % in the elongated state in the natural length and the receptor potential is generated by 10% change in the resting hair bundle resistance @xmath41 , which corresponds to the condition where 30% of the mechanotransducer channels are open . here
we now examine the consistency with an assumption , which we made for deriving the equation of motion ( eq . [ eq : motion ] ) .
recall that the minimal value of the stiffness of the motile element is @xmath105 at @xmath106 ( eq . [ eq : k ] ) . for the present set of the parameter values ,
the value for this stiffness is @xmath107 n / m , larger than @xmath108 n / m for the intrinsic stiffness @xmath14 .
thus this set of parameter values is consistent with the assumption made earlier .
the consistency of the one - dimensional model is tested by comparing its predicted values and experimental values for force generation @xmath48 and the stiffness @xmath56 .
the predicted values are within experimental errors ( table [ table : paramvals ] ) .
here we assume small periodic changes with an angular frequency @xmath109 from a resting resistance @xmath110 of the hair bundle resistance , @xmath111,\ ] ] leads to the receptor potential @xmath112 $ ] .
[ eq : potential ] leads to @xmath113 with @xmath114 . the amplitude @xmath60 of the potential leads to motile responses as described by @xmath76 and @xmath115 in eqs .
[ eq : pinf - period ] [ eq : oscil - motion ] .
the receptor potential with amplitude @xmath60 elicited by small periodic changes in the hair bundle resistance induces changes in the amplitude @xmath76 of changes in the fraction of the extended state of the motile element , resulting in displacements and force generation of the cell . because of mechanoelectric coupling , the load on the motile element reciprocally affects @xmath76 , which , in turn , attenuates the receptor potential @xmath60 .
from here on , energy balance is examined for small periodic stimulation at the hair bundle in a mass - free system ( @xmath35 ) using the parameter values in table [ table : paramvals ] , which corresponds to a best frequency of 4 khz .
non - zero mass introduces an additional time constant , requiring examining a larger number of cases .
in addition , the mass may depends on the frequency because it may include fluid mass @xcite , and hair cell mass may not constitute a major part of the mass _ in vivo_. before evaluating energy , it is useful to examine the membrane capacitance , which is the attenuating factor of the receptor potential .
an external elastic load @xmath15 affects the membrane capacitance through @xmath116 ( see eqs . [
eq : capk ] and [ eq : cm@m=0 ] ) . since @xmath117 for @xmath118 ,
the maximum value of this factor is @xmath119 at @xmath120 . for our set of parameter values ,
the maximum value of this factor is 0.11 , indicating that this factor is rather small even though it is not negligible at low frequencies ( @xmath121 ) . for @xmath122 , nonlinear capacitance diminishes and so does the effect of an elastic load on the membrane capacitance . in the range where @xmath123 is comparable to the viscoelastic characteristic frequency @xmath68
, however , the effect of the elastic load @xmath15 appears mainly through the characteristic viscoelastic frequency @xmath124 . for a given frequency @xmath123 ,
an increase in the elastic load @xmath15 increases @xmath68 through @xmath23 and thus increases the capacitance @xmath77 . in the following , energy output from ohc is examined assuming that the hair bundle is stimulated at a level at which hair bundle conductance undergoes changes with an amplitude 10% of the resting value , where the mechano - channels are assumed to be 30% open .
thus , the change is 3% of its maximal value .
parameter values are given in table [ table : paramvals ] .
this amplitude allows linearized approximation , eqs .
[ eq : oscil - distr ] , [ eq : oscil - motion ] , and [ eq : oscil - voltage ] . in this regime ,
energy output increases with the second power of the receptor potential .
the system has two characteristic frequencies , the frequency @xmath125 of viscoelastic roll - off and the roll - off frequency @xmath126 of the rc circuit . here
the effect of the elastic load @xmath15 is examined .
the stiffness @xmath14 of the cell has been determined by experiments ( table [ table : paramvals ] ) .
two characteristic frequencies of the system leads to two cases , which are examined here : @xmath127 and @xmath128 .
it should be noticed , while @xmath129 is independent of @xmath15 , @xmath125 goes up as @xmath15 increases .
the energy @xmath130 conveyed to the external elastic load @xmath15 per half cycle can be obtained by evaluating @xmath131 and the work @xmath132 against viscous drag per half cycle is @xmath133 . here examine a case
in which the drag coefficient @xmath31 is extremely small and viscous loss is negligible .
let @xmath134 kg / s .
this value satisfies the condition @xmath127 except for diminishing @xmath15 .
this condition is satisfied even with increasing external elastic load @xmath15 to @xmath135 . at a given frequency ,
an increase in the amplitude of the receptor potential with increasing external elastic load @xmath15 is too small to notice in the plot ( fig .
[ fig : small_eta]a ) . with a given elastic load ,
the amplitude of the receptor potential monotonicaly decreases with increasing frequency ( fig .
[ fig : small_eta]a ) .
* a * * b * + kg / s.,title="fig:",width=226 ] kg / s.,title="fig:",width=226 ] the work against the elastic load is evaluated for a half cycle . for a given frequency ,
elastic energy output per half cycle has a maximum with respect to the external elastic load @xmath15 at the load ratio @xmath136 ( fig .
[ fig : small_eta]b ) .
the optimal ratio is not significantly affected by the frequency .
work output monotonically decreases with the frequency ( fig .
[ fig : small_eta]b ) .
the plot also shows that decrease of energy with frequency is less steep at larger load .
asymmetry increases with increasing frequency , while the peak ratio @xmath137 remains virtually unchanged
. if the viscous drag is larger , the receptor potential decreases with increasing the stiffness ratio @xmath137 for a given frequency ( fig .
[ fig : large_eta]a ) .
this result may appear counterintuitive because it is the opposite an increase , be it rather small , under the condition of small drag .
this reversal is due to a change in the time constant .
the membrane capacitance increases with increasing elastic load , owing to decreasing @xmath139 in eq .
[ eq : cm@m=0 ] .
an increase in the membrane capacitance increases rc attenuation , resulting in a reduction of the receptor potential .
the frequency dependence of the receptor potential is monotonic .
however , smallest elastic load makes the slope steeper at higher frequencies ( fig .
[ fig : large_eta]a ) . *
a * * b * + + * c * * d * + + the value of @xmath140kg / s for the drag coefficient used in the plots is chosen to examine a condition , which nearly maximizes the work against viscous drag , as we will see later .
for each given frequency , the work against the elastic load has a maximum with respect to the stiffness ratio ( fig .
[ fig : large_eta]b ) .
the load ratio @xmath137 that maximizes the work is close to two for low frequencies , unlike with lower viscous drag .
in addition , for higher frequencies the ratio that maximizes energy output increases significantly ( fig . [
fig : large_eta]b ) .
the maximal value at the frequency @xmath141(=4khz ) is about 2 zj ( = @xmath142 j ) at @xmath143 .
the work against viscous drag per half cycle decreases with increasing frequency ( fig .
[ fig : large_eta]c ) .
it also decreases monotonically with increasing stiffness ratio @xmath137 unlike the work against the elastic load . for this reason ,
the comparison of magnitude with elastic work heavily depends on the ratio @xmath137 .
the work against viscous drag is much larger than the work against elastic load if the elastic load is small . at @xmath143 ,
the value at the frequency @xmath141 is 6 zj , still larger than than the value 2 zj for the elastic work .
power output can be obtained by multiplying the work against drag per half cycle by twice the frequency @xmath144 .
it is less frequency dependent , even though it still decreases with increasing frequency ( fig .
[ fig : large_eta]d ) .
here we examine the effect of the viscous load for fixed values of the elastic load .
two values of the external load would be of interest : one of them is the case in which the stiffness of the external elastic load is similar to the internal stiffness of the cell .
the other is the case without an external elastic load .
the former condition is presumably close to the physiological condition and the output in the form of elastic energy would be appreciable .
the latter case is also of interest because it provides the maximal work against the viscous drag .
here we use @xmath145 kg / s as the unit of drag coefficient .
the receptor potential significantly increases with increasing viscous drag ( fig .
[ fig : drag_effect]a ) .
a higher drag coefficient leads to less steep decline of the membrane potential with increasing frequency in the middle range of the plot ( fig .
[ fig : drag_effect]a ) .
the work against the elastic load per half cycle decreases with increasing drag as well as frequency ( fig .
[ fig : drag_effect]b ) , as intuitively expected . at frequency @xmath141
, it is up to @xmath1473.4 zj for low drag . at @xmath148 ,
the value is about 1.8 zj .
* a * * b * + * c * * d * + the work against the viscous drag per half cycle monotonically decreases with the frequency .
however , it has a maximum with respect to viscosity at a given frequency ( fig .
[ fig : drag_effect]c ) . at @xmath141 ,
the maximal value is about 6 zj at @xmath148 .
power output due to the work against viscous drag ( obtained by multiplying the dissipative energy output per half cycle by @xmath144 , twice the frequency ) has a less steep frequency dependence ( fig .
[ fig : drag_effect]d ) .
the maximal output at the frequency @xmath141 is 0.05 fw . at a given frequency ,
work against viscous drag is maximal at @xmath46 , i.e. in the absence of an external elastic load ( fig .
[ fig : large_eta ] ) . for this reason it is interesting to examine the system without external elastic load to evaluate the limit even though such a condition may not be physiological .
* a * + + * b * * c * + the receptor potential increases with increasing viscous load and decreases with increasing frequency ( fig .
[ fig : drag_only]a ) . at a given frequency ,
the work against the viscous drag per half cycle has a maximum ( fig .
[ fig : drag_only]b ) . the peak position shifts to at smaller drag with increasing frequency .
the maximal work per half cycle is about 12 zj for @xmath141(=4khz ) .
the power output shows dependence on the drag and frequency similar to the work per half cycle does , even though the frequency dependence is less steep ( fig . [
fig : drag_only]c ) .
the maximal power output is about 0.1 fw .
here we discuss energy balance in an ohc that operates at 4 khz in the absence of mechanical resonance by examining the elastic load and viscous drag , to which the cell is likely subjected .
that is followed by discussion on cells that operates at higher frequencies . of the internal mechanical load
, there is no need to discuss the stiffness of the ohc because it has been determined experimentally .
some discussion would be needed regarding the internal drag of an ohc . for that purpose ,
the cell body is approximated by a cylinder .
the velocity of the fluid is null in the middle of the cell and linearly increases towards the two ends , exactly the same as the plasma membrane .
for this reason , poiseuille s law , for example , does not apply and viscous drag must be very small
. the magnitude of the drag on the outer surface could be estimated by stokes law for a sphere , @xmath149 , where @xmath150 is the radius .
the resulting drag coefficient is @xmath151 kg / s , assuming @xmath152 m and @xmath153 kg/(m@xmath154s ) of water for the viscosity @xmath102 . since this value is smaller than the value @xmath155 that maximizes dissipative energy output , the assumed drag coefficient @xmath31 in fig .
[ fig : large_eta ] and @xmath156 in figs .
[ fig : drag_effect ] and [ fig : drag_only ] , are dominated by the external viscous load and are not intrinsic to the cell .
the numerical examination shows that the receptor potential is affected by an increase of elastic load and that of drag quite differently . while an increase in drag always increase the receptor potential ( fig .
[ fig : drag_effect]a ) , an increase in the elastic load increases the receptor potential only slightly if drag is small ( fig .
[ fig : small_eta]a ) but it decreases the receptor potential if drag is larger ( fig .
[ fig : large_eta]a ) .
these observations suggest that drag is more effective than elastic load in enhancing the receptor potential that powers the cell s motility . the functional significance of ohcs as the cochlear amplifier depends on the balance between energy input and output .
energy output from the cell body has been evaluated in earlier sections .
energy input required for stimulating the hair bundle to generate the receptor potential consists of two components .
one is elastic and the other is dissipative .
the elastic component is recovered in the next cycle of stimulation during sustained oscillation .
the work against viscous drag can not be recovered .
therefore the balance of this energy has been considered critical for the sensitivity and sharp tuning of the ear @xcite .
the present model allows an evaluation of ohc s output , which compensates for the dissipated energy for hair bundle stimulation .
examination of force or energy balance in the cochlea , however , can not be made without a certain set of assumptions including the mode of motion in the cochlea @xcite . in the following , energy balance of a single ohc is examined . for this comparison , output values obtained under the condition @xmath157 and @xmath158
are used so that the two kinds of outputs can be realized at the same time .
the hair bundle drag , which gives rise to the dissipative energy required for input has been evaluated for bull frog saccular hair bundles @xcite .
it consists of external friction between the hair bundle and the bulk fluid and internal friction due to relative motion between the stereocilia .
the former is dominant when a hair bundle is stimulated by force applied to the kinocilium : the drag coefficient of a whole hair bundle is about 1/5 of a single stereocilia @xcite .
if we consider a sphere with a diameter of 8.5 @xmath102 m , which is the hair bundle height , stokes law gives the drag coefficient of 80 nns / m , about the same as the drag coefficient calculated for the bundle in the frequency range higher than 2 khz , where stereocilia moves in synchrony in response to force applied to the kinocilium @xcite .
this value can be even larger if the thickness of boundary layer is added to the diameter .
therefore , the internal drag must be smaller than the total drag at least by an order of magnitude . here
we set 1/10 of the total friction coefficient as an upper bound of the contribution of internal friction . in the physiological conditions for ohcs , unlike a frog hair bundle stimulated by holding kinocilium , the hair bundle is in the subtectorial space between the tectorial membrane and the reticular lamina and hair bundles are stimulated by the shear in the gap . under this condition , the external drag disappears because there is no relative motion between the bulk fluid and the hair bundle .
the energy loss in that system is due to the shear between the two places and the internal drag of the hair bundle is negligible .
this analysis is indeed consistent with earlier reports @xcite . assuming that the internal drag is less than 1/10 of the total , experimental data on frog saccular hair bundles @xcite can be used to estimate an upper bound of energy dissipation .
full gating of frog hair bundle stimulated at 100 @xmath102 m/s with amplitude @xmath147100 nm generates viscous force @xmath14720 pn @xcite .
this value is about twice as large as 80 nns / m for frequencies higher than 2 khz @xcite .
energy dissipation based on this observed force is @xmath159 j / half cycle for @xmath160 hz for full gating .
thus 3% gating corresponds to @xmath161 j / half cycle . if we extrapolate for higher frequencies , @xmath162 j / half cycle for frequency of 4 khz .
since the drag coefficient at 4 khz is less than that at 500 hz @xcite , this value is an overestimate .
a ten - fold reduction of this value leads to @xmath163 zj , which is given as an upper bound of energy dissipation due to internal friction of a _ frog _ hair bundle .
this value is smaller than @xmath164 zj ( @xmath165 j ) with elastic load @xmath146 ( figs .
[ fig : drag_effect ] c and [ fig : energy_balance ] ) , and @xmath91 zj without elastic load ( figs .
[ fig : drag_only]b and [ fig : energy_balance ] ) . *
a * * b * + the internal friction coefficient of a ohc hair bundle is likely much smaller than that of the frog hair bundle , given the difference in their geometries .
if we assume that internal friction is due to the shear between the stereocilia , the internal friction of an ohc is @xmath1471/4 of that of a frog hair bundle ( supplement ) .
thus an upper bound of energy loss is @xmath166 zj , less than 1/10 of the energy output of an ohc .
even if we assume that the ohc is isolated and subjected to an external drag equivalent to stokes drag for a sphere with the diameter of 4 @xmath102 m , the maximal height of a hair bundle that operates at 4 khz @xcite _ in vivo _ , the comparison indicates that energy output exceeds energy loss by hair bundle drag ( fig .
[ fig : energy_balance ] ) .
the observed stiffness of the hair bundle of ohcs is between 1 to 3 mn / m @xcite .
the hair bundle displacement required for 3 % change in the bundle conductance ( or resistance ) is @xmath1470.03 nm @xcite .
thus the energy required for changing the conductance by 3% is @xmath167 aj in the most efficient condition based on 1 mn / m bundle stiffness .
the elastic energy output obtained from the model is @xmath1472 zj at 4 khz ( fig .
[ fig : large_eta]b ) .
it is 3.6 zj even for small viscous drag ( fig .
[ fig : small_eta]b ) .
this indicates that output energy is a fraction of input energy .
outer hair cells have , therefore , a minor effect in the elastic energy involved in the oscillation of the system . since
elastic energy is conserved it is unnecessary for the output of an amplifier to match the input .
the analysis described above shows that energy output exceeds energy input for an ohc that operates at 4 khz even without mechanical resonance .
more basal cells that operate at higher frequencies are more labile and harder to obtain reliable experimental data
. however , the energy balance in more basal cells , which operate at higher frequencies , is of great interest because it is harder to achieve a favorable energy balance at a higher frequencies and therefore it is important for understanding the effectiveness of ohc electromotility as the basis of the cochlear amplifier @xcite . if we stimulate the same cell , for example , at 12 khz , a frequency higher by 3 times , the energy output is significantly less .
however , the ohcs that operate at higher frequencies are shorter , with a lower linear membrane capacitance @xmath43 and larger axial stiffness @xmath14 .
in addition , the hair bundle conductance is more sensitive to the strain @xcite . if @xmath43 is halved and @xmath14 is doubled , the energy balance is still favorable ( fig .
[ fig : energy_balance ] ) even without considering a higher sensitivity of the hair bundle transducer .
notice that these comparisons are made without considering mechanical resonance .
the reason for excluding resonance in the present examination is the sharp sensitivity of the membrane capacitance on the resonance frequency @xmath90 as well as the characteristic viscoelastic frequency @xmath125 ( fig .
[ fig : cnl ] ) .
since ohcs in the cochlea works close to mechanical resonance frequencies , it is possible that the membrane capacitance in the operating condition is much smaller , enabling more efficient use of electrical energy @xcite .
this issue , however , can not be addressed quantitatively without precise information , such as the values for @xmath90 , @xmath125 as well as the sensitivity of the hair bundle .
in this paper , a simple model for electromotility of ohcs is proposed to describe its behavior in a dynamic environment .
the model is consistent with the experimental data so far obtained from isolated ohcs .
the model also extends the expression for the membrane capacitance , incorporating the effects of frequency , drag , elastic load , and an associated mass . monitoring
the membrane capacitance could be the easiest means of testing the predictions .
the effect of mass will be described in more detail elsewhere .
this model enables description of the receptor potential and energy production by an ohc while its hair bundle is mechanically stimulated by sinusoidal waveform .
it was found that the receptor potential is more significantly affected by viscous drag than elastic load .
the model predicts that the output of elastic energy is less than the input at the hair bundle .
however , the output of dissipative energy is larger than the input . since
negative drag is a usual amplifying mechanism , these results are consistent with the biological role of the ohcs as an amplifier .
discussion with dr .
charles steele and yanli wang motivated to start the present work at stanford , partially supported by nih 5r01dc00791008 to stanford university . a part of the work was done during my stay at niels bohr institute , copenhagen and max planck institute for physics of complex systems , dresden .
it was also supported in part by the intramural research program of the nih , nidcd .
i acknowledge drs .
charles steele , anthony ricci , thomas heimburg , and frank jlicher for their interest and comments .
i also thank dr .
jong - hoon nam of syracuse university for discussion regarding hair bundle drag and richard chadwick , thomas friedman of nih , and three anonymous reviewers for useful comments .
41 [ 1]`#1 ` ashmore , j. f. , 1987 .
a fast motile response in guinea - pig outer hair cells : the molecular basis of the cochlear amplifier .
_ j. physiol .
_ 388:323347 .
santos - sacchi , j. , 1991 .
reversible inhibition of voltage - dependent outer hair cell motility and capacitance .
_ j. neurophysiol .
_ 11:30963110 .
dong , w. , and e. s. olson , 2013 .
detection of cochlear amplification and its activation .
_ biophys j _ 105:10671078 .
zheng , j. , w. shen , d. z .- z .
he , k. b. long , l. d. madison , and p. dallos , 2000 .
prestin is the motor protein of cochlear outer hair cells .
_ nature _ 405:149155 .
oliver , d. , d. z. he , n. klocker , j. ludwig , u. schulte , s. waldegger , j. p. ruppersberg , p. dallos , and b. fakler , 2001 .
intracellular anions as the voltage sensor of prestin , the outer hair cell motor protein .
_ science _ 292:23402343 .
song , l. , and j. santos - sacchi , 2013 .
disparities in voltage - sensor charge and electromotility imply slow chloride - driven state transitions in the solute carrier slc26a5 .
_ proc natl acad sci u s a _ 110:38833888 .
iwasa , k. h. , 1993 .
effect of stress on the membrane capacitance of the auditory outer hair cell .
_ biophys .
j. _ 65:492498 .
tolomeo , j. a. , and c. r. steele , 1995 .
orthotropic piezoelectric properties of the cochlear outer hair cell wall .
_ j. acoust .
97:30063011 .
iwasa , k. h. , 2001 .
a two - state piezoelectric model for outer hair cell motility .
_ biophys .
j. _ 81:24952506 .
hallworth , r. , 1995 .
passive compliance and active force generation in the guinea pig outer hair cell .
_ j. neurophysiol .
_ 74:23192328 .
iwasa , k. h. , and m. adachi , 1997 .
force generation in the outer hair cell of the cochlea .
_ biophys . j. _
73:546555 .
dallos , p. , r. hallworth , and b. n. evans , 1993 . theory of electrically driven shape changes of cochlear outer hair cells .
_ j. neurophysiol .
_ 70:299323 .
rabbitt , r. d. , s. clifford , k. d. breneman , b. farrell , and w. e. brownell , 2009 .
power efficiency of outer hair cell somatic electromotility .
_ plos comput biol _ 5:e1000444 .
ramamoorthy , s. , and a. l. nuttall , 2012 .
outer hair cell somatic electromotility in vivo and power transfer to the organ of corti .
_ biophys j _ 102:388398 .
frank , g. , w. hemmert , and a. w. gummer , 1999 . limiting dynamics of high - frequency electromechanical transduction of outer hair cells .
usa _ 96:44204425 .
housley , g. d. , and j. f. ashmore , 1992 .
ionic currents of outer hair cells isolated from the guinea - pig cochlea .
_ j. physiol .
_ 448:7398 .
johnson , s. l. , m. beurg , w. marcotti , and r. fettiplace , 2011 .
prestin - driven cochlear amplification is not limited by the outer hair cell membrane time constant .
_ neuron _ 70:11431154 .
ospeck , m. , and k. h. iwasa , 2012 .
how close should the outer hair cell rc roll - off frequency be to the characteristic frequency ?
[ correction : 103:846 - 847 ] .
_ biophys .
j _ 102:17671774 .
adachi , m. , and k. h. iwasa , 1999 .
electrically driven motor in the outer hair cell : effect of a mechanical constraint .
usa _ 96:72447249 .
ospeck , m. , x .- x .
dong , and k. h. iwasa , 2003 . limiting frequency of the cochlear amplifier based on electromotility of outer hair cells . _ biophys .
84:739749 .
mistrk , p. , c. mullaley , f. mammano , and j. ashmore , 2009 .
three - dimensional current flow in a large - scale model of the cochlea and the mechanism of amplification of sound .
_ j r soc interface _ 6:279291 .
maoilidigh , d. o. , and a. j. hudspeth , 2013 .
effects of cochlear loading on the motility of active outer hair cells . _ proc .
usa _ 110:54745479 .
cole , k. s. , 1968 .
membranes , ions , and impulses .
university of california press , berkely , ca .
sokabe , m. , f. sachs , and z. q. jing , 1991 .
quantitative video microscopy of patch clamped membranes stress , strain , capacitance , and stretch channel activation .
_ biophys .
59:722728 .
iwasa , k. h. , 2010 .
chapter 6 .
electromotility of outer hair cells . _ in _
p. a. fuchs , editor , the oxford handbook of auditory science volume 1 : the ear , oxford university press , oxford , uk , 179212 .
iwasa , k. h. , 1997 . current noise spectrum and capacitance due to the membrane motor of the outer hair cell :
_ biophys .
73:29652971 . gale , j. e. , and j. f. ashmore , 1997 .
an intrinsic frequency limit to the cochlear amplifier .
_ nature _ 389:6366 .
dong , x. , d. ehrenstein , and k. h. iwasa , 2000 .
fluctuation of motor charge in the lateral membrane of the cochlear outer hair cell .
_ biophys j _ 79:18761882 .
salt , a. n. , n. inamura , r. thalmann , and a. vora , 1989 .
calcium gradients in inner ear endolymph .
. santos - sacchi , j. , s. kakehata , t. kikuchi , y. katori , and t. takasaka , 1998 .
density of motility - related charge in the outer hair cell of the guinea pig is inversely related to best frequency .
_ neurosci lett .
_ 256:155158 .
freeman , d. m. , and t. f. weiss , 1988 .
the role of fluid inertia in mechanical stimulation of hair cells . _ hearing res . _ 35:201208 .
gold , t. , 1948 . hearing .
the physical basis of the action of the cochlea .
roy . soc . ,
_ 135:492498 .
kozlov , a. s. , j. baumgart , t. risler , c. p. c. versteegh , and a. j. hudspeth , 2011 .
forces between clustered stereocilia minimize friction in the ear on a subnanometre scale .
_ nature _ 474:376379 .
bormuth , v. , j. barral , j .- f .
joanny , f. jlicher , and p. martin , 2014 .
transduction channels gating can control friction on vibrating hair - cell bundles in the ear .
_ proc natl acad sci u s a _ 111:71857190 .
nam , j .- h . , and r. fettiplace , 2008 .
theoretical conditions for high - frequency hair bundle oscillations in auditory hair cells .
_ biophys j _ 95:49484962 .
prodanovic , s. , s. gracewski , and j .- h .
nam , 2015 .
power dissipation in the subtectorial space of the mammalian cochlea is modulated by inner hair cell stereocilia .
_ biophys j _ 108:479488 .
kennedy , h. j. , a. c. crawford , and r. fettiplace , 2005 .
force generation by mammalian hair bundles supports a role in cochlear amplification .
_ nature _ 433:880883
. mountain , d. c. , and a. e. hubbard , 1994 . a piezoelectric model of outer hair cell function .
_ j. acoust .
95:350354 .
mller , m. , 1991 .
frequency representation in the rat cochlea .
_ hear res _
51:247254 .
furness , d. n. , s. mahendrasingam , m. ohashi , r. fettiplace , and c. m. hackney , 2008 .
the dimensions and composition of stereociliary rootlets in mammalian cochlear hair cells : comparison between high- and low - frequency cells and evidence for a connection to the lateral membrane .
_ j neurosci _ 28:63426353 .
lim , d. j. , 1986 .
functional structure of the organ of corti : a review . _ hear res _ 22:117146 .
here internal drag of a hair bundle is estimated assuming it is due to the shear between stereocilia in a hair bundle . to provide an estimate ,
the shear between two stereocilia is approximated by the shear between two plates , assuming the separation is the same as the rootlet separation @xmath168 , and the widths of the plates is given by the diameter @xmath169 of the stereocilia , ignoring end effects .
this model overestimates the separation because the nearest distance between the stereocilia is less than @xmath168 , resulting in an underestimate of the drag . at the same time the planar model overestimates the drag because end effects are not considered .
if the angle @xmath170 of bending and the separation @xmath168 between the stereocilia are small , viscous drag @xmath171 between a pair of stereocilia can be expressed , @xmath172 where @xmath173 is the speed of the relative motion of the stereocilia , @xmath168 root separation , @xmath174 height of the gap , @xmath169 stereocila diameter , and @xmath175 tip height ( fig .
a1 ) . since @xmath176 and the tip velocity is expressed by @xmath177 , the drag coefficient @xmath178 with respect to tip velocity
does not depend on the separation @xmath168 .
if the bundle has @xmath179 sliding pairs , the drag coefficient @xmath180 of the bundle is , @xmath181 ( -2.5,0 ) ( 2.5,0 ) ; ( -0.6,0)(-0.2,2 ) ; at ( -0.1,2.3 ) @xmath174 ; ( -0.6,0)(-0.6,3 ) ; ( -0.3,1.2 ) arc ( 70:90:1 ) ; at ( -0.8,1 ) @xmath170 ; ( 0.6,0)(1.4,4 ) ; at ( 1.8,3.8 ) @xmath175 ; ( 0.5,4)(1.3,4 ) ; at ( 0.8,3.7 ) @xmath182 ; ( -0.2,2)(1.0,2 ) ; ( -0.2,1.95)(0.9,1.7 ) ; at ( 0.3,1.6 ) @xmath170 ; ( 0.5,2 ) arc ( 0:-28:0.5 ) ; at ( 0,-0.2 ) @xmath168 ; ( 0.2,-0.2)(0.6,-0.2 ) ; ( -0.6,-0.2)(-0.2,-0.2 ) ; frog saccular hair bundle is modeled by a hexagonally packed hexagonal array that consists of 62 stereocilia @xcite .
let a displacement applied through the kinocilium at a vertex toward the center of the bundle .
the number of sliding pairs in this direction is 53 .
the number of pairs away from this direction by 60 degree is 106 .
those pairs are subjected a half as large shear as those in the direction of stimulation .
thus the effective number @xmath179 of sliding pairs is 106 .
the mean height @xmath174 of these pairs is @xmath183 .
hence @xmath184 .
if we choose a value 0.4 @xmath102 m for the radius @xmath169 of stereocilia , we obtain @xmath185 ns / m , which is about 1/3 of the total drag coefficient @xcite , consistent with the significance of end effects . if we choose [email protected] @xmath102 m , @xmath187 ns / m , @xmath1861/10 of the total drag coefficient .
it is compatible with the calculation of kozlov et al .
@xcite in the cochlea of rats , 4 khz location is at 80% from the base @xcite , which is in the apical turn .
the ratio @xmath188 is typically 0.5 @xcite .
if we assume the total number of stereocilia is about 60 , similar to chinchilla hair bundles @xcite , the number of sliding pairs is 40 .
if we can use [email protected]@xmath102 m for the effective radius of stereocilia , we obtain @xmath189 ns / m .
|
electromotility of outer hair cells ( ohcs ) has been extensively studied with _ in vitro _
experiments because of its physiological significance to the cochlear amplifier , which provides the exquisite sensitivity and frequency selectivity of the mammalian ear . however , these studies have been performed largely under load - free conditions or with static load , while these cells function _ in vivo _ in a dynamic environment , receiving electrical energy to enhance mechanical oscillation in the inner ear .
this gap leaves uncertainties in addressing a key issue , how much mechanical energy an ohc provides .
the present report is an attempt of bridging the gap by introducing a simple one - dimensional model for electromotility of ohc in a dynamic environment .
this model incorporates a feedback loop involving the receptor potential and the mechanical load on ohc , and leads to an analytical expression for the membrane capacitance , which explicitly describes the dependence on the elastic load , viscous drag , and the mass .
the derived equation of motion was examined in a mass - less model system with realistic parameter values for ohc .
it was found that viscous drag is more effective than elastic load in enhancing the receptor potential that drives the cell .
for this reason , it is expected that ohcs are more effective in counteracting viscous drag than providing elastic energy to the system .
amplifier , mammalian ear , membrane capacitance
|
the laboratory realization of ultracold neutral atomic gases such as bose - einstein condensates @xcite and degenerate fermi gases @xcite quantum gases delivered remarkably versatile experimental systems that can realize physical effects with analogues throughout physics .
the coherence properties of bose - einstein condensates ( becs ) allow them to address concepts from optics and nonlinear optics : classical and quantum atom optics @xcite .
quantum gases have shed light on many effects predicted in the context of traditional condensed matter systems such as the bosonic superfluid to mott transition in optical lattices @xcite , and the bardeen - cooper - schrieffer crossover in degenerate fermi gases @xcite .
even phenomena commonplace in high energy physics can occur in ultracold settings , where higgs modes have been observed @xcite , unconventional `` color '' superfluidity @xcite is possible , and where confinement mechanisms @xcite and axion electrodynamics @xcite have been predicted .
atomic quantum gases are charge neutral , and therefore , they are not affected by external electromagnetic fields the way electrons are .
however , atom - light coupling allows for the creation of versatile gauge potentials that effectively emerge in the atoms dynamics , allowing experimental access to a panoply of new phenomena at the quantum level . using this technology ,
atoms can be subjected to static abelian gauge fields , offering a framework where synthetic electric and magnetic fields can be experimentally tuned with lasers ( see the first experimental works at nist ) .
these setups can also be extended to generate versatile non - abelian gauge potentials @xcite .
these static non - abelian gauge fields could be tailored so as to reproduce the effects of rashba - type spin - orbit couplings , but also , to mimic a variety of properties encountered in the context of high - energy physics .
the first experimental steps towards the realization of a two - dimensional spin - orbit - coupled atomic gas have been reported in refs .
@xcite , where the spin - orbit coupling acts along a single spatial dimension . mimicking magnetic and spin - orbit effects in cold atom laboratories
enables the assembly of quantum simulators of new kinds of exotic quantum matter @xcite .
indeed , cold atomic gases are ideally suited for quantum simulation , as numerous physical parameters governing the systems dynamics are experimentally tunable : particle density , confining potentials , effective dimensionality @xcite , and even the collisional properties @xcite can be easily controlled in the same laboratory . taken together ,
this greatly enlarges the range of systems that can realize richard feynman s vision @xcite for constructing physical quantum emulators of systems or situations that are computationally or analytically intractable .
gauge theories , with their associated gauge potentials , are central in our understanding of the interactions between elementary particles .
electromagnetism is the simplest example , where the scalar and vector potentials together describe the coupling between charged matter and electromagnetic fields . in the standard model , interactions are mediated by more complex gauge fields which often are of a non - abelian character . also , the idea of emergent gauge fields , where the low energy sector of a more complicated system is described by an effective gauge theory , is not new .
mead and truhlar @xcite and berry @xcite noted that the adiabatic motion of quantum particles with internal structure can be described in terms of an effective `` geometric '' gauge potential .
this property was first studied in molecular physics , where the jahn - teller effect revealed the geometric phases and corresponding vector potentials @xcite .
the adiabatic and born - oppenheimer approximations are closely linked to geometric gauge fields in atomic systems .
these geometric vector potentials appear when each atom s external motion is described separately from it s internal dynamics , yet the hamiltonian governing the internal dynamics parametrically depends on the atomic position ( for example , via the light - matter interaction ) . in this context , the possibility of emergent vector potentials was first noted by taichenachev et al @xcite , dum and olshanii @xcite , as well as by visser and nienhuis @xcite .
refs . proposed setups for systematically engineering vector potentials which provide a non - zero artificial ( synthetic ) magnetic field for quantum degenerate gases .
these synthetic magnetic fields were recently experimentally realized @xcite , whereas the effect of geometric scalar potentials in optical lattices was experimentally observed a decade earlier @xcite . when the local atomic internal states `` dressed '' by the laser fields have degeneracies , effective non - abelian gauge potentials can be formed @xcite , often manifesting as a spin - orbit coupling would in material systems .
these artificial spin - orbit couplings lead to the spin hall effect in atomic systems @xcite , as recently demonstrated experimentally @xcite .
artificial gauge fields are therefore a highly versatile tool for creating exotic condensed matter analogs in atomic gases @xcite . these approaches can be extended in a powerful way by adding optical lattice potentials @xcite , where the link to quantum simulation of condensed matter phenomena is particularly evident . here , the artificial magnetic field can be understood as resulting from a laser induced tunneling between the lattice sites @xcite , or lattice shaking @xcite .
numerous theoretical proposals for simulating condensed matter models and realizing strongly correlated systems have been put forth , and recently , artificial gauge potentials corresponding to staggered @xcite and uniform @xcite magnetic fluxes have been produced in optical lattices .
all of these schemes create _ static _ gauge fields , in the sense that they are described by additional terms in the atomic hamiltonian (
although the gauge fields can still have an externally imposed time dependence , leading to effective electric fields @xcite ) .
dynamical gauge fields ( which are described by their own hamiltonian and are not just imposed ) are important in many areas of physics , from particle physics where the gauge fields are the fundamental force carriers ( e.g. , ref . ) , to many - body condensed - matter physics where they appear in effective field theories @xcite . as such ,
a number of proposals exist for creating dynamic gauge fields with ultracold atoms @xcite , but to date their complexity has stymied experimental realization . by emulating a fully dynamical field theory , which includes gauge fields , it is certainly tempting to envisage a quantum simulator that can address open questions from the standard model @xcite . mapping out the complete qcd phase diagram is a formidable task .
it is also an np hard problem , and therefore highly intractable using classical computation . a special purpose quantum computer able to emulate the corresponding machinery from the standard model would significantly contribute to our understanding of the fundamental forces and processes in nature .
although the fundamental nature of being able to create gauge fields for charge neutral ultracold quantum gases , with clear links to particle physics and the forces of nature , is a compelling argument for pursuing such an endeavor , there are also other , more practical , motivations to create these gauge fields .
magnetic fields and spin - orbit coupling , in particular , appear to provide a route towards the preparation of topological states of matter @xcite , with some quite remarkable properties and promises for future applications @xcite .
ultracold gases subject to artificial gauge fields provide an alternative route for reaching such exotic states of matter , with some added benefits from their unique probes and unprecedented flexibility in controlling many experimental parameters in these systems @xcite .
topology is the branch of mathematics that deals with properties of geometric objects that do not change under smooth deformations @xcite .
the great interest in topological states of matter relies on the fact that such states are robust against external perturbations ( e.g. finite temperature , noise , or in general experimental imperfections ) .
there are many intriguing phenomena associated with topological matter .
the most striking is the existence of metallic edge states in a material that is insulating in the bulk @xcite . in the integer quantum hall effect @xcite
, these edge modes carry precisely one quantum of conductance , which leads to the quantization of the hall conductivity . due to the bulk - edge correspondence @xcite ,
much of the properties and information of topological systems can be extracted from the edge states .
depending on the details of the particular physical setup ( e.g. the lattice geometry , the interactions , the number of magnetic flux per particle , the presence of spin - orbit coupling ) , these edge modes may turn out to have very exotic properties @xcite .
for instance , some topological edge states do not satisfy the traditional statistics of fermions or bosons .
these enigmatic _ anyons _
@xcite , have not been identified in nature , but are expected to live as excitations in quantum hall liquids @xcite and topological superfluids @xcite .
the unusual braiding properties associated with the so - called non - abelian anyons @xcite , together with their robustness against imperfections and noise , makes topological matter a promising candidate for building an error - free quantum computer @xcite , which has all the potential to revolutionize modern technology . generating and probing anyonic excitations in cold - atom setups , using artificial gauge fields , is certainly one of the most important goals in the field .
one of the challenges and open questions would be how to develop schemes which allow for the creation and stability of such states , e.g. against various decay and heating processes that are generally present in experiments , and by doing so manipulate anyonic excitations in a well - controlled environment . in this review ,
we summarize different techniques for creating artificial gauge potentials in cold atom systems ( both implemented and proposed ) , pedagogically describing the main physical mechanisms behind each .
we then illustrate the gauge potential s role in a number of applications and highlight the connections between these engineered gauge potentials and other branches of physics . since the publication of a shorter review of modern physics colloquium on artificial gauge fields for ultracold atoms @xcite , there has been a great deal of theoretical and experimental activities in the area , which are reflected in the present review .
the manuscript is organized as follows . in sect .
[ sect : rotation ] , we review the initial technique where rotating gases experienced effective uniform magnetic fields . in sect .
[ sect : gaugefields ] we present a general framework for light - induced gauge potentials . in sect .
[ sect : lightmatter ] we study the basic interaction between laser fields and atoms , providing guidelines for designing realistic artificial gauge potentials in alkali systems .
we then show in sect .
[ sect : schemes ] how light - matter interactions can optically induce abelian gauge potentials for ultracold atoms .
[ sect : soc ] considers the schemes for generating non - abelian gauge potentials and spin - orbit coupling for ultra cold atoms . in sect .
[ sect : interactions ] , we study how collisions are altered by the light - matter coupling , illustrating their role first for ground - state bec s , and for the pairing mechanisms in interacting fermi gases .
[ sect : lattices ] adds optical lattices , and describe how synthetic magnetic fluxes and spin - orbit couplings can be engineered in a lattice environment .
then , in sect . [ sect : simulation ] , we discuss several quantum simulators based on the gauge potential concept , and comment on experimental techniques for detecting the gauge field s effects . in sect . [ sect : dynamical ] , we then briefly discuss the concept of interacting gauge theories with their unconventional current nonlinearities , and also recent proposals for emulating fully dynamical gauge fields in optical lattices .
we conclude by a summary of the current techniques available for creating artificial gauge potentials and indicate potential applications of these ideas .
artificial gauge fields result from spatially and/or temporally inhomogeneous hamiltonians . in this section we consider two situations where one can eliminate the time - dependence of a trapping potential by going to a non - inertial frame of reference .
both conceptually , and experimentally , the most simple example of an artificial gauge field appears in a spatially rotating frame @xcite .
this exploits the familiar equivalence between the coriolis force in a rotating system and the lorentz force acting on a charged particle in a uniform magnetic field . around the @xmath0 axis .
, width=384 ] let us consider in detail the quantum dynamics of an electrically neutral atom in a trap rotating with an angular frequency @xmath1 around the @xmath0 axis , as depicted in fig .
[ fig : rotating_cloud ] . recalling that a spatial rotation by an angle @xmath2 around the rotation vector @xmath3 is described by the transformation @xmath4 involing an orbital angular momentum operator @xmath5 ,
the atomic hamiltonian is @xmath6 where @xmath7 , and @xmath8 are respectively the position and momentum vectors in the inertial frame of reference , and @xmath9 is the position vector in the rotating frame , with @xmath10 and @xmath11 .
the usual canonical commutation relations between the cartesian components of the position and momentum vectors @xmath12 = i \hbar \delta_{l , j}$ ] allows us to represent the momentum vector @xmath13 for problems explicitly expressed in the coordinate representation .
thus any time dependence of the trapping potential @xmath14 emerges exclusively through the temporal dependence of the rotating vector @xmath15 . because the magnitude of the momentum is unchanged by rotations , @xmath16 , the hamiltonian @xmath17 is related to its time - independent counterpart via the unitary transformation @xmath18 @xmath19r_{z}^{\dagger}(t)\,.\label{eq : h - initial-1}\ ] ] the time - dependent equation ( tdse ) @xmath20 governs the system s dynamics . inserting the transformed wavefunction @xmath21 into the tdse yields a rotating frame tdse @xmath22 with the time - independent hamiltonian @xmath23 where the term @xmath24 results from the temporal dependence of @xmath18 . using @xmath25
, the hamiltonian @xmath26 can be represented as @xmath27 the emerging symmetric - gauge vector potential @xmath28 describes the cyclotron motion of the atom in the @xmath29 plane .
an additional anti - trapping ( centrifugal ) potential @xmath30 repels the atom away from the rotation axis @xmath0 .
the hamiltonian ( [ eq : rothamiltonian ] ) has the same form as that for a particle with a unit charge moving in a uniform magnetic field @xcite @xmath31 .
the above analysis does not involve any assumption concerning a specific form of the trapping potential @xmath32 .
thus the creation of an artificial magnetic flux via rotation can be applied not only to the usual trapping potentials @xcite and also to other structures , such as rotating optical lattices @xcite or superfluid atom circuits with a rotating weak link @xcite .
the centrifugal potential can compensate for harmonic trapping potentials @xmath33 when the rotation frequency approaches the trap frequencies @xmath34 and @xmath35 . in this limit
, the problem reduces to that of an unconfined free particle in the constant magnetic field @xmath36 .
interestingly , the associated cyclotron frequency @xmath37 is twice the rotation frequency .
having seen how the single - particle hamiltonian transforms into the rotating frame , we now turn to the question of interactions . for now , consider an arbitrary pairwise interaction @xmath38 which is a function only of the separation between particles . under the transformation to the rotating frame
the potential @xmath39 is unchanged .
this follows from the identities @xmath40 , and @xmath41 which simply state that relative geometry is not changed by rotations .
the potential @xmath38 remains the same , but the two body problem does change due to the emerging vector and centrifugal potentials .
ultracold atoms have been rotated to large angular frequency with spectacular success ( see fig . [
fig_rotation]a ) by several groups , lead by pioneering experiments at jila @xcite , ens @xcite , and mit @xcite in conventional harmonic traps .
these experiments addressed several important technical questions : ( 1 ) how to start an ultracold atomic gas rotating ; ( 2 ) how to keep it rotating ; and ( 3 ) how to detect rotation .
questions ( 1 ) and ( 2 ) are related . as is evident in eq .
[ eq : rothamiltonian ] , the desired rotating frame hamiltonian should have no remnant time - dependance .
the trapping potential must be asymmetric in the @xmath42-@xmath43 plane in order to induce rotation into an initially non - rotating system .
generally this is achieved either by rotating an initially deformed harmonic trap , or by stirring with focused `` tweezer '' lasers .
( the initial jila experiment used an ingenious technique involving transitions between internal atomic states @xcite , but adopted the deformed trap method to great success as described below . ) for experiments featuring the most rapid rotation
equivalently the largest effective magnetic fields the trap potential is generally returned to near - perfect axial symmetry .
any non - rotating component of the potential ( in the lab frame ) transforms to an unwanted rotating contribution ( in the rotating frame ) that can frictionally heat , or slow rapidly rotating clouds ( the edges of which can easily exceed the critical velocity for superfluid flow in the lab frame ) . with sufficient effort ,
it is possible to achieve nearly perfect axial symmetry , and in the later jila experiments , there was no discernible decrease in angular frequency for the lifetime of their atomic ensembles @xcite . , showing a well ordered vortex lattice .
( b ) rotating systems entering the lowest landau level ( lll ) .
for slowly rotating systems , vortices occupy only a small fraction of the system ( black dashed line ) , whereas in the lll the vortex density is constant ( red - dashed line ) . as the chemical potential falls below the effective cyclotron frequency , the bec can be well described by a wavefunction projected into the lll .
figures are included by permission of eric cornell , and first appeared in [ p. engels , _ et al .
_ prl * 90 * , 170405 ( 2003 ) , and v. schweikhard , _ et al .
_ prl * 92 * , 040404 ( 2004 ) ] , refs .
@xcite and @xcite .
, width=432 ] for superfluid bose gases , there are three primary techniques for detecting rotation .
firstly , the presence of vortices in a bec directly indicates the existence of an effective magnetic field , or rotation .
these vortices result in a change in the atomic density over a small region , with a length scale set by the condensate s healing length @xmath44 , where @xmath45 is the chemical potential .
typically @xmath46 is between @xmath47 and @xmath48 : below the usual imaging resolution . as a result
, vortices have not been measured directly in the trap , and are revealed by removing the confining potential and allowing the bec and vortices to expand before imaging .
secondly , the evolution of collective modes directly reveals rotation @xcite .
lastly , rotation can be inferred by observing the weakening of the trap from the centripetal anti - confinement @xcite . in spite of the success creating large vortex lattices ,
rotating systems have not entered the strongly correlated regime where the filling factor @xmath49 .
this results from a technical limitation : at some point it becomes difficult to increase the angular momentum per particle and remain in equilibrium .
the most rapidly rotating bec just barely entered the lowest landau level ( lll ) regime , where the vast majority of the particles reside in the lll ( unlike electron systems , this can occur at quite large @xmath50 ) .
a bec will enter the lll when the occupation parameter @xmath51 becomes large ; the onset of this crossover was observed by the jila group ( see fig .
[ fig_rotation]b ) using a series of ingenious evaporation tricks to increase the mean angular momentum per particle .
a secondary technical challenge for studying strongly correlated systems is to confine the atoms into 2d planes .
the confinement along @xmath52 should be large compared to the radial trapping .
this is generally achieved using one - dimensional optical lattices , but technically it is difficult to create lattices with the required axial symmetry for rapid rotation experiments .
most experiments with rotating systems worked with large atomic gases in a single trapping potential , however , a recent experiment created an optical lattice with @xmath53 atoms per site , where each lattice site was separately rotated @xcite ( as opposed to rotating the optical lattice in its entirety @xcite ) . in this regime ,
very rapid rotation is possible owing to the very strong confinement in individual lattice sites , and the authors argue they have entered the few - atom strongly correlated regime . besides rotation ,
shaking optical lattices is another widely used experimental technique @xcite , which can trigger non - trivial topological effects and gauge structures in a rather direct manner @xcite .
driven - induced gauge fields , using modulated cold - atom systems , is deeply related to the concept of floquet topological states " @xcite and strain - induced magnetic fields @xcite , which are currently explored in solid - state laboratories .
in particular , a shaken atomic system can be tailored so as to reproduce the dynamics of electronic systems subjected to circularly polarized light [ see below and ref .
@xcite ] ; such an observation is instructive since light - driven electronic systems were predicted to produce floquet topological states in materials such as graphene @xcite . to give a general description of the phenomenon ,
let us consider atoms in an arbitrary trapping potential ( e.g. a periodic optical lattice or a harmonic potential ) , which is affected by shaking .
the hamiltonian @xmath17 then has the form of eq .
( [ eq : h - initial ] ) , in which the trapping potential @xmath14 depends on the position vector @xmath54 defined by @xmath55\,.\label{eq : shifted_frame}\ ] ] namely , @xmath56 contains a time - dependent shift @xmath57 with respect to the inertial frame vector @xmath58 , which depends on the shaking protocol @xcite .
in contrast to the case of rotation , the orientation of the coordinate system is not altered .
only the origin @xmath57 changes in time .
if the motion was chosen to be linear in time , @xmath59 , the transformation would imply a simple transition to another inertial frame . here , we are interested in another scenario , where the trapping potential is modulated in a periodic manner , @xmath60 , with @xmath61 being the period of the shaking . for instance , considering a single harmonic , the linear and circular driving are respectively described by @xmath62 where @xmath63 denotes the driving amplitude .
the transformation to the non - inertial frame does not alter the momentum , @xmath64 , so that the transformed hamiltonian reads @xmath65 where the vector potential @xmath66 represents the momentum of a particle moving with the frame velocity @xmath67 .
in the last relation , we have neglected the uniform term @xmath68 , which can be eliminated by including a position - independent phase factor to the atomic wave - function .
the time - dependent vector potential yields a spatially uniform force @xmath69 acting on atoms due to acceleration of the non - inertial frame .
in contrast to the transformation to a rotating frame considered in the previous sect .
[ sect : rotation ] , the vector potential @xmath70 does not have any spatial dependence , but instead , it is time - dependent @xcite .
note also that the vector potential @xmath71 associated with a linear ( resp .
circular ) harmonic shaking given by eq .
( [ eq : linear - circular ] ) is equivalent to the one emerging for a charged particle in a linear ( resp .
circular ) polarized electric field @xcite . as mentioned above
, this analogy with electronic systems emphasizes the possibility to generate driven - induced ( floquet ) topological phases @xcite with shaken optical lattices @xcite .
the effective gauge structures and topological phases emanating from lattice modulations will be further addressed in section [ sec : shaking lattice ] .
geometric gauge potentials arise throughout physics @xcite .
one place they can emerge in cold atom systems is when the atomic center of mass motion is coupled to its internal ( `` spin '' ) degrees of freedom @xcite . to understand these gauge fields , we begin with the full ( and in general time - dependent ) atomic hamiltonian @xmath72 where @xmath73 is the state - independent trapping potential , @xmath58 and @xmath74 are the atomic center of mass coordinate and momentum , and @xmath75 is the unit operator acting on the internal atomic degrees of freedom .
the operator @xmath76 includes the hamiltonian for the atomic internal motion , as well as the atom - light coupling term . as a result
, @xmath77 explicitly depends on @xmath58 , and in general time .
hats over operators ( like @xmath77 and @xmath75 ) signify that they act on the internal atomic degrees of freedom ; center of mass operators such as @xmath58 and @xmath74 , will be hatless . in this section , we adopt the coordinate representation for the center of mass motion , so @xmath58 simply denotes the atomic position , and @xmath78 is the associated momentum operator . in subsequent sections , we will turn to the momentum representation to view the gauge potentials from another angle @xcite .
the operator @xmath79 can be cast in terms of the atomic bare internal states @xmath80 @xmath81 so the position and time dependence of @xmath79 comes exclusively from the matrix elements @xmath82 . here
@xmath83 is the number of atomic internal states involved . of the atomic dressed states @xmath84 .
, width=384 ] the diagonalisation of the operator @xmath79 provides a set of eigenstates @xmath84 ( @xmath85 ) , known as the atomic dressed states , with eigenenergies @xmath86 that depend on the atomic position @xmath58 , as depicted in fig .
[ fig : fig_adiabatic ] .
any atomic state - vector can be expanded in this position - dependent basis @xmath87 where @xmath88 is a wave - function for the centre of mass motion of the atom in the @xmath89-th internal dressed state .
the atomic bare and dressed basis states are connected via a position - dependent unitary transformation @xmath90 which makes the operator @xmath79 diagonal : @xmath91 replacing the original state vector @xmath92 , eq . ( [ eq : state - vector - full ] ) , by the transformed one @xmath93 the hamiltonian governing the transformed tdse @xmath94 reads @xmath95 , giving @xmath96 the vector operator @xmath97 emerges due to the spatial dependence of the atomic dressed states , whereas the additional scalar operator @xmath98 arises because of their temporal dependence . for the sake of simplicity , we have omitted the unit operator @xmath75 multiplying the momentum operator @xmath74 and the state - independent potential @xmath99 in the transformed hamiltonian @xmath100 , eq .
( [ eq : h - transformed - r ] ) .
if a subset containing @xmath101 dressed states is well separated in energy from the remaining ones , it is appropriate to make an adiabatic ( born - oppenheimer ) approximation by projecting the system s dynamics onto the truncated space of the internal states , using the operator @xmath102 .
the hamiltonian @xmath103 describing such a reduced dynamics reads @xmath104 where @xmath105 , @xmath106 and @xmath107 are the projections onto the reduced subspace of the corresponding operators featured in the full hamiltonian @xmath100 , eq .
( [ eq : h - transformed - r ] ) . here
we have omitted the projector @xmath108 multiplying @xmath74 and the state - independent potential @xmath99 . an extra operator @xmath109 with the matrix elements @xmath110 results from projecting @xmath111 onto the selected subspace of the internal dressed states .
the potential @xmath112 can be interpreted as a kinetic energy of the atomic micro - trembling due to off - resonance non - adiabatic transitions to the omitted dressed states with @xmath113 @xcite . in this way , the geometric vector potential @xmath106 and scalar potential @xmath112 emerge from the atomic dressed states spatial dependence .
the potential @xmath107 on the other hand , stems from their time dependence .
the latter @xmath107 describes the population transfer between the atomic levels due to the temporal dependence of the external fields @xcite .
when the truncated space includes a single dressed state ( @xmath114 ) well separated from the others , as is the case in fig .
[ fig : fig_adiabatic ] , the vector and scalar potentials reduce to the ordinary commuting vector and scalar fields , @xmath115 in this case , the resulting artificial electric and magnetic fields are the standard @xmath116 and @xmath117 .
however , for @xmath118 the more general relations discussed in the following are required .
the vector and scalar potentials @xmath106 and @xmath112 featured in the projected hamiltonian @xmath119 , eq .
( [ eq : h - matrix - reduced ] ) , can be related to the operator generalizations of conventional magnetic and electric fields .
the situation is more complicated than for classical electromagnetism , because the scalar potential and the cartesian components of the vector potential are now operators which do not necessarily commute . to elucidate the problem
, we turn to the heisenberg equations of motion governing the projected dynamics .
the velocity operator defined via the heisenberg equation is @xmath120=\frac{1}{m}(\mathbf{p}-\hat{{\mathbfcal a}^{\left(q\right)}})\,.\label{eq : v - explicit}\ ] ] the acceleration is defined via the heisenberg equation for the velocity operator @xmath121/\hbar$ ] .
since the only explicit time dependence of @xmath122 resides in @xmath106 , then @xmath123 , giving @xmath124-\frac{i}{\hbar}[\hat{\mathbf{v}},\hat{v}_{\mathrm{tot}}^{\left(q\right)}].\label{eq : v - derivative}\ ] ] this equivalently gives the cartesian components of the acceleration @xmath125+[\hat{v}_{k},\hat{v}_{l}]\hat{v}_{l})-\frac{i}{\hbar}[\hat{v}_{k},\hat{v}_{\mathrm{tot}}^{\left(q\right)}],\label{eq : v - derivative - components}\ ] ] where a summation over the repeated cartesian indices is again implied .
expressing the velocity commutators @xmath126=i\hbar\hat{{\mathcal{f}}}_{kl}$ ] in terms of the antisymmetric tensor @xmath127\,,\label{eq : v - k -- v - l - commutators}\ ] ] one arrives at the equation of motion @xmath128 the vector operator @xmath129 with components @xmath130 is the artificial magnetic field ( berry curvature ) providing the lorentz force . using eq .
( [ eq : a - nm ] ) for @xmath131 together with the completeness relation , the matrix elements of the curvature can be represented as a sum over the eliminated states @xmath132 hence , the berry curvature is non - zero only for the reduced atomic dynamics @xmath133 when some of the atomic states are eliminated . the same applies to the geometric scalar potential @xmath134 given by eq.([eq : w - nm ] ) .
additionally the atom is affected by an effective electric field @xmath135\,\label{eq : e - prime}\ ] ] that contains a commutator @xmath136 $ ] together with the usual gradient @xmath137 and induced @xmath138 contributions .
when all the cartesian components of @xmath106 commute with each other , the vector potential is said to be abelian ( this is always the case when @xmath114 ) .
otherwise , when some components do not commute , @xmath139 \ne 0 $ ] for some @xmath140 and @xmath141 , we will state that the system exhibits a non - abelian gauge potential @xmath106 . \ne 0 $ ] , or by the non - commutativity of successive loop operations , see e.g. @xcite .
however , in this review , a vector potential satisfying the criterium @xmath142 \ne 0 $ ] will generally be referred to as a non - abelian gauge potential " .
this issue will be further addressed in the section [ non - abelian_wilson ] . ] for abelian vector potentials , we recover @xmath143 and @xmath144 . an abelian geometric vector potential @xmath106 has already been engineered for ultracold atoms in the cases @xmath114 @xcite and @xmath145 @xcite .
for a single adiabatic state ( @xmath114 ) the spatial dependence of @xmath106 yield an artificial magnetic field @xcite and its temporal dependence generate an effective electric field @xcite .
we shall return to the latter issue in the sec .
[ sec : electric field ] on experimentally creating the artificial electric field .
now we turn to an interplay between geometric gauge potentials and rotation , motivated by a possibility for simultaneously generating an abelian vector potential ( via rotation ) and a light - induced non - abelian geometric vector potential ( equivalent to spin - orbit coupling ) .
here we outline a general framework for adding an abelian gauge potential via rotation to a non - abelian geometric gauge potential @xcite .
this scenario has several associated subtle issues : ( 1 ) in proper rotation experiments , the whole hamiltonian should be static in the rotating frame . in this context , both the center of mass and coupling terms should be rotating , implying that the whole laser system should rotate .
( 2 ) generally , techniques for creating artificial gauge fields involve raman - resonant transitions , and in such cases the `` orientation '' of zeeman terms results from a relative phase ( for the @xmath42 and @xmath43 components ) and detuning from resonance ( for the @xmath52 component ) , not geometry .
the following discussion focuses on the broad picture and will not treat these points .
let us suppose that the system is described by the hamiltonian ( [ eq : h - full ] ) , in which both the center of mass and internal contributions rotate with the frequency @xmath1 around @xmath0 .
both the state - independent and state - dependent potentials are assumed to be time - independent in the rotating frame of reference , so that in the lab frame the time - dependence of @xmath146 and @xmath147 come exclusively from the rotating radius vector @xmath148 .
this means the trapping potential as well as all the lasers producing the atom - light coupling @xmath77 should rotate with the same frequency @xmath1 around the rotation vector @xmath149 .
reference includes a comparison with an experimentally more feasible situation where the center of mass degrees of freedom are set into rotation , but the lasers ( and hence internal degrees of freedom ) are not .
note that realizing the rotating laser configuration is a challenging task , but this geometry can be realized by extending techniques for creating dynamical @xcite and rotating @xcite optical lattices .
figure [ fig_rotatinggaugefields ] depicts the required experimental geometry .
this setup uses a pair of scanning galvanometer - based mirrors that direct the raman lasers to intersect at the depicted cloud of atoms with wave - vectors @xmath150 and @xmath151 , with difference @xmath152 that rotates sinusoidally in the @xmath153 plane .
assuming that the internal degrees of freedom are characterized by the total angular momentum @xmath154 , the transformation to the rotating frame of the original hamiltonian @xmath155 [ eq . ( [ eq : h - full ] ) ] is described by the unitary operator @xmath156 $ ] involving the sum of the center of mass angular momentum @xmath5 and the internal angular momentum @xmath157 . transforming to the rotating frame the full atomic state - vector @xmath158 ,
one arrives at a tdse for @xmath159 described by the hamiltonian @xmath160 .
the latter is connected to the laboratory frame hamiltonian @xmath155 as @xmath161 and hence @xmath162\hat{i}+\hat{m}\left(\mathbf{r}\right)-\boldsymbol{\omega}_{\mathrm{rot}}\cdot\hat{\mathbf{\mathbf{f}}}\,,\label{eq : h - full - rotating - frame}\ ] ] where the rotation vector potential @xmath163 and the antitrapping ( centrifugal ) potential @xmath164 are as in eq .
( [ eq : a - rot ] ) .
the additional term @xmath165 is analogous to the hamiltonian describing the atomic internal spin in a magnetic field proportional to the rotation vector @xmath149 .
this is not surprising , because the rotation manifests itself like the magnetic field not only for the centre of mass motion , but also for the internal dynamics of the atom .
consequently the operator @xmath166 introduces a @xmath167 shift of the atomic spin states . for typical rotation experiments ,
the shift @xmath168 is small compared to the other frequencies characterizing the atomic internal dynamics .
the state - dependent operator @xmath169 featured in @xmath160 has a set of time - independent eigenstates @xmath170 with eigenenergies @xmath171 .
subsequently , like in the sec .
[ sub : formulation - gauge - potentials ] , the ( rotating frame ) atomic state - vector @xmath159 is transformed to @xmath172 by means of the unitary transformation @xmath173 converting the atomic bare states into the dressed ones .
the transformed hamiltonian @xmath174 has the same form as eq.([eq : h - transformed - r ] ) with the geometric vector potential @xmath175 replaced by @xmath176 and the centrifugal potential @xmath177 added : @xmath178 where we put @xmath179 because the dressed states are time - independent in the rotating frame .
the adiabatic approximation outlined in sect .
[ sub : adiabatic - approximation ] gives the effective hamiltonian for the atomic dynamics in the reduced internal space @xmath180 where the geometric scalar potential @xmath112 is as in eqs.([eq : w - q - general])-([eq : w - nm ] ) .
thus , rotation adds an abelian potential @xmath181 to the geometric vector potential @xmath106 , and also introduces the centrifugal potential @xmath164 .
in this section , we study the basic interaction between laser - light and atoms , before explicitly analyzing light - induced gauge potentials .
specifically , we consider the far off - resonant coupling @xcite between an alkali atom in its electronic ground state manifold and an oscillatory optical field , providing guidelines for designing realistic artificial gauge fields in alkali systems .
the same basic line of reasoning is generally valid in other atomic systems , but the ground and excited states will of course differ , and specific conclusions may not cross over @xcite .
we focus on the largest light - matter coupling term the electric dipole
( see ref .
@xcite , for example ) that links electronic motion to the optical electric field @xmath182 with vector components @xmath183 ( @xmath184 ) and frequency @xmath185 @xmath186 where the summation over repeated cartesian indices is assumed . here
, @xmath187 is the electric dipole operator , @xmath188 is the electron s charge , and @xmath189 is the position of the @xmath190th electron within the atom
. we will study the impact of this coupling term on the atomic ground state manifold to second order in perturbation theory , in a step - by - step manner including different effects in order of decreasing importance .
we analyze the lowest energy electric dipole transition in an alkali atom , between the ground @xmath191 electron orbital and the excited @xmath192 orbital with excitation energy @xmath193 , as pictured in fig .
[ fig : alkalilevelstructure ] .
the atom is then described by the hamiltonian @xmath194 where @xmath195 is the fine - structure coupling constant ; @xmath196 are the projectors onto the space of ground or excited states ; and @xmath197 and @xmath198 are the total electronic orbital angular momentum and the spin , respectively .
( the electronic orbital momentum @xmath197 should not be confused with the orbital momentum of the atomic center of mass motion @xmath199 appearing in the analysis of the rotating systems in the sects .
[ sect : rotation ] and [ sect : gaugefields ] . )
the orbital angular momentum @xmath197 has eigenstates @xmath200 , @xmath201 , with eigenvalues @xmath202 and @xmath203 for @xmath204 and @xmath205 , respectively ; the electron spin @xmath198 along the quantization axis @xmath52 has the eigenstates @xmath206 . in a similar manner ,
the atomic nuclear spin @xmath207 provides the eigenstates @xmath208 . because the electronic degree of freedom [ @xmath209 or @xmath210 is uniquely defined by the total electronic angular momentum quantum number , the operator @xmath211 projects onto the space of ground electronic states [ @xmath191 , with @xmath212 and @xmath213 projects onto the set of electronic excited states [ @xmath192 , with @xmath214 , as long as only these two sets of states are involved . to @xmath215 transition
( with their typical @xmath216 transition energies , and labeled by the archaic d1 and d2 designations for the two fine - structure resolved group of states ) required for the discussed herein .
this transition is driven by the far - detuned coupling laser indicated in red .
also pictured is the much smaller @xmath217 fine - structure splitting ( ranging from @xmath218 in @xmath219 to @xmath220 in @xmath221 ) that enables state - dependent `` vector '' light shift . , width=480 ] in the alkali atoms , there are three excited - state energy scales : the ground - excited splitting @xmath193 , the fine structure splitting @xmath217 ( with eigenstates of the combined angular momentum @xmath222 ) , and the typical scale of hyperfine splittings @xmath223 ( with eigenstates of the total angular momentum @xmath224 ) . in
any given alkali atom , these are ordered @xmath225 . in an experiment , the detuning from the excited states is generally large compared to @xmath223 .
consequently , to analyze the effect of far off - resonant light fields on ground state atoms , it suffices to include just the electronic excited level structure as corrected by the @xmath226 fine - structure term .
the hyperfine - structure of the excited level enables further transitions , but these can only be resolved when the lasers are tuned so near to the atomic resonance that the rate of spontaneous emission prohibits any practical use for engineering artificial gauge fields , or quantum simulation in general .
we focus on optical fields not too far from resonance , @xmath227 . in that case
it is convenient to transform the atomic excited state vectors @xmath228 into the rotating frame @xmath229 via the unitary transformation @xmath230 where the transformed state vector is @xmath231 , and the projection operators @xmath232 and @xmath233 are as defined above . in this temporally rotating frame , the dipole hamiltonian [ eq . ( [ eq : h - dip ] ) ] is @xmath234,\label{eq : h - dip - rot}\ ] ] where we introduced the complex electric field @xmath235 and applied the rotating wave approximation ( rwa ) by removing the terms oscillating at the frequencies @xmath185 and @xmath236 in the second relation of eq .
( [ eq : h - dip - rot ] ) . in the rotating frame and after the rwa , the full atomic and coupling hamiltonian is @xmath237 where the rotating frame atomic hamiltonian @xmath238 acquires an extra term @xmath239 $ ] from the the unitary transformation s temporal dependence .
hence , the excited state energy @xmath193 is supplanted by the detuning @xmath240 in the transformed hamiltonian @xmath241 within the ground state manifold ( adiabatically eliminating the excited states ) , the effect of the rwa electric dipole term @xmath242 is described to second - order by the effective atomic hamiltonian @xmath243 which can be represented as @xmath244 here we have introduced the rank-2 cartesian tensor operator @xmath245 acting on the ground state manifold . in the second relation of eq .
we have omitted the excited state projectors @xmath233 , because the atomic ground states do not have a permanent electric dipole moment : @xmath246 .
operators of this form can be expressed as the sum @xmath247 of irreducible tensor operators of rank-0 ( transforms as a scalar under rotation ) @xmath248 rank-1 ( transforms as a vector under rotation ) @xmath249 and rank-2 @xmath250 where the summation over the repeated cartesian indices is again implied .
this separation allows one to classify the light - matter interaction in a powerful way .
let us begin by considering the simple case when @xmath251 , implying that the excited states are degenerate , and allowing us to completely neglect electronic and nuclear spin . in this case
, we replace the atomic hamiltonian @xmath252 with @xmath253 , and the tensor @xmath254 in eq
. takes the particularly simple form @xmath255 in the present case , let us study how eq .
transforms under rotations as effected by the unitary operator @xmath256 .
because the ground s state has @xmath257 , it is clear that @xmath258 , therefore @xmath259 thus @xmath254 transforms like a scalar under rotation and must be proportional to the unit tensor .
consequently only the zero - rank contribution @xmath260 is non - zero @xmath261 where @xmath262 is proportional to the atoms ac polarizability ( without including the fine - structure corrections ) @xmath263 here @xmath264 is the commonly used reduced matrix element and @xmath265 is the bohr radius ( the reduced matrix element is a single number , even though the conventional notation makes it look like it should have three vector components ) . equations and yield the _ scalar light shift _ @xmath266 alternatively _ @xmath267 _ can be expressed as @xmath268 where we defined the intensity @xmath269 , used the speed of light @xmath270 , the transition s natural linewidth @xmath271 , the electric constant @xmath272 , and the expression @xmath273 for the linewidth @xcite .
this result is valid whenever it is safe to ignore the excited state fine structure ; a good zero - order approximation when the @xmath240 detuning from the optical transition is much larger than @xmath217 .
this result reminds us of several important facts : the scalar light shift is a potential that is independent of optical polarization , scales like @xmath274 , and for red - detuned beams ( @xmath275 ) is attractive .
let us now turn to the effects of the fine structure by recalling that the excited eigenstates @xmath276 of the hamiltonian @xmath252 eq .
corresponding to @xmath277 are characterised by the quantum numbers @xmath140 and @xmath278 of the combined angular momentum @xmath222 , with @xmath279 and @xmath280 .
specifically we have @xmath281 , where @xmath282 and @xmath283 are the detunings for the corresponding fine structure transitions .
next we shall determine the tensor @xmath284 describing the light - induced coupling between the atomic ground states , without explicitly involving the excited fine - structure states @xmath276 .
for this we use eq . for @xmath252 and write down a dyson - type equation for the inverse atomic hamiltonian @xmath285 projected onto the excited - state manifold , @xmath286 with @xmath287 and @xmath288 . calling on the communtation relations @xmath289=i\hbar\epsilon_{ijk}\hat{d}_{k}$ ] @xcite together with @xmath290 , we replace @xmath291 with the commutator @xmath292=i\hbar\epsilon_{i\ell m}\hat{p}_{g}\hat{d}_{m}\hat{\sigma}_{l}$ ] . using the relation @xmath293 , eqs . and yield @xmath294 where the reduced matrix element @xmath295 has already been featured in the previous subsection . as eq .
contains operators that act only within the electronic ground state , we may now omit the projectors @xmath232 .
furthermore , since the only operators acting in the electronic ground state are the pauli matrices , @xmath296 must take the form @xmath297 where we also used the fact that @xmath296 can be decomposed into a scalar and vector component ( no rank-2 component is possible with pauli matrices alone ) . replacing the proposed solution , permuting one levi - civita symbol and exploiting the identity @xmath298 , eq .
gives @xmath299 the resulting pair of linear equations has solutions @xmath300 in the initial notation with @xmath301 and @xmath240 , we associate the denominators appearing in the scalar coefficient @xmath302 with the detuning from the excited - state fine - structure split levels @xmath303 and @xmath304 , giving @xmath305 additionally introducing an average @xmath306 and the fine - structure splitting @xmath307 , the complete tensor operator takes the form @xmath308\,,\end{aligned}\ ] ] where we used @xmath309 as is suitable in the @xmath257 electronic ground state .
this leads to the light shift @xmath310\hat{p}_{g}\label{eq : h_eff - result}\end{aligned}\ ] ] in terms of the scalar and vector polarizabilities @xmath311 when @xmath312 , we recover our previous result , eqs . - .
as an illustration , consider an atom illuminated with circularly @xmath313 ( along @xmath52 ) polarized light , @xmath314 .
for this field @xmath315 , so the vector light shift is described by the operator @xmath316 acting on the ground state manifold .
consequently @xmath317 provides opposite light shifts for the spin - up and spin - down atomic ground states . in the presence of an external magnetic field , an alkali atom in its electronic ground state manifold
is described by the hamiltonian @xmath318 where @xmath319 is the bohr magneton ( henceforth the projector @xmath232 is kept implicit in the ground state operators ) . the first term in eq
. takes into account the coupling between the electron and nuclear spins of the atom , and is described by the magnetic dipole hyperfine coefficient @xmath320 .
the second ( zeeman ) term includes separate contributions from the electronic spin @xmath321 and the nuclear angular momentum @xmath322 , along with their respective land factors @xmath323 and @xmath324 . for the alkali atoms @xmath325 , so we will safely neglect the nuclear contribution @xmath326 to the zeeman term .
the previous subsection showed that laser fields induce the scalar and vector light shifts featured in the effective ground - state hamiltonian @xmath327 , eq . .
the two terms can be independently controlled by choosing the laser frequency @xmath185 and polarization . evidently , the vector light shift is a contribution to the total hamiltonian acting like an effective magnetic field @xcite @xmath328 that affects @xmath329 but not the nuclear spin @xmath322 .
thus the complete effective hamiltonian for the ground state atoms affected by the magnetic field and light reads @xmath330 where @xmath331 acts as a true magnetic field and adds vectorially with @xmath332 . instead of using the full breit - rabi equation @xcite for the zeeman energies , we assume that the zeeman shifts are small in comparison with the hyperfine splitting corresponding to the linear , or anomalous , zeeman regime . in this case
, the effective hamiltonian for a single manifold of total angular momentum @xmath333 states is obtained by replacing @xmath334 in the magnetic field interaction term @xcite , giving @xmath335 where we have introduced the hyperfine land g - factor @xmath336 and omitted the last term @xmath337 which is constant for the single hyperfine manifold with fixed @xmath338 .
for example , in @xmath339 s for which @xmath340 and @xmath279 in the electronic ground state , the lowest energy hyperfine manifold with @xmath341 corresponds to @xmath342 . in this way , at second order of perturbation the light shift contains the term @xmath343 described as a rank-0 ( scalar ) tensor operator resulting directly from the electric dipole transition , as well as the term @xmath344 described by the rank-1 ( vector ) tensor operator emerging from adding excited state fine structure . since the latter operator is linear with respect to the total momentum @xmath157 , it can induce only transitions which change @xmath345 by @xmath346 .
adding effects due to the excited state hyperfine coupling leads to an additional rank-2 tensor contribution to the light - induced coupling within the atomic ground - state manifold .
because the scale of this term is set by the @xmath347 to @xmath348 scale excited - state hyperfine couplings , the smallness makes these light shifts unsuitable for most equilibrium - system applications in quantum gases .
we now turn to a situation frequently encountered in the current experiments on artificial gauge potentials @xcite where an ensemble of ultracold atoms is subjected to a magnetic field @xmath349 and is simultaneously illuminated by possibly several laser beams with two frequencies @xmath185 and @xmath350 .
the frequency separation @xmath351 differs by a small detuning @xmath352 ( with @xmath353 ) from the linear zeeman shift between @xmath345 states .
the effective magnetic field induced by the complex electric field @xmath354 $ ] can be cast into its frequency components using eq .
@xmath355 where @xmath356 are the corresponding amplitudes . the first term @xmath357 adds to the static bias field @xmath358 , and the remaining time - dependent terms @xmath359 can drive transitions between different @xmath345 levels .
the real magnetic field is assumed to be much greater than the effective one @xmath360 , and the frequency difference @xmath361 is taken to be large compared to the kinetic energy scales . in this case , the time - dependence of the hamiltonian @xmath362 in eq .
can be eliminated via the unitary transformation @xmath363 and the subsequent application of the rotating wave approximation ( rwa ) by removing the terms oscillating at the frequencies @xmath361 and @xmath364 in the transformed hamiltonian @xmath365 .
the transformation does not alter @xmath366 , whereas the raising and lowering operators @xmath367 change to @xmath368 .
the resulting hamiltonian can be represented as @xmath369 the state - independent ( scalar ) potential is given by @xmath370 and @xmath371 where @xmath372 and @xmath373\,\label{eq : omega_pm_explicit}\ ] ] are the circular components of the rwa effective zeeman magnetic field @xmath374 .
the latter zeeman field is explicitly @xmath375\ez\nonumber \\ & -\frac{u_{v}g_{f}}{\hbar g_{j}}{\rm im}\left[\left({\mathbf{e}}_{\omega_{-}}^{*}\shorttimes{\mathbf{e}}_{\omega_{+}}\right)\!\cdot\!\left(\ex - i\ey\right)\right]\ex\label{eq : omega - final - general}\\ & -\frac{u_{v}g_{f}}{\hbar g_{j}}{\rm re}\left[\left({\mathbf{e}}_{\omega_{-}}^{*}\shorttimes{\mathbf{e}}_{\omega_{+}}\right)\!\cdot\!\left(\ex - i\ey\right)\right]\ey.\nonumber \end{aligned}\ ] ] the form of the effective coupling operator @xmath376 shows that , while it is related to the initial vector light shifts , the rwa effective zeeman magnetic field @xmath377 is composed of both static and resonant couplings in a way that goes beyond the restrictive @xmath378 form . in this way ,
the operator @xmath366 featured in the vector coupling @xmath376 , eq .
induces the opposites light shifts for the spin - up and spin - down atomic ground states , whereas the operators @xmath379 describe the raman transitions which change @xmath345 by @xmath346 . if the detuning from atomic resonance @xmath380 is large compared to the excited state fine structure splitting @xmath217 , the magnitude of this vector coupling proportional to @xmath381 drops off as @xmath382 , not like the scalar shift @xmath383 which drops off as @xmath274 .
since the amount of the off - resonant light scattering is also proportional to @xmath384 , the balance between the off - resonant scattering and the vector coupling is bounded , and can not be improved by large detuning .
while this is a modest problem for rubidium ( @xmath385 fine structure splitting ) , it turns to be a serious obstacle for atoms with smaller fine structure splitting such as potassium ( @xmath386 ) and lithium ( @xmath387 ) .
-@xmath43 plane with frequency @xmath185 and @xmath388 .
( b ) physical level diagram for three - level total angular momentum @xmath341 case , as is applicable for the common alkali atoms @xmath389li , @xmath390na , @xmath391 , @xmath392k , and @xmath393rb ( all of which have @xmath394 . the generally small quadratic zeeman effect that makes the @xmath395-@xmath396 energy difference different from the @xmath396-@xmath397 splitting , is not included in this diagram.,width=432 ] within this formalism , let us consider a straightforward example of two counter propagating raman beams depicted in fig .
[ fig_layout ] .
such a setup is often used in current experiments on synthetic gauge fields @xcite and spin - orbit coupling@xcite for ultracold atoms . in this case
@xmath398 describe the electric field of two lasers counterpropogating along @xmath42 with equal intensities , crossed linear polarization and wave - number @xmath399 , giving @xmath400 and @xmath401 .
consequently @xmath402 and @xmath403 where @xmath404 is the rabi frequency of the raman coupling .
the resulting effective zeeman field @xmath374 entering the vector light shift is @xmath405.\label{eq : onedimensioneffectivefield}\ ] ] together , @xmath406 and @xmath407 describe a constant scalar light shift along with a spatially rotating zeeman field
. it is evident from eqs . and that for this spatially rotating zeeman field to be non - zero , the effective zeeman field @xmath408 must not be parallel to the quantizing axis defined by the applied magnetic field @xmath349 . in the present case
, this implies that the counter propagating beams can not be aligned along @xmath52 . physically , any projection of @xmath408 along @xmath52 yields a spin - dependent optical lattice @xcite rather than a spatially rotating zeeman field .
but since the raman beams have different frequencies , the spin - dependent lattice is moving fast and vanishes after performing the rwa .
note also that in order to produce the vector light shifts like the one given by eq .
( [ eq : onedimensioneffectivefield ] ) , there is no need for the two raman beams to counter - propagate .
when the beams intersect with opening angle @xmath190 , the momentum exchange is changed from @xmath409 to @xmath410 $ ] . since this altered wave - vector simply changes the energy and length scales of the problem , we shall concentrate on the counter propagating setup , eqs . .
to obtain an exact correspondence with the notations used in most of the experimental papers on the artificial gauge fields @xcite and the spin - orbit coupling @xcite , one needs to interchange the @xmath52 and @xmath43 axes in our analysis .
it is instructive to note that the @xmath411 component of the effective zeeman field @xmath374 , eq .
, does not depend on the intensity of the raman beams .
the situation is not universal .
for example , consider a pair of raman laser beams with complex electric fields @xmath412 intersecting with angle @xmath413 , and where the @xmath414 beam is circularly polarized @xcite . in this case
@xmath415 , @xmath416 , and @xmath417 giving @xmath418 so @xmath419 becomes intensity - dependent , providing an extra control of the light - induced gauge potential .
finally , more sophisticated spatial dependence of the zeeman field @xmath374 can be created using additional raman beams with specially chosen polarizations , for example leading to optical flux lattices @xcite , to be considered in the section [ sub : flux - latt ] .
in this section , we describe how the atom - light coupling can be exploited to generate abelian gauge potentials in cold - atom systems .
we present useful schemes based on the adiabatic motion of atomic dressed states , and we eventually describe the specific experimental setup used at nist to generate synthetic gauge fields .
let us begin our analysis of abelian gauge potentials with the illustrative case of atoms characterized by the @xmath420 type level structure , where the laser beams couple two atomic internal states @xmath421 and @xmath422 with a third one @xmath423 , see figure [ lambda ] . the atom light coupling operator @xmath77 featured in the full atomic hamiltonian ( [ eq : h - full ] ) has then the following form : @xmath424 where the frequencies @xmath352 and @xmath425 characterize the detuning between the atomic states .
the rabi frequencies @xmath426 and @xmath427 are generally complex and position - dependent . in the context of the light - induced gauge potential for ultracold atoms ,
the @xmath420 scheme was first explored in the late 90 s @xcite .
subsequently it was shown that the scheme can yield a non - zero artificial magnetic field ( inducing the lorentz force for electrically neutral atoms ) using non - trivial arrangements of laser fields @xcite or position - dependent detuning @xcite . in most of these treatments
the state @xmath423 is assumed to be the atomic excited state coupled resonantly to the ground states @xmath421 and @xmath422 by laser beams , as shown in fig .
[ lambda ] .
the involvement of the excited states is inevitably associated with a substantial dissipation due to the spontaneous emission .
scheme of the atom - light coupling in which the laser beams induce atomic transitions @xmath428 and @xmath429 characterized by the rabi frequencies @xmath430 and @xmath431 . , width=302 ] [ lambda ] to avoid such losses , one can choose @xmath423 to be an atomic ground state coupled to the other two ground states @xmath421 and @xmath422 via the raman transitions @xcite . in that case @xmath430 and
@xmath431 featured in the coupling operator ( [ eq : m - lambda ] ) represent the raman rabi frequencies .
this forms a ladder scheme which is equivalent to the @xmath420 sheme .
in particular , if @xmath432 and @xmath433 , the hamiltonian ( [ eq : m - lambda ] ) reduces to the hamiltonian given by eq . for an atom in the @xmath434 ground state manifold affected by the rwa effective zeeman field @xmath435 .
the latter @xmath374 is induced by the raman transitions and the real magnetic field . in this case
the states @xmath421 , @xmath423 and @xmath422 corresponds to the magnetic sublevels with @xmath436 , @xmath437 and @xmath438 , respectively . in the next subsection we shall analyse in detail the gauge potentials resulting from such a coupling between the magnetic sublevels .
if @xmath439 the atom - light coupling operator ( [ eq : m - lambda ] ) can no longer be represented as a hamiltonian for the atomic spin ( or quasispin ) interacting with the effective magnetic field .
assuming exact resonance between the atomic levels @xmath440 and @xmath441 ( @xmath442 ) , the interaction operator @xmath443 given by eq .
( [ eq : m - lambda ] ) has a single dressed eigenstate @xmath444 known as the dark or uncoupled state : @xmath445 where @xmath446 is a relative phase between the two rabi frequencies .
the state @xmath444 contains no contribution from the excited state @xmath423 and is characterised by a zero eigenvalue @xmath447 .
dark states are frequently encountered in quantum optics .
they play an important role in applications such as the electromagnetically induced transparency ( eit ) @xcite and the stimulated raman adiabatic passage ( stirap ) @xcite , relying on the fact that the excited level @xmath423 is not populated for the dark state atoms and spontaneous decay is therefore suppressed . as we have seen in section [ sub :
formulation - gauge - potentials ] , if the atom is in a selected internal dressed state well separated from the remaining dressed states , an abelian geometric vector potential emerges for the centre of mass motion . for atoms in the internal dark state
the adiabatic motion takes place if the total rabi frequency @xmath448 exceeds the characteristic kinetic energy of the atomic motion .
the corresponding vector potential @xmath449 and the associated magnetic field @xmath117 are @xmath450 @xmath451 the geometric scalar potential for the dark state atoms is given by @xmath452 one easily recognizes that the vector gauge potential yields a non - vanishing artificial magnetic @xmath453 only if the gradients of the relative intensity and the relative phase are both non - zero and not parallel to each other .
therefore the synthetic magnetic field can not be created for the dark state atoms of the @xmath420 scheme using the plane - waves driving the transitions @xmath454 and @xmath455 @xcite .
however , plane waves can indeed be used in more complex tripod @xcite or closed loop @xcite setups to generate non - abelian gauge fields for a pair of degenerate internal dressed states , as we shall see in the section [ sect : soc ] .
equation ( [ eq : b - eff-2 ] ) has a very intuitive interpretation @xcite .
the vector @xmath456 $ ] connects the `` center of mass '' of the two light beams and @xmath457 is proportional to the vector of the relative momentum of the two light beams .
thus a nonvanishing @xmath453 requires a _ relative orbital angular momentum _ of the two light beams .
this is the case for light beams carrying optical vortices @xcite or if one uses two counterpropagating light beams of finite diameter with an axis offset @xcite , as depicted in fig .
[ lambda-1 ] .
a more detailed analysis of these schemes is available in the cited references and the previous reviews @xcite .
scheme shown in fig .
[ lambda].,scaledwidth=45.0% ] let us now turn to a situation where the atom - light hamiltonian can be represented as an interaction of a spin @xmath458@xmath157 with an effective zeeman field @xmath459 where we have parametrized the zeeman vector @xmath460 in terms of the spherical angles @xmath461 and @xmath462 shown in fig .
9.@xmath463 where @xmath464 is the length of the zeeman vector .
such a coupling can be produced using bichromatic laser beams considered in the previous section , leading to eq . which has the same form as eq.([eq : m - r - spin - interaction - general ] ) .
for the spin @xmath440 case , the hamiltonian ( [ eq : m - r - spin - interaction - general ] ) corresponds to the @xmath420 ( ladder ) scheme described by the hamiltonian ( [ eq : m - lambda ] ) as long as @xmath465 and @xmath433 .
note that the operator @xmath157 does not necessarily represent the true atomic spin operator ; it can be other atomic vector operators with cartesian components obeying the angular momentum algebra . in terms of the spherical angles
@xmath461 and @xmath462 .,width=192 ] [ fig : omega - spherical ] the coupling hamiltonian @xmath77 can be diagonalised via a unitary transformation @xmath466 the first operator @xmath467 rotates the spin around the @xmath411-axes by the angle @xmath462 eliminating the phase @xmath462 in @xmath376 , eq.([eq : m - r - spin - interaction - general ] ) . the subsequent transformation using the second operator
@xmath468 rotates the spin around the @xmath469-axes by the angle @xmath461 making the operator @xmath376 proportional to @xmath366 .
although the third operator @xmath470 no longer affects the transformed operator @xmath376 , its inclusion ensures that the whole transformation @xmath471 reduces to the unit operator for zero polar angle @xmath472 and arbitrary azimuthal angle @xmath462 .
thus one has @xmath473
the transformed hamiltonian @xmath474 has eigenstates @xmath475 characterized by the total angular momentum @xmath338 and its @xmath52 projection @xmath89 . in the following ,
we use both notations @xmath89 and @xmath476 to denote the projection of the total angular momentum .
therefore the eigenstates @xmath477 of the original atom - light hamiltonian @xmath77 read @xmath478 the corresponding eigenenergies being @xmath479 where both the eigenenergies @xmath480 and also the local eigenstates @xmath481 are position - dependent through the position - dependence of the effective zeeman vector @xmath482 .
it is to be noted that similar kinds of eigenstates give rise to artificial gauge potentials describing the rotation of diatomic molecules @xcite and the physics of atomic collisions @xcite .
if an atom is prepared in a dressed state @xmath483 and if its characteristic kinetic energy is small compared to the energy difference between adjacent spin states @xmath484 , the internal state of the atom will adiabatically follow the dressed state @xmath483 as the atom moves , and contributions due to other states with @xmath485 can be neglected .
projecting the atomic dynamics onto the selected internal eigenstate @xmath486 yields a reduced equation for the atomic center of mass motion affected by the geometric vector potential @xmath487 and the scalar potential @xmath488 which are generally defined by eq .
( [ eq : a - v q - eq-1 ] ) . in the current situation
these potentials explicitly read @xcite @xmath489 and @xmath490\left[\sin^{2}\theta\left(\nabla\phi\right)^{2}+\left(\nabla\theta\right)^{2}\right],\label{eq : w - spin}\end{aligned}\ ] ] the geometric vector potential @xmath491 yields the artificial magnetic field @xmath492 which is non - zero if @xmath493 and @xmath494 are not parallel to each other . both @xmath491 and @xmath495
have the largest magnitude for maximum absolute values of the spin projection @xmath89 and are zero for @xmath437 . on the other hand ,
the geometric scalar potential @xmath496 is maximum for @xmath437 and reduces with increasing @xmath497 .
the atomic motion is affected by three distinct scalar potentials : ( a ) the state independent potential @xmath32 representing the `` scalar light shift '' given by eq . for bichromatic fields , ( b ) the `` adiabatic scalar potential '' @xmath498 arising from spatial variations in the magnitude of the zeeman vector @xmath499 , and ( c )
the `` geometric scalar potential '' @xmath496 described above .
all three contribute to the potential energy of atoms in the dressed state basis . for two counterpropagating raman beams , the effective zeeman field @xmath374 , eq .
, defining the gauge fields , is characterised by the azimuthal angle @xmath500 with the gradient @xmath501 equal to the raman recoil wave - vector @xmath502 . on the other hand ,
the polar angle @xmath461 and the length of the zeeman vector @xmath464 are determined by the detuning @xmath352 and the raman rabi frequency @xmath503 with @xmath504 typically experiments are performed in the lowest energy dressed state , where @xmath505 assumes its maximum absolute value .
equations ( [ eq : vect - spin ] ) and ( [ eq : grad - phi])-([eq : cos - theta ] ) illustrate two important points : ( 1 ) the magnitude of the vector potential @xmath506 is defined by the recoil momentum of the lasers as well as the spin projection @xmath507 , and ( 2 ) the vector potential is strictly bounded between @xmath508 , where the maximum possible value is the recoil momentum @xmath509 times the spin @xmath338 .
we note that employing an extra spatially uniform radio - frequency magnetic field adds a constant term to the spatially oscillating @xmath510 component of effective zeeman field @xmath374 , eq . .
in that case both the adiabatic energy @xmath511 and the geometric scalar potential @xmath512 become spatially oscillating functions , thus creating a composite lattice potential in addition to the vector potential @xmath491 @xcite .
we shall discuss this issue in more detail in the section [ sec : other schemes ] .
returning to eqs .
( [ eq : b - spin])-([eq : cos - theta ] ) , we see that in order to have a non - zero @xmath453 , either the optical intensity or the detuning must depend on @xmath469 or @xmath411 , i.e. their gradient(s ) should not be parallel to the wave - vector @xmath513 .
the position - dependent detuning can be generated by the vector light shift itself , which is absent for co - propagating light beams .
this provides an additional term analogous to @xmath352 , but proportional to the raman rabi frequency .
this is the case if the laser beams propagate at the angle @xmath413 , one of them being circularly polarised @xcite , as one can see in eq . of the previous section .
therefore , by simply making @xmath442 and having the intensity - dependent detuning ( which adds to @xmath352 ) depend on position in eq . , it is possible to create artificial gauge potentials which are truly geometric in nature , as the absolute light intensity then completely vanishes from the azymuthal angle @xmath461 .
the first experimental implementation of the synthetic magnetic field @xcite was based on a different insight , namely that with constant @xmath503 a gradient of the detuning @xmath352 due to a magnetic field gradient also generates a non - zero artificial magnetic field @xmath453 providing an artificial lorentz force .
the generated field @xmath453 is not purely geometric , and depends on the relative strength of @xmath352 and @xmath503 in addition to their geometry .
it is noteworthy that the maximum magnetic flux produced by means of two counterpropagating laser beams amounts to @xmath514 dirac flux quanta and is thus proportional to the systems length @xmath515 @xcite .
this is a drawback in creating very large magnetic fluxes necessary for reaching the fractional quantum hall regimes .
the use of the optical flux lattices considered in the following section overcomes this drawback making the induced flux proportional to the area rather than the length of the atomic cloud .
note also that instead of counterpropagating laser fields one can employ co - propagating beams carrying optical vortices with opposite vorticity @xcite .
for the first - order laguerre - gaussian beams the azimuthal angle @xmath462 entering the vector and scalar potentials is then twice the real space azimuthal angle . on the other hand ,
the raman rabi frequency @xmath503 , entering in eqs .
( [ eq : grad - phi])-([eq : cos - theta ] ) for the polar angle , linearly depends on the cylindrical radius in the vicinity of the vortex core .
non - zero artificial magnetic fields can therefore be generated without making the detuning position - dependent .
using vortex beams the number of dirac flux quanta imparted onto the atomic cloud is determined by the vorticity of the beams @xcite and is thus normally much smaller than that induced by counter - propagating laser fields .
raman transitions providing the optical flux lattice for @xmath516,scaledwidth=40.0% ] .
optical flux lattices were introduced by n. r. cooper in ref .
this concept is based on the observation that singularities in the synthetic vector potential could be created and exploited so as to produce periodic structures with non - trivial magnetic fluxes @xcite .
such optical flux lattices could be created using a bichromatic laser field @xcite , as considered in the previous section . following the proposal by cooper and dalibard @xcite
, we consider that the @xmath516 ground state magnetic sublevels are coupled via raman transitions ( fig .
[ fig : raman - scheme ] ) involving laser fields with two frequencies .
one frequency component @xmath517 shown in green represents a circular @xmath518 polarized plane wave propagating along the @xmath411 axis .
another frequency field @xmath519 shown in red has all three circular polarizations .
dalibard and cooper considered a situation where the field @xmath519 represents a superposition of three linearly polarized plane waves propagating in the @xmath42-@xmath43 plane and intersecting at 120 degrees , with polarizations being tilted by some angle with respect to the propagation plane @xcite . in that case
one can produce triangular or hexagonal state - dependent lattices for the adiabatic motion of laser - dressed atoms affected by a non - staggered magnetic flux .
a square optical flux lattice is formed if the first frequency component @xmath517 is again a circular @xmath518 polarized plane wave propagating along the @xmath411 axis , yet the second frequency field @xmath519 represents a sum of polarization - dependent standing waves with a @xmath520 time - phase difference between the standing waves oscillating in the @xmath42 and @xmath43 directions @xcite . in that case one arrives at an atom - light coupling described by the state - dependent hamiltonian @xmath521 with a spatially periodic zeeman field @xmath522 : @xmath523 where a dimensionless ratio @xmath524 is controlled by changing the polarisations of the standing waves @xcite . atoms with a fixed spin projection @xmath89 ( along the zeeman field @xmath522 ) then undergo an adiabatic motion in a square lattice potential @xmath525 with a periodicity in the @xmath42 and @xmath43 directions twice smaller than the periodicity @xmath526 of the atom - light hamiltonian @xmath527 .
the adiabatic motion is accompanied by a vector potential @xmath506 containing two aharonov - bohm type singularities ( per elementary cell ) corresponding to the points where @xmath528 and @xmath529 . in the vicinity of each singular point
, the vector potential @xmath530 has a dirac - string piercing the @xmath42-@xmath43 plane and carrying @xmath531 dirac flux quanta ( with @xmath532 ) @xcite @xmath533 note that these singular fluxes are gauge - dependent and non - measurable .
one can shift each singular points to another location via a gauge transformation @xcite , yet the singularities can not be removed from the @xmath42-@xmath43 plane . due to the periodicity of the system , the total flux over the elementary cell is equal to zero @xmath534 where @xmath535 is the total magnetic flux density composed of the continuous ( background ) magnetic flux density @xmath536 and a set of above - mentioned gauge - dependent singular fluxes @xmath537 of the aharonov - bohm type corresponding to @xmath538 , eq .
( [ eq : a_m - ab ] ) .
thus it is strictly speaking impossible to produce a non - zero effective magnetic flux over the elementary cell using the periodic atom - light coupling .
yet this does not preclude having a non - staggered continuous magnetic flux density @xmath495 over the elementary cell , because of the existence of non - measurable dirac strings . deducting the latter ,
the continuous physical flux over the elementary cell is @xmath539 where the summation is over the singular points of the vector potential ( emerging at @xmath540 ) around which the contour integration is carried out .
to summarize , the optical flux lattice contains a background non - staggered magnetic field @xmath453 plus an array of gauge - dependent dirac - string fluxes of the opposite sign as compared to the background .
the two types of fluxes compensate each other , so the total magnetic flux over an elementary cell is zero , as it is required from the periodicity of the hamiltonian .
however , the dirac - string fluxes are non - measurable and hence must be excluded from any physical consideration . as a result , a non - staggered magnetic flux over the optical flux lattice appears , and topologically non - trivial bands with non - zero chern numbers can be formed @xcite . by choosing the proper laser beams producing the optical flux lattice
, one can produce a configuration where the lowest energy band is topologically non - trivial and nearly dispersionless , leading to a possible formation of fractional quantum hall states @xcite .
note also that unlike other lattice schemes involving laser - assisted - tunneling methods , such as proposed in refs .
@xcite ( see section [ sect : lattices ] ) , the concept of optical flux lattices is based on the adiabatic motion of the atoms .
finally , it is also convenient to reinterpret optical flux lattices as tight - binding models defined in reciprocal space @xcite .
this complementary picture offers an efficient method to establish the atom - light coupling configurations leading to non - trivial topological bands : following ref .
@xcite , any tight - binding model exhibiting non - trivial chern insulating phases , such as the haldane model @xcite , could constitute the roots of an interesting optical flux lattice .
the preceding discussion followed our initial introduction to artificial gauge fields ( sect .
[ sect : gaugefields ] ) , where we first explicitly solved the state - dependent part of the hamiltonian @xmath376 , thereby introducing a vector potential @xmath541 for the adiabatic motion of atoms in the dressed - state manifold .
this elegant description clearly illustrates the geometric origin of these gauge fields , but the projection process is an approximation , and in many cases , relevant information can be obtained by directly solving the total hamiltonian ( including the kinetic energy part and the atom - light coupling @xmath77 ) .
this can be done explicitly if the atom - light coupling is induced by plane waves @xcite , such as in the case of two counterpropagating laser beams considered above . in order to establish a general framework that accommodates the effects of inter - particles collisions , we find convenient to use the second - quantized formalism , where the hamiltonian for a particle in a uniform time - independent magnetic field normal to a 2d plane is @xmath542 ^ 2 + \left[k_y - \frac{{\mathcal a}_y}{\hbar } \right]^2\right\}\hat \psi({\bf x}),\end{aligned}\ ] ] and @xmath543 is the field operator for the creation of a particle at position @xmath544 , with momentum represented by @xmath545 .
we have in mind a situation which explicitly realizes eq .
( [ fieldham ] ) in a specific gauge .
the actual gauge will depend on the details of the experimental setup , and will be in this case the landau gauge with @xmath546 .
the goal is to arrive at a hamiltonian where the minimum of the energy - momentum dispersion relation @xmath547 becomes asymmetric @xcite and is displaced from zero momentum as a function of spatial position .
the dressed single - particle states are spin and momentum superpositions whose state decomposition depends on the effective vector potential @xmath548 . the canonical momentum associated with this vector potential
can be observed by probing the internal state decomposition of the dressed states , as was done by lin et al @xcite , see sect .
[ sec : eximplementation ] .
this approach relies on a collection of atoms with two or more electronic ground states which interact with two counter - propagating raman coupling lasers aligned along @xmath42 .
if both raman beams are far detuned from the ground to excited state transition there is negligible population in the excited state .
the raman beams then induce a coupling @xmath549 between ground states which can lead to synthetic magnetic fields .
we will first consider a coupled two level system with internal states @xmath550 and @xmath551 , where exact solutions can be obtained @xcite .
physically , these two states might be two @xmath476 levels in the ground state manifold of an alkali atom .
for example , the @xmath434 manifold of rb@xmath393 at large enough field such that the quadratic zeeman effect can resolve two out of the three zeeman sublevels for the raman or radio frequency transitions . in
the frame rotating with angular frequency @xmath552 , the raman fields are detuned @xmath553 from resonance . the atom - light coupling term in the rotating wave approximation is then @xmath554\right .
\\ & + \left.\frac{\hbar\delta}{2}\left[\hat\phi_+^\dagger(k_x , y)\hat \phi_+(k_x , y ) - \hat\phi_-^\dagger(k_x , y)\phi_-(k_x , y)\right]\right\}\end{aligned}\ ] ] the notation @xmath555 denotes the creation of a particle with wave vector @xmath556 along @xmath42 at position @xmath469 , with @xmath557 .
here @xmath558 also includes the raman detuning terms . in the following @xmath559 and @xmath352
will be treated as spatially varying functions of @xmath469 , but not @xmath510 .
this is the explicit expansion of the @xmath560 contribution to eq .
( [ rwa_hamiltonian ] ) with the effective magnetic field from eq . in terms of field operators for the two - level case . with no coupling ,
the hamiltonian is given by @xmath561 , which represent motion along @xmath42 , motion along @xmath43 , the external potential , and interparticle interactions respectively . when expressed in terms of the real space field operators
@xmath562we obtain @xmath563 the contact interaction for collisions between ultracold atoms in 3d is set by the 3d s - wave scattering length @xmath564 , here assumed to be state independent .
strong confinement in one direction results in an effective 2d coupling constant @xmath565 where @xmath566 is the harmonic oscillator length resulting from a strongly confining potential along @xmath52 .
finally , @xmath567 is an external trapping potential , which we assume also to be state - independent .
this problem can be solved exactly when considering free motion along @xmath42 , i.e. , treating only @xmath568 and @xmath558 , and going to the momentum representation for the atomic motion along the @xmath510 axis .
the second quantized hamiltonian for these two contributions can be compactly expressed in terms of the operators @xmath569 . using this set
, @xmath570 reduces to an integral over 2@xmath5712 blocks @xmath572 labeled by @xmath573 and @xmath469 .
the expression in terms of @xmath574 operators instead of @xmath575 is a gauge transformation which boosts the @xmath550 and @xmath551 states in opposite directions .
the dependence of the two - photon coupling @xmath559 and detuning @xmath352 on @xmath469 has been suppressed for notational clarity .
the resulting hamiltonian density for motion along @xmath42 at a fixed @xmath469 is then @xmath576for each @xmath573 , @xmath577 can be diagonalized by the unitary transformation @xmath578 .
unlike the approximate solutions discussed in the context of adiabatic gauge fields in subsections [ subsect : lambda]-[sub : spin - in - magn - field ] , these represent the exact solution for motion along @xmath42 .
the resulting eigenvalues of @xmath577 , @xmath579\end{aligned}\ ] ] give the effective dispersion relations in the dressed basis , @xmath580 . for each @xmath581
the eigenvectors of @xmath582 form a family of states @xcite . in terms of the associated real - space operators @xmath583 these diagonalized terms of the initial hamiltonian are @xmath584 in analogy with the term `` crystal - momentum '' for particles in a lattice potential , we call the quantum number @xmath573 the `` quasi - momentum '' . here
@xmath585 is the real - space representation of the quasi - momentum @xmath573 .
the symbol @xmath586 is a differential operator describing the dispersion of the dressed eigenstates , just as the operator @xmath587 describes quadratic dispersion along @xmath42 of a charged particle moving in a magnetic field @xmath588 in the landau gauge .
to lowest order in @xmath589 and second order in @xmath590 , @xmath591 can be expanded as @xmath592 atoms in the dressed potential are significantly changed in three ways : ( 1 ) the energies of the dressed state atoms are shifted by a scalar shift analogous to the sum of the adiabatic and geometric potentials @xmath593 and @xmath594 . ( 2 ) atoms acquire an effective mass @xmath595 .
this change is absent in the adiabatic picture .
( 3 ) the centre of the dispersion relation is shifted to @xmath596 . just as with the adiabatic case considered in the previous subsections [ subsect : lambda]-[sub : spin - in - magn - field ] ,
@xmath597 can depend on @xmath469 by a spatial dependence on @xmath559 , or , via @xmath598 as described below . in either case ,
the effective hamiltonian is that of a charged particle in a magnetic field expressed in the landau gauge . , and the vertical axis is the dressed state energy @xmath599 .
the black lines are the exact eigenvalues of eq . , and the red dashed lines are the analytic approximation .
( a ) bare potentials with raman beams on resonance @xmath600 .
( b ) dressed potentials with @xmath601 , @xmath602 .
( c ) bare potentials with raman beams off resonance @xmath603 and @xmath604 ) .
( d ) dressed potentials with raman beams off resonance ( @xmath601 , @xmath604).,width=384 ] figure [ 2dressedstates ] shows the dressed state dispersion relations in this model .
panels ( a ) and ( c ) show the undressed case ( @xmath605 ) for detuning @xmath606 and @xmath607 respectively .
panels ( b ) and ( d ) depict the same detunings , for @xmath601 , where the exact results ( solid line ) are displayed along with the approximate dispersion ( red dashed line ) .
[ 2dressedstates]b shows the strongly dressed states for large @xmath559 , each of which is symmetric about @xmath608 .
when detuned as in panel c , the dispersion is displaced from @xmath608 .
if space dependent this displacement leads to a non - trivial gauge potential . in the limit of very small @xmath559
the dressed curve @xmath609 forms a double - well potential as a function of @xmath556 . in a related raman - coupled system ,
bose condensation in such double - well potentials were studied theoretically @xcite , and has been explored in detail in the context of spin - orbit coupling @xcite .
the raman coupling leads to a dressed dispersion along @xmath42 , where motion along @xmath43 is largely unaffected .
when the detuning is made to vary linearly along @xmath43 , with @xmath610 , an effective single particle hamiltonian contains a 2d effective vector potential @xmath611 .
the synthetic magnetic field is therefore @xmath612 .
figure [ limitations ] shows the vector potential as a function of detuning @xmath352 .
as expected , the linear approximation discussed above ( dashed line ) is only valid for small @xmath352 .
the synthetic field therefore decreases from its peak value as @xmath352 increases ( top inset ) .
this technique also modifies the trapping potential along @xmath43 . in the adiabatic picture , this comprises the sum of the adiabatic and geometric potentials @xmath593 and @xmath594 which have no individual identity in this exact solution . when the initial potential @xmath613 is harmonic with trapping frequencies @xmath614 and @xmath615 , the combined potential along @xmath43 becomes @xmath616 y^2/2 $ ] , where @xmath617 .
this contribution to the overall trapping potential is not unlike the centripetal term which appears in a rotating frame of reference , where a synthetic magnetic field @xmath453 arises as well . in the case of a frame rotating with angular frequency @xmath464 ,
the centripetal term gives rise to a repulsive harmonic term with frequency @xmath618 . in the present case
the scalar trapping frequency can be rewritten in a similar form @xmath619 .
the scalar potential may be attractive or repulsive , and it increases in relative importance with increasing @xmath559 . versus detuning @xmath620 for @xmath621 .
the solid line is the exact result and the dashed line is the lowest order expansion in @xmath352 .
the inset shows the synthetic magnetic field @xmath622 , indicating the predicted degree of field inhomogeneity predicted quantities relevant to experiment , computed for rb@xmath393 with a detuning gradient @xmath623 , and @xmath624 raman lasers .
, width=324 ] we have so far omitted motion along @xmath43 , interactions , and the effect of an external potential .
the inclusion of these effects are discussed in detail in ref .
the outcome is that in the limit of large @xmath559 these are unchanged by the transformation into the dressed basis .
together these allow the construction of the real space hamiltonian in the basis of localized spin - superposition states @xmath625 .
the density distribution of these localized states is not perfectly localized , but are some fraction of an optical wavelength @xmath626 in extent . for interactions , this is important because dressed state atoms separated by this distance will in fact interact . for particles starting in higher bands , transitions to lower bands
are energetically allowed and a fermi s golden rule argument thus gives rise to decay from all but the lowest energy dressed state @xcite .
this technique is not without its limitations .
foremost among them is the range of possible @xmath627 shown in fig .
[ limitations ] where @xmath621 . while the linear expansion ( dashed )
is unbounded , the exact vector potential is bounded by @xmath628 .
for example , the hybridized combination of @xmath629 and @xmath630 can not give rise to dressed states with minima more positive than @xmath631 where the energy of the @xmath550 states is minimized without dressing , fig .
[ 2dressedstates ] ( c ) .
the minima can also not be more negative than @xmath632 .
this limitation does not effect the maximum attainable field , only the spatial range over which this field exists .
specifically , a linear gradient in @xmath598 gives rise to the synthetic field @xmath633 which is subject to @xmath634 .
this simply states that the vector potential bounded by @xmath635 is the integral of the magnetic field .
note however , that along @xmath42 the region of large @xmath636 has no spatial bounds .
a second limitation of this technique is the assumption of strong raman coupling between the zeeman split states . as already pointed out at the end of the sec .
[ sec : tworaman - general ] , for alkali atoms , when the detuning from atomic resonance @xmath637 is large compared to the excited state fine structure @xmath217 , the two - photon raman coupling for @xmath638 transitions drops as @xmath639 , not @xmath640 as for the ac stark shift . as a result , the ratio between the raman coupling @xmath559 and off - resonant scattering can not be increased by using larger detuning , and the only solution is to consider atoms with large fine structure splitting @xmath217 , such as rubidium . for three - level atoms dressed by counter - propagating raman beams .
the horizontal axis is quasi - momentum @xmath573 , and the vertical axis is energy .
( a ) bare potentials ( undressed ) with raman beams on resonance .
( b ) dressed potentials ( @xmath641 , @xmath602 ) .
( c ) bare potentials ( undressed ) with raman beams off resonance ( @xmath604 ) .
( d ) dressed potentials with raman beams off resonance ( @xmath641 , @xmath642 ) .
, width=324 ] the range of possible effective vector potentials can be extended by coupling more states , for example the @xmath476 states of an @xmath643 manifold in the linear zeeman regime .
we can follow the two - level example above , except there are no compact closed - form solutions . in the adiabatic picture ,
eqs ( [ eq : vect - spin ] ) and ( [ eq : grad - phi])-([eq : cos - theta ] ) , additional levels extended the range of the vector potential from @xmath644 ( the two level case ) to @xmath645 for arbitrary @xmath338 .
this can be illustrated by considering an optically - trapped system of rb@xmath393 atoms in the @xmath341 manifold in a small magnetic field which splits the three @xmath476 levels by @xmath646 .
the @xmath647 blocks @xmath648 describing the three internal states of the @xmath341 manifold are @xmath649 in this expression , @xmath352 is the detuning of the two photon dressing transition from resonance , @xmath650 accounts for any quadratic zeeman shift ( not included in the sect . [
sub : spin - in - magn - field ] ) , @xmath559 is the two - photon transition matrix element , and @xmath573 , in units of the recoil momentum @xmath651 , is the atomic momentum displaced by a state - dependent term @xmath652 for @xmath653 , @xmath654 for @xmath655 , and @xmath656 for @xmath657 . here , the three eigenvalues are denoted by @xmath658 and @xmath659 . as with the two - level case , the states associated with eigenvalues @xmath660 experience an effective vector potential which can be made position - dependent with a spatially varying detuning @xmath352 .
the @xmath659 also experiences a synthetic field , but at higher order in @xmath661 .
a magnetic field gradient along @xmath43 gives @xmath662 , and generates a uniform synthetic magnetic field normal to the plane spanned by the dressing lasers and real magnetic field @xmath663 .
the preceding sections described an overall procedure by which light - induced artificial gauge fields can be created . in this section , we look at three effects : the introduction of a spatially uniform vector potential @xmath548 , the use of a temporal gradient to induce an electric field @xmath664 , and lastly the inclusion of a spatial gradient which gives rise to a magnetic field @xmath665 . for more technical details
we refer the reader to the original publications @xcite .
components differing in momentum by @xmath666 , displaced from zero @xmath581 when @xmath667 .
the horizontal arrow depicts the displacement of the @xmath655 cloud from zero corresponding the the non - zero vector potential .
( b ) measured momentum of @xmath668 component as a function of detuning @xmath352 . with the gauge choice @xmath669 , this constitutes a direct measurement of the artificial vector potential @xmath670 .
the dashed curve is the predicted vector potential for the experimental parameters .
data first appeared in y .- j .
lin , _ et al .
_ prl * 102 * , 130401 ( 2009 ) , and y .- j .
lin , _ et al .
* 7 * , 531 ( 2011 ) , refs .
@xcite , and @xcite .
, width=432 ] figure [ fig_vectorpotential ] shows the basic principle for creating artificial vector potentials . in these experiments
an optically trapped @xmath339 bec was adiabatically transferred from the initial @xmath671 state into the @xmath672 dressed state , with quasimomentum @xmath608 , which corresponds to the lowest energy dressed band at @xmath442 , as identified by the black arrow in the top - left panel to fig .
[ fig_vectorpotential]a . once loaded into this state ,
the atoms were held for a brief equilibration time , at which point both the raman lasers and the trapping lasers were suddenly turned off . after the turn off , the atoms were allowed to travel ballistically along @xmath42 , while a magnetic field gradient stern - gerlach separated the three @xmath476 components along @xmath43 .
this process allowed the direct detection of the spin - momentum superpositions which comprise the raman dressed states .
the top - right panel of fig .
[ fig_vectorpotential]a shows that for @xmath442 the ground state bec is located at @xmath608 , and that the dressed state is made up of three different @xmath476 states each with momentum given by @xmath673 .
the lower two panels of fig .
[ fig_vectorpotential]a depict an experiment where the bec is loaded as described above .
the detuning is slowly ramped from @xmath674 to @xmath675 . during such a ramp , the bec adiabatically remains at the minimum of @xmath676 .
the magnitude of the resulting shift is consequently associated with the artificial vector potential .
the results of many such experiments is shown in fig .
[ fig_vectorpotential]b which compares the measured vector potential . in this case
the raman lasers intersected at @xmath677 degrees .
this has the effect of decreasing the effective recoil momentum to @xmath678 and the recoil momentum to @xmath679 .
the prediction of the model is obtained by numerically solving eq . with no free parameters .
it is important to note that the bec was always at rest in these experiments
. the uniform and static vector potential would therefore normally not be experimentally detectable . in this case , however , the vector potential s difference from the @xmath442 case is encoded into the spin - momentum superposition making up the dressed state , and can be accessed in experiment .
see also ref .
@xcite for more details of this effect . .
the symbols indicate data in which the vector potential was changed from @xmath680 to @xmath681 ( `` + for red and ' ' - " for blue ) , at which point the bec s momentum was measured .
the black crosses ( bottom inset ) reflect data for which @xmath682 , where the bec was allowed to oscillate in the harmonic trap . the amplitude in momentum of these oscillations is plotted .
the grey line is the expected outcome , with slope @xmath683 .
( b ) the in trap oscillation data , fitted to a sinusoidal model .
the blue circles depict data where @xmath684 .
as depicted by the top inset in ( a ) , the measured canonical momentum oscillates around a non - zero value given by @xmath685 .
the black squares depict the average velocity of the three components measured after tof multiplied by the effective mass @xmath686 .
data first appeared in y .- j .
lin , _ et al .
* 7 * , 531 ( 2011 ) , ref .
, width=432 ] with access to an artificial vector potential it is natural to consider what kind of mechanical forces one can induce on the atoms with suitable gradients of this potential .
several possibilities are available .
one can for instance change the vector potential from an initial @xmath687 to a final @xmath688 by suddenly changing @xmath352 .
if @xmath548 behaves as a real vector potential , then this change should be associated with a force that accelerates the atoms , changing their _ mechanical _ momentum by an amount @xmath689 .
there are currently two experiments by lin et al . ,
@xcite which illustrate the effect of an artificial electric field .
the atoms were first prepared at an initial detuning @xmath352 producing a non - zero @xmath687 , and then suddenly making @xmath690 .
this results in @xmath691 which depends on the sign of @xmath352 , see fig .
[ fig_vectorpotential]b . combined with free expansion of the cloud the final momentum can be measured , shown in fig .
[ fig_electricfield]a .
alternatively , @xmath548 can be changed only slightly , as illustrated in fig .
[ fig_electricfield]b . by taking the average velocity of all three spin - momentum components the quasimomentum and
also the group velocity can be monitored . the subsequent evolution in the harmonic confining potential
is then monitored .
[ fig_electricfield]b depicts the mechanical momentum @xmath692 , related to the group velocity , by multiplying the average velocity by the effective mass @xmath693 .
this quantity has the same amplitude as the canonical momentum oscillations , but is centered on zero .
along @xmath43 from a magnetic field gradient , creates an artificial magnetic field .
this causes a shearing of the atomic cloud and allows the entry of vortices into the bec .
( b ) with the gradient present , the vector potential depends on @xmath469 as @xmath694 . when the raman lasers are turned off , @xmath695 .
this process introduces an electric field along @xmath42 that depends on @xmath469 , and as a result the bec undergoes a shearing motion during tof .
( c ) above a critical gradient ( grey band ) it becomes energetically favourable for vortices to enter the bec .
data in ( a ) and ( c ) first appeared in y .- j .
lin , _ et al .
_ nature . *
462 * , 628 ( 2009 ) , ref .
@xcite . , width=432 ] with a synthetic abelian vector potential present , it is also natural to consider creating an artificial magnetic field @xmath696 .
the key ingredient for obtaining a non - zero effective magnetic field is the dressed state , given by a detuning which varies linearly along @xmath43 , with @xmath610 .
this gives @xmath697 when @xmath698 . to create an artificial magnetic field such as in ref .
@xcite , the geometry shown in fig .
[ fig_layout ] can be used , with the three @xmath476 states of @xmath339 s hyperfine ground state zeeman split by a biasing magnetic field along @xmath43 .
the states are coupled with a pair of raman lasers giving a momentum exchange along @xmath42 . using a pair of quadrupole coils in an anti - helmholtz geometry aligned along @xmath52 ,
produces a magnetic field gradient of the form @xmath699 which adds to the bias field @xmath700 .
the resulting zeeman shift is proportional to @xmath701 giving the desired detuning gradient along @xmath43 , which in turn results in a synthetic magnetic field @xmath702 .
figure [ fig_magneticfield]a depicts images of atoms after tof , in which the appearance of an artificial magnetic field is marked in two ways .
firstly , the initially symmetric cloud acquires a shear as @xmath703 increases . with the gradient present
, the vector potential depends on @xmath469 as @xmath694 . when the raman lasers are turned off at the beginning of the time of flight , @xmath695 .
this process introduces an electric field along @xmath42 proportional to @xmath469 , and as a result the bec undergoes a shearing motion during the expansion . above a critical gradient vortices
spontaneously enter into the non - rotating bec . in this regime ,
the bec is described by a macroscopic wave function @xmath704 , which obeys the gross - pitaevskii equation .
the phase @xmath462 winds by @xmath705 around each vortex , with amplitude @xmath706 at the vortex center .
the magnetic flux @xmath707 results in @xmath708 vortices and for an infinite , zero temperature system , the vortices are arrayed in a lattice @xcite with density @xmath709 . for finite systems vortices are energetically less favourable , and their areal density is below this asymptotic value , decreasing to zero at a critical field @xmath710 .
[ sect : soc ] let us now turn to the case of non - abelian geometric gauge potentials . as already mentioned in section [ sect : gaugefields ] , they emerge when the centre of mass motion of atoms takes place in a manifold of degenerate or quasi - degenerate internal dressed states .
the non - abelian gauge potentials provide a coupling between the centre of mass motion and the internal ( spin or quasi - spin ) degrees of freedom , thus causing an effective spin - orbit " coupling . in this section
, we shall overview several schemes for creating these gauge potentials in cold - atom setups .
we refer to sect .
[ non - abelian_wilson ] for a discussion on genuine non - abelian gauge structures .
( with @xmath184 ) are coupled to the excited state @xmath423 via the rabi frequencies @xmath711 of the laser fields . , width=264 ] [ tripod - figure ] let us begin with a tripod setup ( shown in fig .
[ tripod - figure ] ) for the atom - light coupling , comprising three lower atomic levels coupled with an excited level via the laser fields with the raman frequencies @xmath711 . compared to the @xmath420 setup analyzed in the previous section , now there is an additional third laser driving the transitions between an extra ground state @xmath712 and the excited state @xmath674 .
the tripod scheme has two degenerate dark states representing the superpositions of the three ground states immune to the atom - light coupling .
the tripod scheme was initially considered by olshanii and co - authors @xcite in the context of velocity selective coherent population trapping . in another development klaas bergmann and co - workers
theoretically @xcite and experimentally @xcite explored the adiabatic evolution of the dark - state atoms in the time - dependent laser fields ( the tripod - stirap ) , governed by the matrix valued geometric phase @xcite .
ruseckas et al considered the adiabatic motion of the dark - state atoms in the spatially varying laser fields @xcite providing non - abelian gauge potentials for the centre of mass motion using the tripod scheme .
assuming exact resonance for the atom - light coupling , the hamiltonian of the tripod atom reads in the interaction representation @xmath713 where @xmath714 is the total rabi frequency and @xmath715 is the atomic bright state , representing a superposition of the atomic ground states directly coupled to the excited state @xmath423 .
the hamiltonian @xmath77 has two eigenstates @xmath716 and @xmath717 known as the dark ( or uncoupled ) states which are orthogonal to the bright state @xmath718 , contain no excited state contribution and are characterized by zero eigenenergies ( @xmath719 ) .
the bright state @xmath720 is position dependent due to the position dependence of the amplitude and the phase of the laser rabi frequencies @xmath721 .
thus the adiabatic elimination of the bright and excited states leads to the adiabatic centre of mass motion of the dark state atoms affected by the vector potential @xmath106 ( with @xmath145 since there are two dark states ) to be labelled simply by @xmath175 .
the explicit expressions for the vector potential @xmath175 and the accompanying geometric scalar potential are available in the previous original @xcite or review @xcite articles .
if two rabi frequencies @xmath722 represent co - propagating first order laguerre - gaussian beams with opposite vorticity along the propagation axis @xmath411 and the third rabi frequency @xmath723 represent the first order hermite - gaussian beam propagating in @xmath510 direction , the induced vector potential @xmath175 contains a contribution due to a magnetic monopole @xcite times a pauli matrix . on the other hand , if all three beams co - propagate , two of them carrying opposite vortices , the third beams being without the vortex , a persistent current can be generated for ultracold atoms @xcite . by properly choosing the laser fields , the cartesian components of the vector potential @xmath175 do not commute , leading to the non - abelian gauge potentials ( see also sect .
[ non - abelian_wilson ] ) .
we note that this can happen even if the rabi frequencies @xmath711 of the tripod setup represent ( non - collinear ) plane waves @xcite or the properly chosen standing waves @xcite . in that case
, one can produce a uniform vector potential whose cartesian components are proportional to the pauli matrices , thus generating a spin - orbit coupling ( soc ) of the rashba - dresselhaus type for ultracold atoms .
the rapidly progressing field of spin - orbit - coupled atomic gases will be further explored in the following .
spin - orbit coupling ( soc ) is pervasive in material systems . in some cases it leads to parasitic effects such as reduced spin coherence times @xcite , while in other contexts , like topological insulators , it is essential @xcite .
topological insulators non - interacting fermionic systems
represent a first realization of time - reversal ( tr ) invariant systems with topological order @xcite . in analogy with the progression from the tr - violating single - particle integer quantum hall effect ( iqhe ) to the interaction driven fractional quantum hall effects ( fqhes ) ,
the next important step is realizing strongly interacting cousins to the topological insulators , of which topological superconductors are a first example @xcite .
ultracold atoms are an ideal platform to study strongly interacting soc systems , both with bosonic @xcite and fermionic atoms @xcite . since ultracold atoms lack intrinsic soc , numerous techniques ( including the tripod setup considered above ) have been suggested for generating soc ( generally equivalent to non - abelian gauge potentials @xcite ) , with optical @xcite , rf @xcite or pulsed magnetic @xcite fields .
a first step towards this goal has been experimentally achieved with the recent implementation of an abelian soc ( i.e. a soc term generated along one spatial direction only ) in several laboratories @xcite , see sec . [ sub : exp - soc ] . in sect .
[ sect : gaugefields ] , we considered how non - trivial artificial gauge fields can appear for atoms adiabatically moving in a restricted set of `` target '' states .
then in sect .
[ sect : schemes ] , we studied an experimentally relevant case where the artificial gauge field @xmath548 was a vector of real numbers , i.e. a gauge potential related to an abelian ( u(1 ) ) gauge structure . here
we will focus on the most simple extension where the vector potential is non - abelian , but is spatially uniform , such as the one generated using the tripod setup mentioned above .
because of the @xmath724 term appearing in the expression ( [ eq : b ] ) for the non - abelian magnetic field , a uniform non - abelian vector potential can have measurable effects @xcite .
soc can be a simple example of non - abelian vector potentials : suppose @xmath725 is a spatially uniform vector potential with non - commuting elements each described by @xmath726 matrices acting on a pseudospin - degree of freedom .
we can expand the general hamiltonian @xmath727^{2}+v(\hat{{\bf r}})\\ & = \frac{\hbar^{2}\mathbf{k}^{2}}{2m}-\frac{\hbar}{m}\left[\hat{\mathbfcal a}_{x}k_{x}+\hat{\mathbfcal a}_{y}k_{y}+\hat{\mathbfcal a}_{z}k_{z}\right]+\frac{1}{2m}\hat{\mathbfcal a}\cdot\hat{\mathbfcal a}+v({\bf r}),\end{aligned}\ ] ] where each of the terms @xmath728 couple the spin to the atom s linear motion .
in general , because @xmath729\neq0 $ ] for some @xmath730 , this uniform vector potential can not be eliminated by a gauge transformation . as we argued in sect .
[ sect : gaugefields ] , non - trivial gauge fields require that the atomic motion be restricted to a target subspace of the initial hamiltonian . in the present case
this implies that the full hamiltonian be spanned by a minimum of three internal states . the tripod scheme involving four laser - coupled levels is the most commonly used theoretical model in which non - abelian gauge fields appear . here
we will present a slightly simpler `` ring coupling '' model containing only three sequentially coupled levels @xcite which displays the same physics ( interestingly , the familiar tripod scheme @xcite reduces to the three level ring model when the excited state is far from resonance and thus can be adiabatically eliminated . )
we consider @xmath712 ground or metastable atomic `` spin '' states @xmath731 coupled together with complex valued matrix elements @xmath732 $ ] , linking each state to each other state . here
, @xmath464 describes the optical coupling strength , @xmath733 is the respective momentum acquired in the @xmath734 atomic transition . throughout this section ,
spin indices are taken @xmath735 , implying periodic boundary conditions @xmath736 for spin states . including the motional degrees of freedom
, the momentum representation second - quantized hamiltonian @xmath737\bigg\}\label{eq : main}\end{aligned}\ ] ] describes a system of 2d atoms in the momentum representation , where all summations over @xmath140 range from @xmath440 to @xmath712 . here , @xmath738 is the spinor field operator describing the creation of a particle with momentum @xmath739 in internal state @xmath740 . in what follows ,
we require that @xmath741 , so that no momentum is transferred to an atom upon completing a closed - loop transition @xmath742 . in this case
, the momenta - exchange can be represented in terms of the differences @xmath743 , and we require @xmath744 to have zero average .
the displacement vectors @xmath745 define these transformations explicitly . in the spirit of sects .
[ subsect : twolevel ] and [ subsect : threelevel ] , we substitute @xmath746 into the hamiltonian in eq .
( [ eq : main ] ) which separates into an integral @xmath747 over @xmath748 blocks @xmath749 where each block is labeled by the quasi - momentum @xmath750 .
we can relate the coupling term in eq .
( [ eqndisplaced ] ) to the situation of a 1d periodic tight binding hamiltonian with a hopping matrix element @xmath751 , and where the three sites correspond to the three internal states of the atom . for @xmath464
much larger than the kinetic energy the coupling term can be diagonalised using a basis conjugate to the spin - index @xmath140 with field operators @xmath752 the corresponding eigenenergies are given by @xmath753 , where @xmath754 is analogous to the usual crystal momentum .
the ground state is two - fold degenerate for @xmath755 for states at @xmath756 and @xmath757 .
if the displacement vectors @xmath744 are chosen such that @xmath758 , the full hamiltonian matrix becomes @xmath759,\label{dressedfullhamiltonian}\end{aligned}\ ] ] where momentum is expressed in units of @xmath760 and energy in units of recoil energy @xmath761 .
we will in the following consider the two nearly degenerate states with @xmath756 and @xmath757 , yielding the pseudospins @xmath762 and @xmath763 . in the subspace spanned by the lowest energy pair of dressed states ,
we consequently get a zeroth order hamiltonian of the rashba form @xmath764 where @xmath765 are the standard pauli matrices . for finite coupling @xmath464
it is instructive to adiabatically eliminate the excited states order - by - order in perturbation theory , which gives additional effective terms @xmath766 @xcite .
the key to the experimental realization of spin - orbit coupling , is in the fact that the laser geometry in eq .
can also be used to obtain a coupling of the form , @xmath767 where @xmath516 was chosen , together with a pseudo - spin rotation @xmath768 .
the hamiltonian in eq .
( [ eq : onedimensionsoc ] ) describes a spin - orbit coupled system consisting of an equal sum of rashba and dresselhaus terms .
this corresponds to the abelian gauge potential @xmath769 containing a single cartesian component .
the situation is non - trivial due an additional zeeman term @xmath770 , and leads to a number of interesting single- and many - body effects recently studied both experimentally @xcite and theoretically@xcite . , where the transition between the double minimum and the single minima configurations are illustrated .
( b ) the quasi momentum obtained by adiabatically loading a spin mixture into a raman dressed state with the desired @xmath503 .
data first appeared in y .- j .
lin , _ et al .
* 471 * , 83 ( 2011 ) , ref .
, width=453 ] in fig .
[ fig_soc]a it is shown how the spin - orbit coupling dispersion , as a function of the laser coupling strength , is dramatically different from that of a free particle . for small @xmath503 ,
the lowest dressed band consists of a double - well in momentum space @xcite , with two distinct momenta where the group velocity is zero .
the states near the two minima are the dressed spin states , @xmath771 and @xmath772 . as @xmath464 increases , the two dressed spin states merge . in this limit
the picture corresponds to spinless bosons with a tunable dispersion relation@xcite . in the experiment by lin _
_ @xcite the effective two - level system was created in @xmath339 s @xmath341 manifold where the @xmath773 and @xmath774 states were isolated
. this could be achieved by using sufficiently large magnetic field ( @xmath775 ) such that the quadratic zeeman shift detuned the @xmath776 state appreciably from the raman resonance .
figure [ fig_soc]a shows the quasimomentum measured by first preparing an equal mixture with @xmath777 , and then slowly increasing @xmath503 . in this process , each dressed spin state adiabatically follows its associated minima . the resulting momentum can be measured using time of flight , which will reveal the presence of a non - trivial spin - orbit coupling in the system . up to now
only a 1d spin - orbit coupling corresponding abelian gauge potential has been generated @xcite in spite of numerous proposal for creating the 2d @xcite and 3d @xcite rashba type spin - orbit coupling ( soc ) in cold atomic gases .
implementation of the 2d or 3d rashba soc would allow for the study of rich ground state physics proposed in systems of many - body fermions @xcite and bosons @xcite , of which many properties have no solid - state physics analogue .
this will be considered in the next section .
[ sect : interactions ] synthetic magnetic fields and spin - orbit couplings result from engineering the single particle hamiltonian , however , ultracold gases can be strongly influenced by collisions .
absent artificial gauge fields , these systems are a unique playground where the inter - particle interaction strength can be tuned virtually at will @xcite . in this section ,
we describe how interacting becs and degenerate fermi gases behave in the presence of an external gauge potential , focusing on the case where atoms are subjected to a synthetic soc .
the interplay between interactions and synthetic magnetic fields induced by rotation were described in the review by cooper @xcite ; for a more complete overview on interacting spin - orbit - coupled atomic gases , consider the reviews by h. zhai @xcite and x. zhou et al @xcite , as well as other articles recently published in the special issue on non - abelian gauge fields edited by gerbier et al .
@xcite . in a weakly interacting bose gas where @xmath778 ( @xmath779 is the @xmath780-wave scattering length and @xmath781 is the atom density ) , only two - body scattering processes are relevant .
the interaction between atoms can be described by the contact potential v_int(*r*-*r*)=(*r*-*r*)= g ( * r*-*r * ) . in most cases ,
the dynamics is accurately described by the mean - field gross - pitaevskii equation @xcite i(r , t)=(-^2+v(r)+g|(r , t)|^2 ) ( r , t ) , a nonlinear equation describing the condensate wavefunction @xmath782 , with a potential @xmath783 and a nonlinearity proportional to the density @xmath784 .
interactions captured by this nonlinearity change the dynamics dramatically compared to the non - interacting gas , as evidenced by the elementary excitation spectra@xcite and the nucleation of vortices @xcite . in the context of artificial gauge potentials
we ask : what effects are added for interacting gases with soc ?
this scenario has been studied by a number of authors @xcite , showing that soc dramatically alters the ground state properties . in such situations ,
the general second - quantized hamiltonian is h = d*r * ^= h_0 + h _ , [ so - gp1 ] where the soc is described by the gauge potential @xmath725 ; @xmath785 is a multicomponent field operator ; and @xmath786 is the confining potential with frequency @xmath185 .
the exact form of the interaction term @xmath787 depends on the details of the physical setup @xcite . typically , the field operator @xmath785 associated with a dressed state
a superposition of the bare atomic levels and the two - body scattering lengths entering @xmath788 depend on the different bare states . in this dressed - state picture , this can yield spin - dependent collision terms , e.g. , @xmath789 with @xmath790 , independent of the gauge potential @xmath725 , see refs . . furthermore ( and beyond the scope of the discussion here )
the atomic contact interaction can acquire momentum dependence a finite range from the momentum - dependence of the dressed states as expressed in the basis of bare atomic states @xcite . here
, we review the ground - state properties of a quasi - spin-1/2 system with isotropic 2d rashba soc , where @xmath791 , and where @xmath725 is simply expressed in terms of the pauli matrices as @xmath792 .
the synthetic gauge potentials giving rise to spin - orbit coupling are remarkably flexible in the sense that the precise shape of the gauge potential can often be controlled at will .
the rashba type coupling is one example , which is also often encountered in solid state scenarios .
such spin - orbit couplings can be achieved optically @xcite , but also using periodically - driven cold - atom systems , such as those based on pulsed - magnetic fields @xcite , see also section 8 . in this configuration , the single - particle energy spectrum of the homogeneous ( @xmath793 ) system has the two energy branches @xmath794 illustrated in fig .
[ figurebandso ] .
the main effect of the rashba coupling @xmath725 is to replace the unique condensation point @xmath795 by a continuum of minima located on the ring @xmath796 , see fig .
[ figurebandso ] .
these degenerate single - particle ground states affect the many - body problem , where an interacting condensate s ground state will be chosen through a complex competition between the interactions and the parameters of the non - interacting model ( e.g. the trap frequency @xmath185 and the rashba coupling strength @xmath63 ) .
hence , in the presence of rashba soc , interactions play a major role even when they are weak @xcite .
dispersion for a homogeneous gas with rashba soc : @xmath797 .
@xmath798 denotes momentum in the @xmath42-@xmath43 plane . when @xmath799 , the lowest band is minimal on the ring @xmath800 . ]
the mean - field phase diagram of the homogenous , @xmath801 , spin - orbit - coupled bec is already altered by a simple interaction term of the form @xcite _ int = d*r * g n_1 ^2 + g n_2 ^2 + g_12 n_1
n_2 , [ simpleterm ] where the two scattering parameters @xmath802 and @xmath803 account for spin - dependent collisions in a minimal manner , and where @xmath804 denote the density operators . the mean - field phase diagram described below has been obtained by wang _
@xcite , through a minimization of the gross - pitaevskii free energy for the two - component order parameter @xmath805 , derived from the soc hamiltonian - . when @xmath806 , the densities associated with the two spin components @xmath807 are uniform in space , but the phase of the condensate @xmath805 varies periodically along one direction , spontaneously breaking rotational symmetry .
this regime characterizes the `` plane wave phase , '' where the macroscopic wave function of the condensate takes the simple form @xcite ( r)= e^i x ( 1 , 1)^t[planewavecondensage ] . without loss of generality
, we have taken the plane - wave to be along @xmath42 , reflecting the spontaneous symmetry breaking . when @xmath808 , the condensate @xmath809 is described by a superposition of two such plane waves , ( r)= = ( c(x ) + i(x ) ) , [ stripecond ] yielding a periodic spatial modulation of the spin - density @xmath810 this phase is known as the `` stripe superfluid '' or `` standing wave phase '' @xcite .
the mean - field phase diagram of the homogeneous system , obtained from the gross - pitaevskii equation derived from eqs .
- is robust against deformation of the soc term @xmath725 ( without loosing the cylindrical symmetry ) , and generally holds in the presence of additional fields , e.g. , zeeman terms @xcite .
the phase diagram is enriched by the harmonic confinement @xmath786 generally present in cold - atom experiments @xcite .
remarkably , novel phases emerge in trapped systems @xcite , even under the strong ( and generally unphysical ) assumption that all the interaction processes are described by a single scattering parameter , i.e. , when the interaction term reduces to the spin - independent form @xmath811 , where @xmath812 is an effective scattering length describing all collisions @xcite .
this strong simplification aims to isolate the main effects introduced by the trap in spin - orbit coupled gases . into the problem , in the spirit of eq .
, which could lead to even more involved phase diagrams than the one presented here .
] the external trap introduces an additional energy scale @xmath813 into the homogenous problem discussed above , hence it is natural to describe the resulting phase diagram in terms of the dimensionless parameters @xmath814 and @xmath815 , where @xmath816 . in the non - interacting regime where soc dominates ( @xmath817 , @xmath818 )
, the upper branch @xmath819 can be neglected , and , going to the momentum representation , one obtains the single particle wavefunction @xmath820 , where _ -(k)= ( c 1 + ) =( c 1 + e^i ) is an eigenstate of the homogeneous system in the lowest branch @xmath821 ; and @xmath822 denotes the polar angle of @xmath798 . adding the confining potential to the single - particle hamiltonian , @xmath823 , and solving the corresponding equation using the ansatz @xmath824 , yields the eigenenergies @xcite e_nl=+n+ , [ so - trap ] where @xmath825 labels the radial excitations around the minima of the mexican - hat potential @xmath826 ( see fig .
[ figurebandso ] ) and @xmath827 . the degenerate ground states _
l=\{0,-1}(k)e^-(k-)^2 _0 ^ 2/2+i l * u*_- ( ) have quantum numbers @xmath828 and @xmath829 .
in real space , and using polar coordinates @xmath830 , the two degenerate ground states become @xmath831 these half - vortex states @xcite appear in various contexts , including topological quantum computing @xcite , superfluid @xmath832he @xcite and triplet superconductors . the degenerate states in eq . and
any linear combination thereof , yield the rotationally symmetric density distribution illustrated in fig .
[ luis1 ] ( a ) . as pointed out in ref .
, this half - vortex regime ( hv 1/2 ) survives for finite interaction strength @xmath833 , where @xmath834 is related to the energy difference between the ground ( @xmath835 ) and higher energy ( @xmath836 ) states .
indeed , for large soc strength @xmath837 , the energy difference between successive angular excitations [ eq . ] is small compared to radial excitations .
thus , the population of higher angular momentum states ( @xmath838 ) should be favoured at greater , but still reasonably small , interaction strength @xmath839@xcite . a condensate formed with higher angular momentum states satisfying @xmath836 , called the ( hv 3/2)phase , occurs in a certain range of the interaction strength @xmath840 , and has the ring - shaped density distribution depicted in fig .
[ luis1 ] ( b ) .
for larger interaction strength , @xmath841 , the gas enters a `` lattice phase '' , where its density is modulated in the form of a hexagonal lattice , see fig .
[ luis1 ] ( c ) .
interestingly , rotational symmetry is broken and the lattice period is independent of the interaction strength @xmath842 .
in fact , the distance between the lattice minima is solely dictated by the soc strength @xmath843 .
consequently , the resulting momentum distribution has a robust ring - like structure with six peaks , see fig .
[ luis1 ] ( d ) .
the three phases discussed above all have in common that their densities satisfy ( continuous or discrete ) rotational symmetry .
this is no longer the case for large interaction strength ( @xmath844 ) . in this
highly "- interacting regime , the spin - orbit coupled cloud enters a `` striped phase , '' where the density is sinusoidally modulated in a specific direction , see figs .
[ luis1 ] ( e)-(f ) .
this anisotropic state shares similarities with the `` stripe phase '' in the homogeneous system with spin - dependent interactions @xmath845 ( see eq . and sect . [ homogso ] ) .
mapping out the full phase diagram for the trapped spin - orbit coupled gas , by scanning the dimensionless parameters @xmath842 and @xmath843 , reveals a surprisingly intricate picture , see fig .
[ luis3 ] . for readers further interested in the effects of finite temperature , anisotropic rashba coupling @xmath846 and beyond - mean - field descriptions
, we refer to ref .
@xcite , and references therein .
finally , the interplay between soc and dipolar interactions , leading to diverse spin and crystalline structures , have been explored in refs .
@xcite .
trapped spin - orbit coupled gas .
( a)(c),(e ) : the ground state density @xmath847 , for increasing interaction strengths @xmath848 .
the dimensionless soc strength is @xmath849 .
( d ) the momentum distribution for @xmath850 .
( f ) density @xmath851 of a single component for @xmath852 . reproduced with permission from s. sinha , r. nath and l. santos , phys .
* 107 * , 270401 ( 2011 ) . ]
ground state phase diagram of the harmonically trapped spin - orbit coupled bose gas .
reproduced with permission from s. sinha , r. nath and l. santos , phys .
lett . * 107 * , 270401 ( 2011 ) . ]
degenerate atomic fermi gases are a remarkably rich platform for studying strongly - correlated phases common in condensed - matter systems @xcite .
today , the atomic systems are widely investigated , both theoretically and experimentally , to deepen our understanding of ( high-@xmath853 ) superconductivity and topological phases ( see also sect . [
sect : simulation ] ) . degenerate fermi gases offer unique advantages compared to their solid - state brethren .
the interactions between fermions can be tuned at will through feshbach resonances @xcite , leading to rich phase diagrams that describe the evolution from strongly repulsive to strongly attractive fermi systems . for two - component ( spin-1/2 )
fermions interacting through @xmath780-wave collisions , the system evolves from a bcs phase of cooper pairs ( a fermonic superfluid ) when the scattering length is small and negative @xmath854 , to a bec of tightly - bound molecules ( a bosonic superfluid ) when @xmath855 .
the crossover between the bcs and the bec phases @xcite takes place as @xmath856 travels through zero ( i.e. the unitary limit @xmath857 ) , the threshold for the formation of bound molecular states in the two body problem @xcite . here
we discuss how these phase transitions are modified by the inclusion of synthetic soc .
this problem was already considered in solid - state physics , where it was shown that rashba soc significantly alters the physical properties of superconductors displaying strong electric fields near their surface @xcite . for non - zero soc ,
the cooper - pair wavefunction is a mixture of singlet and triplet components , which results in a non - trivial and anisotropic spin susceptibility tensor that reflects the response of the system to an external magnetic field .
+ vyasanakere and shenoy @xcite first demonstrated that the inclusion of a soc term drastically modifies the properties of the interacting fermi gas , especially on the bcs side of the resonance ( where the scattering length is negative , @xmath858 ) . in the presence of a non - abelian gauge potential ,
two - body bound - states are present _ even in the bcs side _ @xmath858 of the resonance . in other words ,
the soc shifts the threshold for the formation of bound - states towards finite negative values @xmath858 .
moreover , for certain configurations of the gauge potential ( e.g. sufficiently large soc strength ) , bound states are formed for _ any _ value of the scattering length .
these soc - induced bound states , referred to as rashbons " , are described by a two - body wavefunction featuring singlet and triplet components , similarly to the cooper - pair wave function in the presence of soc @xcite .
the rashbons lead to superfluids with nematic spin structures at low temperature , similar to superfluid @xmath832he @xcite . since bound states are allowed within the bcs regime and the binding energy @xmath859 strongly
depends on the gauge potential s strength , a subsequent work by vyasanakere , zhang , and shenoy @xcite showed that a bcs - bec crossover could be induced by varying the rashba spin - orbit strength while keeping the scattering length @xmath854 _ fixed _ within the bcs regime .
this rashba - induced transition , from the bcs state to the molecular bec formed by rashbons , is accompanied by a topological modification of the fermi surface @xcite .
the general structure of the two - body wavefunction , with singlet and triplet components , is maintained during the transition : as the rashba coupling is increased , the cooper - pair wave function converges towards the rashbon wave function , suggesting that the crossover nature of the transition is preserved @xcite .
the crossover has been further established , by confirming that the ground state is indeed protected by a gap throughout the transition @xcite .
a complete description of the rashba - induced bcs - bec crossover , in terms of various gauge potential configurations , is presented in ref .
( see also ref .
@xcite ) . surprisingly , at zero temperature , the condensed @xmath860 and superfluid @xmath861 fractions behave oppositly as the rashba coupling strength @xmath63 is increased , see ref . .
the condensate fraction increases monotonically with @xmath63 , eventually reaching @xmath862 .
this enhanced condensate formation , even present for @xmath858 ( bcs side ) , is compatible with the formation of the rashbons discussed above .
in contrast , the superfluid fraction @xmath861 is generally _ reduced _ by rashba soc . depending on the value of the scattering length @xmath564 , the fraction @xmath863 decreases monotonically or non - monotonically with increasing rashba soc .
superfluidity is never totally suppressed by rashba soc @xcite , furthermore , on the bcs side of resonance , the superfluid transition temperature @xmath853 as well as the pairing gap @xmath425 increase as a function of the rashba coupling strength @xcite .
these two quantities respectively converge towards the significantly higher molecular - bec superfluid temperature @xmath864 and the two - body binding energy @xmath865 .
l. he and x .-
huang @xcite investigated the rashba - induced bcs - bec crossover in two dimensions , where the berezinskii - kosterlitz - thouless transition temperature is also enhanced by the rashba coupling .
similar properties have been studied for fermi gases trapped in optical lattices , where the interplay between interactions and the pseudo - relativistic ( dirac ) spectrum generated by the rashba coupling leads to interesting behavior@xcite . taken together , these results highlight the major deviations from standard superconductivity , which is offered by the large and tunable rashba coupling possible in atomic systems
. methods for probing the unusual properties of atomic bcs superfluids in the presence of rashba coupling and rashbon condensates ( i.e. molecular bec induced by rashba coupling in the bcs limit @xmath854 ) have been proposed in ref .
, based on bragg spectroscopy and density profiles .
other effects related to the presence of rashba ( or equal rashba - dresselhaus ) soc during the bcs - bec crossover could also be measured through the isothermal compressibility and the spin susceptibility @xcite .
+ in most studies , the scattering length @xmath564 was chosen to be independent of the synthetic soc , and thus , it is treated as a constant along the rashba - induced bcs - bec crossover . refs .
@xcite describe how interactions are modified in the presence of strong rashba coupling , offering a relation between physical scattering lengths and mean - field interactions .
+ another important topic concerns the realization of topologically - ordered superfluids using fermi gases subjected to rashba soc @xcite .
this line of research is motivated by the possibility to create and control non - abelian excitations such as majorana fermions in a versatile setup , with possible application to topological computation ( see ref . and sect .
[ sect : simulation ] ) . to achieve this goal in a fermi gas ,
the following basic ingredients are required : ( a ) a rashba soc @xmath866 ( or in 1d , any form of soc ) , ( b ) a zeeman term @xmath867 and ( c ) @xmath780-wave interactions with scattering length @xmath564 @xcite .
based on this minimal and realistic system , several authors have investigated the rich interplay between the rashba - induced bcs - bec crossover , the population imbalance produced by the zeeman term , and the emergence of topological order , which could be observed by varying the control parameters @xmath868 @xcite .
these simple systems present a new path for exploring the fflo finite - momentum paired states @xcite .
finally , we note that finite - momentum ground - states can be generated in rashba - coupled fermi gases also in the _ absence _ of a zeeman splitting term , provided that the collisional interaction is strong enough @xcite .
in the previous sections , we have considered the creation of artificial gauge fields for ultracold atoms moving in continuous space ( i.e. typically , atoms in a harmonic trap ) . here
, we discuss the emergence of lattice gauge structures ( e.g. artificial magnetic fluxes ) , which can be realized by taking advantage of the discrete motion of atoms in optical lattices .
we begin with a brief overview of lattice gauge structures and lattice models , and we then describe their experimental implementation with optical lattices . in theoretical physics , lattice models
are generally developed for two different reasons .
the first motivation for discretizing configuration space is to apply analytical or numerical tools that are unavailable for continuum systems .
the results obtained from these lattice descriptions are relevant for large systems with vanishingly small lattice periods .
a prime example of this approach are lattice gauge theories , which deepen our understanding of quantum electrodynamics ( qed ) and quantum chromodynamics ( qcd ) @xcite . in contrast , many systems have crystalline order and are therefore quite naturally described by lattice models .
this situation frequently arises in condensed - matter systems , where electrons move about in the periodic potential formed by an orderly array of ions @xcite which together comprise a crystalline material . here
, the lattice is physical : the lattice spacing is finite and the hopping parameters are determined by the material under scrutiny . when electrons at the fermi - energy are tightly bound to the ionic potential , the system is well described by the fermi - hubbard hamiltonian @xmath869 a single - band tight - binding model , in which @xmath870 indicates that the summation is over neighboring lattice sites . here , @xmath871 and @xmath872 describe the tunneling amplitude and on - site interactions , respectively ; and @xmath873 describes the creation of a fermion at lattice site @xmath140 with spin @xmath874 .
the hubbard model , and its generalizations , have shed light on numerous physical phenomena such as metal - insulator transitions @xcite , high-@xmath853 superconductivity @xcite , transport properties in graphene @xcite , and recently topological insulating phases @xcite .
the emulation of lattice models , using quantum analog simulators @xcite , makes possible the exploration of effects common in condensed - matter and high - energy systems .
ultracold atoms trapped in optical lattices can nearly perfectly realize the hubbard model @xcite in the laboratory @xcite .
spinless bosonic or fermionic atoms are described by the general hamiltonian @xmath875 \hat \psi ( \bs r ) \nonumber \\ & + g \int d\bs{x } \
\hat \psi^{\dagger}(\bs r ) \hat \psi^{\dagger}(\bs r ) \hat \psi ( \bs r ) \hat \psi ( \bs r ) , \label{hamatoms}\end{aligned}\ ] ] where @xmath876 describes the creation of an atom at position @xmath544 ; @xmath877 is a periodic optical potential ; @xmath878 is an external confining potential ; and @xmath802 characterizes the interatomic interaction . as a consequence of @xmath877 s lattice structure , it is convenient to express the field operator @xmath879 in terms a sum over wannier orbitals @xmath880 at site @xmath881 and band @xmath626 , which together constitute a set of orthogonal functions @xcite . for a sufficiently deep lattice potential ,
this expansion can be restricted to the lowest band , namely @xmath882 , in which case the general hamiltonian reduces to the tight - binding hamiltonian @xmath883 a spinless bosonic hubbard hamiltonian , where @xmath884 is the on - site interaction strength and @xmath885 is the energy offset @xcite .
the atomic hubbard hamiltonian features the tunneling amplitudes @xmath886 w(\bs r_k)$ ] between the nearest - neighboring sites @xmath887 , which are typically much larger than the tunneling matrix elements between more distant sites @xcite .
spinful hubbard models can be implemented using fermionic or bosonic atoms with two or more internal states @xcite .
experimental realization of hubbard type systems with ultracold atoms in optical lattices has produced important insight , starting with the experimental observation of quantum phase transitions : superfluid to mott insulator for bosons @xcite and fermi - liquid to mott insulator for fermions @xcite .
in addition , different lattice topologies give rise to pseudo - relativistic dirac fermions @xcite , and systems far from equilibrium can be realized @xcite . today
, significant experimental effort is being devoted to realizing strong synthetic magnetic fields and to creating new kinds of spin - orbit couplings in optical lattices @xcite .
such systems are well suited for studying quantum hall states , topological insulators and superconductors ( see section [ sect : simulation ] ) .
optical lattices penetrated by staggered magnetic fluxes have also been realized with a view to studying frustrated magnetism @xcite . in this section ,
we focus on different experimental methods for creating synthetic gauge fields in optical lattices .
first , we recall how general gauge structures are defined in the lattice framework , mainly discussing the case of a uniform magnetic field ( _ i.e. _ , the hofstadter model @xcite ) .
we then comment on generalizations of this concept to other families of lattice gauge structures , such as non - abelian gauge fields .
finally , we present methods for implementing the hofstadter model and its generalization in the lab . here
, we review the physics of quantum particles constrained to evolve on a lattice and subjected to classical ( static ) gauge fields : firstly , focusing on abelian gauge potentials , i.e. magnetic fields , in two - dimensional lattices , and then moving to non - abelian gauge fields . consider a particle subject to a gauge potential @xmath888 ( where the coupling constant @xmath581 would be the electric charge in conventional electromagnetism ) confined to the @xmath153 plane @xmath889 , described by the single - particle hamiltonian h= ^2,[gaugecont ] when the particle follows a path from a reference point @xmath140 to a point @xmath890 ,
whose coordinates are denoted @xmath881 and @xmath891 , respectively , it acquires a `` magnetic phase factor '' @xcite ( r_k)= ( _ j^k l ) _ 0 ( r_k)=u_jk _ 0 ( r_k),[linkvardef ] where @xmath892 denotes the wavefunction in the absence of the gauge potential , as in fig .
[ fig_gaugelattice ] ( a ) . here , we introduced the _ link variable _
@xcite u_jk= ( _ j^k l ) = e^i _ jk = ( u_kj)^ * ( 1),[linkvardef2 ] embodying the effect of the gauge potential on the fictitious link connecting the points @xmath140 and @xmath890 .
from the perspective of fibre - bundle theory , the link variables define the parallel transport on a principle fibre bundle @xmath893 , where the connection @xmath894 is determined by the gauge potential @xmath895 ( cf . ref . ) .
the link variables @xmath896 are generally called `` peierls phases '' in condensed - matter physics ( see sect . [ sectionmodels ] ) , where they naturally appear in the description of solids subjected to magnetic fields @xcite . under gauge transformations ,
the link variables change according to @xmath897 .
\label{transf}\end{aligned}\ ] ] to develop lattice models , we introduce the notion of a _ plaquette _ , a closed region in space delimited by a set of points @xmath898 , connected by links .
when the particle performs a loop @xmath899 around such a plaquette , it acquires an aharonov - bohm phase @xcite ( r_1 ) ( _ l ) ( r_1)= ( 2 i _ ) ( r_1 ) , where @xmath900 is the number of magnetic flux quanta @xmath901 penetrating the plaquette @xmath899 ; and @xmath902 is the magnetic field associated with the gauge potential @xmath903 , see fig .
[ fig_gaugelattice]bc .
this is the well - known aharonov - bohm effect @xcite , which can also be expressed in terms of the link variables @xmath896 defined in eq . as e^i 2 _
= u_12 u_23 u_34
u_l-1 l u_l1=_u_jk=(i _ _ jk).[fluxdef ] using eq .
, we see that the magnetic flux obtained through the loop " product is a gauge - invariant quantity associated with the link variables e^i 2 _=_u_jk= _ = 1 _ u_jk = e^i 2 _ .
[ invariance ] here , @xmath904 is physically equivalent to @xmath905 for @xmath906 , since @xmath907 . + .
( a ) when a particle follows a path @xmath908 , from a reference point @xmath909 to a point @xmath910 , it acquires a magnetic phase factor " @xmath911 determined by the gauge potential @xmath895 .
( b ) when a particle performs a loop @xmath899 , it realizes the aharonov - bohm effect .
the wave function acquires a berry phase proportional to the number of magnetic flux quanta @xmath912 penetrating the region enclosed by the loop .
the loop @xmath899 defines a u(1 ) transformation , called holonomy @xcite .
( c ) gauge structures on the lattice : the links @xmath913 are all associated with link variables @xmath911 , which are directly related to the magnetic flux @xmath912 penetrating the plaquettes , see eq . .
, width=480 ] these general considerations suggest that the lattice description of a quantum system subject to a gauge potential @xmath895 should include a. a set of lattice sites @xmath914 , b. a set of links @xmath915 connecting the sites and defining plaquettes @xmath899 , c. a set of link variables @xmath916 associated with the links @xmath915 , see fig .
[ fig_gaugelattice ] ( c ) . here ,
the information in @xmath917 is fully contained in the link variables @xmath916 .
the physical gauge - invariant quantity is the magnetic flux penetrating each plaquette @xmath912 , which can be derived from the link variables [ see eq . , and fig .
[ fig_gaugelattice]c ] .
the gauge - invariance relation is interesting from a quantum - simulation point of view , as it defines the simplest set of link variables @xmath916 for any magnetic flux configuration .
the generalization of these concepts to higher dimensions is straightforward .
@xmath918 is the largest magnetic field that can routinely be applied to materials ; because the typical lattice spacing is @xmath919 , the flux per elementary plaquette @xmath920 is tiny . at higher fields , with @xmath921 ( requiring a fantastically large magnetic field @xmath922 ) ,
the interplay between magnetic and lattice structures is captured by the hofstadter model discussed in sect .
[ hofsection ] .
in contrast , synthetic magnetic fields for cold atoms trapped in optical lattices ( sect . [ experimentallattice ] ) can be engineered in the high - flux regime @xmath921 .
the proceeding discussion focused on abelian gauge potentials and magnetic fields , where the link variables @xmath923 are phase factors that do not couple to internal degrees of freedom .
these concepts can be generalized to a wide family of multi - component systems with electronic spin , color , flavor or atomic spin degrees of freedom . in the following
, we will generically refer to these as `` spin '' degrees of freedom .
in this context , gauge potentials and link variables may act differently on the different spin states , and they are therefore matrix - valued objects ( see refs . ) .
the link variables belong to a lie group , e.g. @xmath924 or @xmath925 , while the non - abelian gauge potential is an element of the corresponding lie algebra , e.g. @xmath926 or @xmath927 .
a formal generalization of eq . defines the link variable u_jk= \ { } , [ linkvardef3 ] in terms of the non - abelian gauge potential @xmath928 ( the coupling constant is again subsumed into the definition of @xmath929 ) .
the path - ordered integral @xmath930 is required because the matrices @xmath931 at different points of the path do not necessarily commute @xcite .
the link variable @xmath932 and the non - abelian gauge structures are clearly connected in the continuum limit , where the sites @xmath933 are sufficiently close to each other , i.e. @xmath934 and @xmath935 , where @xmath936 , @xmath937 is the unit vector along the @xmath45 direction and @xmath938 . in this case
, the link variables are @xcite @xmath939 , \label{defna } \\ & \hat u_{\bs r + \epsilon { \bf e}_{\mu } , \bs r } = \hat u_{-\mu } ( \bs r + { \bf e}_{\mu})=\exp\left[-i \frac{\epsilon}{\hbar } \hat{\mathcal a}_{\mu } ( \bs r ) \right ] , \nonumber\end{aligned}\ ] ] which lead to the loop " product around a unit square plaquette , @xmath940.\label{naloop}\end{aligned}\ ] ] equations - define the yang - mills field strength @xcite ( analogous to the antisymmetric tensor @xmath941 defined by eq . for the geometric gauge potentials ) _
( r)= _ _ ( r ) - _ _ ( r ) + [ _ ( r ) , _ ( r)],[fsdef ] to leading order in @xmath650 ( cf . refs . and for derivations ) . in general , the commutator , @xmath942\ne0 $ ] , which is the hallmark of non - abelian gauge theories @xcite . the loop product
thus generalizes eq . to the non - abelian case
this result connects the link variables , defined in the lattice context , and the field strength which plays a fundamental role in continuum theories . as for the abelian case
, a lattice system with a non - abelian gauge structure can be entirely described by ( a ) the lattice topology : the sites coordinates , and the links , delimiting the unit plaquettes ; and ( b ) the link variables @xmath943 associated with the links . under local gauge transformations , the link variables @xmath944 transform according to @xcite u_jk u_jk= t_j u_jk t_k^ , where @xmath945 is a local transformation ( a rotation in spin space at the lattice site @xmath891 ) .
therefore , the loop product around a unit plaquette , delimited by the lattice sites @xmath946 , u_= _
u_jk = u_12 u_23 u_34
u_l-1 l u_l1,[loop_operator_definition ] is not in general gauge invariant @xcite , since u_u_= t_1 u_t_1^.[loop_not_invariant ] this is related to the fact that the field strength @xmath947 , or curvature @xcite , of the continuum theory is not gauge - invariant .
we therefore introduce the wilson loop @xcite , a gauge - invariant quantity associated to each plaquette @xmath899 , w()=tr ( u _ ) , [ wilsonloop ] where @xmath948 indicates the trace of the @xmath949 matrix .
the gauge invariant wilson loop generalizes the notion of flux per plaquette @xmath912 to non - abelian gauge fields .
genuine non - abelian properties result from the non - commutativity of the gauge structure .
when a particle performs two successive loops @xmath950 and @xmath951 , that both start and end at the same point @xmath952 , it will acquire a geometric phase factor @xmath953 . if , for every pair of loops @xmath954 , this geometric phase factor does not depend on the order of the operations @xmath955 ( where @xmath956 [ resp .
@xmath957 corresponds to the situation where the loop @xmath950 [ resp .
@xmath951 ] has been performed first ) , then the system is _ abelian _ in nature .
the observable effects related to these phase factors are those of a _ commutative _ gauge theory . on the other hand ,
if the phase factors differ @xmath959 for some loops @xmath950 and @xmath951 , then the gauge theory is genuinely _ non - abelian_. the manifestations of the underlying non - abelian gauge structure do not have any abelian counterpart @xcite . in continuum gauge theories
, the non - abelian property @xmath959 can be traced back to the non - commutativity of the field strength , @xmath960 \ne 0 $ ] , which naturally stems from the non - commutativity of the gauge potential s components @xmath961 \ne 0 $ ] , see ref .
@xcite . as already pointed out in section [ abvsnonab ] , the criterion @xmath961 \ne 0 $ ] is generally used in the literature to specify non - abelian gauge potentials " , but it is not sufficient to attest that the system hosts genuine non - abelian properties , which are captured by the field strength @xmath947 or the loop operators @xmath872 , as described above @xcite . to avoid any ambiguity , we stress that the term _ non - abelian gauge fields _ should be entirely based on the non - commutativity of the field strength or loop operators [ see also eq .
below ] , as described in this section .
an analogous criterium can be introduced in the lattice framework , in terms of the loop operator @xmath962 defined in eq . .
the lattice gauge structure associated with the link variables @xmath943 is said to be _ genuinely non - abelian _ if there exists a pair of loops @xmath963 and @xmath964 , both starting and ending at some site @xmath965 , such that u__1 u__2 u__2 u__1.[non - comm_property ] note that the non - abelian criterium , based on the non - commutativity property , is gauge invariant , as can be verified using eq . .
therefore , the link variables @xmath943 which lead to non - commutating loop operators , provide genuine non - abelian effects @xcite on the lattice .
+ besides , one can introduce a simple criterium to detect lattice configurations in which all the loop matrices @xmath962 can be simultaneously gauge - transformed into a simple phase factor u_=(i 2 ) _
n n , . [ trivial_lattice_gauge ] if such a reduction was possible over the whole lattice , then the multi - component system would behave as a collection of uncoupled abelian subsystems subjected to the same flux @xmath905 @xcite . on the square lattice , denoting @xmath966 the link variables along @xmath42 and @xmath43
, the condition @xmath967\ne 0 $ ] would constitute a natural , but incomplete criterium to detect non - abelian structures .
for example , the link variables @xmath968 have @xmath967= - 2i \hat\sigma_z\ne 0 $ ] , while the loop matrix around a unit cell is @xmath969 : this configuration therefore corresponds to a two - component lattice subjected to a uniform ( abelian ) flux per plaquette @xmath970 . to a su(n ) tight - binding model with constant hopping operators @xmath971 has been explored in ref . @xcite . ] as shown in ref .
@xcite , the wilson loop provides an unambiguous criterium to determine whether the set @xmath943 leads to uncoupled abelian subsystems captured by eq . .
indeed , the loop operator @xmath972 reduces to a simple phase factor @xmath973 if and only if the wilson loop satisfies @xmath974 .
the criterium based on non - trivial wilson loops @xmath975 is a necessary condition for attesting that the link variables @xmath943 produce a non - abelian gauge field [ eq . ] .
however , we stress that this condition is not _
sufficient_. for instance , a spin-1/2 lattice satisfying @xmath976 in all its plaquettes is clearly abelian in the sense of eq . , while it is generally associated with a non - trivial wilson loop @xmath977 , since the different spin components feel an opposite magnetic flux @xmath978 , i.e. the trivialization is not satisfied in this case .
we note that in the context of the non - abelian aharonov - bohm effect @xcite , such a gauge field , although _ abelian _ , is loosely referred to as a non - abelian flux " . again , to avoid any ambiguity , we stress that _ non - abelian gauge fields _ defined on a lattice are those that satisfy the strict criterium based on non - commutating loop operations [ eq . ] .
the non - relativistic abelian and non - abelian lattice gauge structures discussed above derive from applied gauge potentials ( or equivalently , link variables ) ; these are classical and non - dynamical and are captured by the non - relativistic tight - binding hamiltonian @xmath979 in this general expression , the operator @xmath980 describes the creation of a particle at the lattice site @xmath881 , in the spin state @xmath981 , and the link variables @xmath982 contribute to the hopping between connected sites @xmath983 .
the tunneling s strength is characterized by the hopping rate matrix element @xmath871 .
for simplicity , we take @xmath871 to be uniform ; and since we are focusing on non - interacting particles , we omitted the interaction term in eq . .
in the next paragraphs , we review a few relevant lattice models corresponding to specific configurations of the lattice and link variables .
a square lattice with a uniform magnetic field @xmath984 is an iconic model called the hofstadter model @xcite of a two - dimensional electron gas ( 2deg ) in a strong magnetic field , and is the most simple case of a non - trivial gauge field in a lattice . here
, particles undergo the traditional ( i.e. abelian ) aharonov - bohm effect when they circulate around the unit square plaquettes , @xmath985 .
this model is described by eq .
, with the specific form @xmath986 where the lattice sites are located at @xmath987 , @xmath988 ; and @xmath779 is the lattice period . here , the link variables @xmath989 \in { \rm u(1)}$ ] , commonly known as peierls phases " , satisfy , @xmath990 } , \nonumber \\ & = e^{i 2 \pi \phi}.\end{aligned}\ ] ] in a tight - binding treatment @xcite , eq . can be obtained through the _ peierls substitution _ @xmath991 , where @xmath992 is the tight - binding model s field - free dispersion relation . , where @xmath779 is the lattice spacing . performing the peierls substitution
, @xmath993 yields the effective single - particle hamiltonian @xmath994 $ ] , where @xmath995 and where @xmath996 are translation operators on the lattice , @xmath997 @xcite . ]
the hofstadter hamiltonian can alternately be obtained from the continuum hamiltonian , by discretizing the spatial coordinates and the derivative operators @xcite .
this continuum approach is only rigorously valid in the limit @xmath998 , namely for small flux @xmath999 , where the energy structure only slightly deviates from the landau levels .
away from this continuum limit , @xmath1000 , the tight - binding hamiltonian displays new features originating from the underlying lattice structure .
the hopping phases in the landau gauge are @xmath1001 for a rational flux @xmath1002 , with @xmath1003 , the hamiltonian commutes with the magnetic translation operators @xmath1004 and @xmath1005 . in this gauge ,
the system is described by @xmath1006 magnetic unit cells ( gauge dependent ) and its energy spectrum splits into @xmath581 ( gauge independent ) subbands , see fig .
[ fig_hofstadter]b .
the spectrum has a fractal ( self - similar ) pattern set of eigen - energies , called the hofstadter butterfly @xcite , see fig .
[ fig_hofstadter]c , and it has quantum hall phases with topological order , to be discussed in sect . [ quantumsimulation ] [ see also fig .
[ fig_hofstadter ] ( b ) ] .
at the special
@xmath1007-flux " case @xmath970 , the hofstadter lattice reduces to a two - band model displaying conical intersections at zero energy @xcite , similar to graphene s pseudo - relativistic spectrum @xcite . in this singular case ,
the system satisfies time - reversal symmetry @xmath1008 .
the hofstadter model can easily be extended to other two - dimensional lattices @xcite , and in particular to the honeycomb lattice @xcite , making it an efficient tool to investigate the electronic properties of graphene or other exotic materials in strong magnetic fields @xcite .
the effects of interactions in the hofstadter model have been analyzed in refs . @xcite .
the hofstadter model has been realized with cold atoms in modulated optical lattices @xcite . .
the quantized hall conductivity @xmath1009 is shown for fermi energies located within the bulk gaps .
( c ) the hofstadter butterfly and its energy spectrum as a function of the magnetic flux @xmath905 .
( d ) the haldane model on the honeycomb lattice . ( e ) energy spectrum of the haldane model . in the absence of nnn
hopping ( @xmath1010 ) , the spectrum has two dirac cones . in the presence of complex nnn hopping ( @xmath1011 ) , a bulk gap opens , leading to an anomalous quantum hall phase with the chern number @xmath1012 .
note that the lowest bands of the haldane and hofstadter models presented in ( b ) and ( e ) are topologically equivalent .
however , in contrast with the haldane model , the lowest band of the hofstadter spectrum shown in ( b ) exhibits an almost completely flat dispersion ; these topological - flat - band configurations are good candidates for realizing fractional chern insulators @xcite . ] in the early 1980 s , the quantum hall effect was discovered in two - dimensional electronic systems at modest magnetic fields @xcite , and until a seminal work by haldane @xcite , it was believed that the presence of a uniform magnetic field was necessary to produce this effect . in ref .
, haldane introduced a lattice model demonstrating that the essential ingredient was not the magnetic field , but rather the associated breaking of time - reversal - symmetry ( trs ) . in haldane s model , while trs is broken _ locally _ , the magnetic flux per unit cell is zero .
figure [ fig_hofstadter]d shows this model s honeycomb lattice topology , featuring both nearest - neighbor ( nn ) and next - nearest - neighbor ( nnn ) hopping . while the nn hopping is trivial ( positive and real valued ) , the nnn hopping terms are accompanied by non - trivial peierls phases , indicating the presence of a trs - breaking gauge potential .
the corresponding hamiltonian is @xmath1013 where @xmath1014 and @xmath1015 denotes the nn and nnn sites respectively of a honeycomb lattice , and @xmath1016 depending on the orientation of the path connecting the nnn sites @xmath1015 .
the peierls phases @xmath1017 are complex for nnn sites , leading to non - zero _
local _ magnetic flux @xmath1018 within triangular subplaquettes " @xmath190 , even though the total flux @xmath1019 inside each honeycomb unit cell . for @xmath1011 ,
an energy gap opens in the spectrum , see fig .
[ fig_hofstadter]e , leading to non - trivial topological orders and quantum anomalous " hall phases ( see also sect . [ quantumsimulation ] ) .
the haldane model played a surprising role in the prediction of the quantum spin hall effect , an early precursor to tr preserving topological insulators ( see below ) .
the haldane model has been recently implemented with cold atoms using time - modulated optical lattices @xcite .
the kane - mele model was introduced as a tight binding framework to explore the effects of spin - orbit couplings ( socs ) in graphene @xcite .
based on this model , kane and mele predicted a quantum spin hall effect , leading to the current interest in tr - invariant topological insulatrs ( cf . also refs . and ) .
interestingly , the intrinsic spin - orbit coupling term can be modeled by a nnn haldane hopping term , acting oppositely on the two spin components @xmath1020 . in the presence of soc
, each spin component feels a local magnetic " flux , yet trs is preserved because the fluxes are opposite for @xmath1021 and @xmath1022 .
the corresponding link variables can be represented as @xmath1023 matrices , with @xmath1024 . in solids ,
the rashba soc produced by an external electric field @xcite , is modeled by a nn hopping term that mixes the spin components non - uniformly .
it involves su(2 ) link variables such as @xmath1025 , where @xmath1026 denotes the unit vector along the link connecting the nn sites @xmath1027 .
the total kane - mele hamiltonian is @xmath1028 where the three terms correspond to : direct hopping between nn sites , intrinsic spin - orbit coupling , and rashba spin - orbit coupling @xcite . the kane - mele model for graphene
can therefore be viewed as the direct su(2 ) analogue of the haldane model with an extra rashba term characterised by the strength @xmath1029 featured in eq .. just as the haldane model produces quantum hall states , the kane - mele model produces quantum _ spin _ hall states , which can be viewed as two superimposed , and opposite , spin - filtered qh phases . a three - dimensional generalization of the kane - mele model , where the link variables @xmath1030 are defined along the links of a diamond lattice @xcite reveals the exciting physics of 3d topological insulators @xcite .
in the same spirit , goldman et al .
@xcite introduced a spinful su(2 ) analogue of the hofstadter model on a square lattice . in this model , particles with spin @xmath1020 experience a uniform magnetic flux per plaquette @xmath1031 opposite in sign for the two spin components .
the corresponding hamiltonian is @xmath1032 where the su(2 ) link variables @xmath1033 act on the two - component field operator @xmath1034 , defined at lattice site @xmath987 . for @xmath1035 ,
this models corresponds to two decoupled copies of the spinless hofstadter model - .
the effect of the link variable @xmath1036 is therefore analogous to the intrinsic spin - orbit coupling in eq . .
for @xmath1037 , the two spin components are mixed as they tunnel from one site to its nearest neighbor : the link variable @xmath1038 plays a role similar to the rashba coupling in eq . .
this model therefore captures the essential effects of the kane - mele trs - invariant model in a multi - band framework , but offers the practical advantage of only involving nn hopping on a square lattice .
the optical - lattice implementation of this model has been reported by aidelsburger et al .
@xcite and investigated by kennedy et al . @xcite .
the effects of interactions in the su(2 ) hofstadter model have been analyzed by cocks , orth and coworkers in refs .
@xcite .
the interplay between abelian and non - abelian gauge fields can be studied in a simple two - component model defined on a square lattice @xcite , where the link variables along @xmath42 and @xmath43 are @xmath1039 the abelian part of the gauge structure given by the u(1 ) phase @xmath1040 in @xmath1041 , corresponds to a uniform magnetic flux per plaquette @xmath905 ( cf .
[ hofsection ] ) .
the su(2 ) part is controlled by the parameters @xmath190 and @xmath1042 . in two limiting cases
this system is purely abelian , i.e. @xmath1043 : when @xmath1044 or @xmath1045 ; and when @xmath1046 and @xmath1047 . in both cases ,
the system reduces to two uncoupled ( abelian ) models . for arbitrary values of @xmath1048
the system features non - abelian fluxes [ see section [ non - abelian_wilson ] ] : the wilson loop is uniform and non - trivial @xmath1049 .
this model was investigated in the contexts of quantum hall physics @xcite , transition to the mott - insulating phase @xcite and non - abelian anyonic excitations @xcite .
alternative non - abelian lattice models have been proposed and studied @xcite , showing the rich properties stemming from non - abelian structures in non - relativistic quantum systems .
let us now consider the schemes enabling to create lattice gauge structures for ultracold atoms moving in optical lattices . in this context ,
the main ingredient for generating peierls phases along the links of the lattice is the so - called _ laser - assisted - tunneling method _ , initially introduced by ruostekoski - dunne - javanainen @xcite , and jaksch - zoller @xcite .
this method is based on the possibility to couple atoms living on neighboring sites of an optical lattice , hence controlling their tunneling over the lattice , as illustrated in fig . [ fig_assisted_shaking ] ( a ) .
peierls phases are then engineered by controlling the phase of the coupling , which allows to simulate lattice ( tight - binding ) hamiltonians with non - trivial link variables @xmath1050 , for example leading to the hofstadter hamiltonian in 2d .
this method was further developed by several authors , such as mueller @xcite , gerbier - dalibard @xcite , anisimovas et al .
@xcite , goldman et al .
@xcite and mazza et al .
this method , which can be extended to non - abelian structures @xcite and more exotic lattice geometries @xcite , can be summarized as follows : a. gauge fixing : what link variables @xmath1050 should be generated ?
b. prevent spontaneous hopping along the links for which @xmath1051 .
c. atom - light coupling : induce the hopping externally and engineer the desired link variables @xmath1050 by tuning the coupling lasers , see fig . [ fig_assisted_shaking ] ( a ) .
any physical scheme accomplishing the crucial steps ( b)-(c ) , relies on the specific properties of the atoms used in the experiment .
for instance , gerbier and dalibard proposed an elegant method exploiting the unique properties of alkaline - earth or ytterbium atoms @xcite [ see also anisimovas et al .
@xcite ] . in their proposal ,
the induced - hopping ( c ) involves a coupling between the ground state and a long - lived metastable excited electronic state . for these atomic species ,
metastable states have remarkably long lifetimes , and thus , the single - photon transitions to the excited state have greatly reduced spontaneous emission rates ( as compared , e.g. , to schemes based on two - photon transitions ) . for the more common alkali atoms ,
e.g. li , k , rb , alternative schemes based on raman couplings , i.e. two - photon transitions , are required @xcite .
we point out that the concept of laser - induced tunneling , which is based on the coupling of atoms living on neighboring lattice sites [ fig . [ fig_assisted_shaking ] ( a ) ] ,
is intimately related to methods exploiting time - periodic modulations ( shaking " ) of the optical lattice @xcite .
this analogy is illustrated in fig .
[ fig_assisted_shaking ] ( b ) . here , atoms are trapped in an optical superlattice potential , where neighboring sites are shifted in energy by the offset @xmath425 so as to inhibit the natural hopping .
a resonant modulation of the lattice then allows to re - establish the hopping in a controllable manner . in this framework , and in contrast with the standard laser - assisted method described in sect .
[ laser - assisted - section ] , the atoms can be prepared in a _ single _ internal state . designing a modulation that also transfers momentum to the atoms , e.g. using two raman " laser beams ,
also allows to generate peierls phases in the effective hopping matrix elements ( see section [ mainingredients ] and works by kolovsky @xcite , creffield - sols @xcite , bermudez et al .
@xcite , lim et al .
@xcite , baur - schleier - smith - cooper @xcite and goldman et al . @xcite . ) .
this shaking " scheme was successfully implemented in the munich @xcite and mit @xcite experiments to imprint space - dependent peierls phase in a 2d optical lattice , with a view to creating the hofstadter model with cold atoms .
we point out that off - resonant potential modulations can also be considered to engineer peierls phases in optical lattices , as recently demonstrated in hamburg @xcite [ see section [ sec : other schemes ] for a more detailed discussion on periodically - driven cold - atom systems ] .
+ below , we present the main ingredients for engineering abelian gauge structures in optical lattices , based on the experiment performed in munich in 2011 @xcite .
the generalization to non - abelian ( matrix - valued ) link variables @xmath1052 is briefly discussed in sect .
[ matrixlinkvariables ] .
other interesting schemes are further discussed in section [ sec : other schemes ] . and
@xmath1053 , are trapped in state - dependent optical lattices @xmath1054 . a resonant laser with frequency @xmath1055 couples the two states and induces effective tunneling matrix elements @xcite .
the configuration where @xmath1056 can be simply implemented for yb atoms , by using lasers at the anti - magic " wavelength @xcite .
( b ) shaking an optical superlattice .
atoms are trapped in a superlattice potential @xmath99 , displaying an energy offset @xmath425 between neighboring lattice sites and chosen so as to inhibit the hopping [ i.e. @xmath1057 , where @xmath871 is the natural hopping amplitude ] .
a resonant modulation of the lattice potential , with frequency @xmath1058 , re - activates the hopping in a controlled manner @xcite .
the lattice modulation can be created by two ( far - detuned ) running - wave beams with frequencies @xmath1059 , such that @xmath1060 .
the effective tunneling matrix element can be decorated with non - zero peierls phases , when the modulation further transfers momentum @xmath1061 to the atoms , which can be realized using
raman " lasers with wave vectors @xmath1062 , such that @xmath1063 ( see section [ mainingredients ] ) . an extension of this scheme was recently implemented in the munich @xcite and mit @xcite experiments .
see also fig . [ fig_assistedtunneling ] .
, width=384 ] [ [ choosing - the - gauge ] ] choosing the gauge + + + + + + + + + + + + + + + + + + the first step consists in selecting the gauge , thus dictating the explicit form of the link variables @xmath1050 . for the hofstadter model ,
we typically choose the landau gauge , in which the link variables @xmath989 $ ] are non - trivial along @xmath43 only ( i.e. @xmath1064 and thus the hopping along @xmath42 can be the lattice s native tunneling ) .
it is convenient to create non - trivial link variables @xmath1050 along the links connecting the nn sites of the lattice @xcite , so as to maximize the overlap between the wannier functions ( see eq .
+ [ [ preventing - natural - hopping ] ] preventing natural hopping + + + + + + + + + + + + + + + + + + + + + + + + + + the native hopping along the links for which @xmath1051 must be eliminated . for example , in the landau gauge , this would be the natural hopping along @xmath43 of a square lattice .
this simple , but important , task allows the subsequent external induction and control of hopping . in the munich experiment of 2011 @xcite
, spontaneous hopping is prohibited using a superlattice along @xmath43 , namely , by applying an energy offset @xmath425 much larger than the natural hopping amplitude @xmath871 between alternating sites , see fig .
[ fig_assistedtunneling]a .
schemes involving state - dependent lattices are also envisaged @xcite , where atoms located in neighboring sites experience distinct optical potentials , see fig . [ fig_assisted_shaking ] ( a ) .
state - dependent lattices are convenient for ytterbium atoms @xcite , and for bosonic alkali atoms . however ,
for fermionic alkali atoms , they generally lead to large spontaneous rates , and thus alternative schemes are required @xcite .
is permitted along @xmath42 , and is inhibited along @xmath43 by a superlattice potential ( with offset @xmath425 ) .
a pair of lasers , with wave vectors @xmath1062 and frequencies @xmath1059 , induce hopping along @xmath43 when @xmath1065 .
the hopping matrix elements @xmath1066 $ ] along @xmath43 realize the peierls substitution .
( b ) when @xmath1067 , this generates synthetic magnetic fluxes @xmath1068 in the plaquettes ( in units of the flux quantum ) . in this configuration ,
the flux per plaquette @xmath905 is staggered alternating sign along @xmath43 .
a flux rectification " leading to a uniform magnetic flux over the lattice necessitates a slightly alternative setup ( see the experiments in refs . @xcite and proposals in refs .
@xcite ) . , width=576 ] [ [ induced - tunneling ] ] induced - tunneling + + + + + + + + + + + + + + + + + the tunneling between neighboring sites @xmath1069 must be induced and controlled externally , leading to an effective hopping matrix element @xmath1070 . in this section ,
we consider the modulated - superlattice configuration used in the munich experiment @xcite , which is similar to the schemes implemented in the subsequent munich and mit experiments @xcite .
we refer the reader to refs .
@xcite for a description of the laser - induced tunneling methods using different internal states together with state - dependent lattices and inter - species coupling . in the superlattice
illustrated in fig .
[ fig_assistedtunneling ] ( a ) , atoms in neighboring sites occupy wannier functions @xmath1071 and @xmath1072 , respectively ( here the indices @xmath141 and @xmath1073 refer to the low energy and high energy sites in the superlattice ) . for a given energy offset @xmath425 ,
tunneling can be induced by an external time - dependent perturbation v_coupl(r , t)= ( q r - _ l t),[couplingraman ] from a pair of lasers with wave vectors @xmath1062 and frequencies @xmath1059 .
the @xmath1074 energy difference allows resonant coupling between the staggered lattice sites , see fig .
[ fig_assistedtunneling ] ( a ) . here , @xmath464 denotes the rabi frequency characterizing the strength of the atom - light coupling .
the key ingredient of the scheme is that the momentum transfer @xmath1075 can be adjusted by tuning the angle between the laser beams . the resulting hopping amplitude , from a _ low _ energy site @xmath881 to a _ high _ energy site @xmath891 ,
is then given by the integral @xcite ( see also @xcite ) @xmath1076 where @xmath1077 is the vector connecting neighboring sites @xmath140 and @xmath890 .
the effective hopping is therefore characterized by the amplitude @xmath1078 , along with a complex peierls phase factor determined by the momentum transfer @xmath1079 , see fig . [ fig_assistedtunneling ] ( a ) . in other words ,
the time - dependent perturbation generates a link variable given by @xmath1080 $ ] , where @xmath1081 . the effective hopping in eq .
can also be written in the symmetric " notation j^eff_jk ( r_j , r_k)= e^iq ( r_j + r_k)/2 w^*_h ( r - _ kl/2 ) w_l ( r + _ kl/2 ) e^i q r dr = ^eff_0 e^iq ( r_j + r_k)/2 , which is appropriate for certain geometries , especially for the non - square lattices @xcite .
importantly , the effective hopping from a _ high _ energy site @xmath891 to a _
low _ energy site @xmath881 @xmath1082 results from a momentum transfer @xmath1083 reversed in sign as compared to the _ low_-to-_high _ case and @xmath1084 are real @xcite . ] .
finally , we point out that the effective tunneling amplitude can be expressed in terms of a bessel function of the first kind @xmath1085 , with @xmath1086 , as expected for shaken lattice systems ( see refs .
@xcite , and also section [ sec : shaking lattice ] where off - resonnant driving is discussed in a related context ) .
[ [ staggered - flux - configuration ] ] staggered flux configuration + + + + + + + + + + + + + + + + + + + + + + + + + + + + equations - and fig .
[ fig_assistedtunneling ] show that the induced tunneling is necessarily accompanied with alternating peierls phases : @xmath1087 . generating such peierls phases on a square lattice , @xmath1088 leads to a staggered flux configuration , where successive plaquettes along @xmath43 are penetrated by fluxes @xmath1089 . here , the synthetic magnetic flux @xmath905 is governed by the raman coupling lasers @xmath1090 , and it can thus be set in the high - flux regime @xmath921 , equivalent to the effects of a gigantic magnetic field @xmath1091 for electrons in a crystalline lattice . however , this staggered configuration does not break time - reversal symmetry , and is only equivalent to the hofstadter model in the @xmath1007-flux " limit @xmath1092 .
the experimental realization of the staggered flux model in 2011 @xcite constituted a first important step towards the realization of the hofstadter model ( i.e. uniform magnetic flux over the whole lattice ) . staggered magnetic fluxes have also been realized in off - resonant shaken triangular optical lattices in 2013 @xcite .
[ [ the - uniform - flux - configuration ] ] the uniform flux configuration + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + this alternating field must somehow be rectified to produce a uniform field .
this rectification requires individually addressing successive hoppings @xmath1093 and @xmath1094 , for example using more elaborate potential landscapes and additional laser frequencies @xcite . in such landscapes ,
the energy offsets @xmath1095 and @xmath1096 between successive nn sites are different and the induced hoppings are produced independently .
flux rectification " , which is necessary to break time - reversal symmetry and produce quantum hall states on the square lattice , has been realized in 2013 by the groups of i. bloch @xcite and w. ketterle @xcite , using a potential gradient to individually address adjacent tunneling processes ( as originally suggested in the jaksch - zoller proposal @xcite ) .
the uniform - flux configuration has also been realized with an all - optical setup , which has been used to extract the chern number of hofstadter bands @xcite .
additional coupling lasers ( see above ) are unneeded for special lattice geometries , such as the honeycomb lattice . as was shown in refs .
@xcite , @xcite and @xcite , it is possible to generate local haldane - like fluxes using two intertwined triangular sublattices , denoted @xmath1097 and @xmath1098 , coupled together by a single laser coupling @xmath1099 . in this setup , the natural hopping between the nnn sites of the honeycomb lattice ( i.e. the nn sites of the triangular sublattices @xmath1097 and @xmath1098 ) remains , while the laser - induced hopping with peierls phases acts between nn sites . for arbitrary values of the coupling lasers wave vectors
@xmath1100 , this configuration has non - trivial fluxes inside the subplaquettes of the unit hexagonal cells , and therefore reproduces the haldane model s quantum hall phases
. a complete description of the phase diagrams and flux configurations stemming from this cold - atom setup can be found in ref . .
this scheme could also be easily implemented with yb atoms , in which case it is convenient to set the lattice potential at a so - called anti - magic " wave - length @xcite . in this case , the two internal states are automatically trapped in two intertwined triangular and honeycomb lattices , and the coupling can be directly addressed with a single - photon transition , see anisimovas et al . @xcite .
the haldane model has been realized experimentally using time - modulated honeycomb optical lattices @xcite , as already mentioned in section [ haldanesection ] . +
this honeycomb system could be also extended to the spinful kane - mele model for @xmath1101 topological insulators @xcite . in this case
, each sublattice must trap atoms in two internal atomic states ( see sect .
[ matrixlinkvariables ] ) coupled independently by lasers such that the tunneling operators ( @xmath1102 matrices acting between nn sites @xmath140 and @xmath890 ) are @xmath1103 . in this spinful honeycomb lattice configuration , non - trivial @xmath1101 topological phases , featuring helical edge states exist @xcite .
this system can include time - reversal breaking perturbations , testing the robustness of the @xmath1101 topological phases and exploring phase transitions between helical and chiral edge textures @xcite . the general atom - coupling method presented in sect .
[ mainingredients ] for creating peierls phases @xmath1104 can be extended to produce matrix - valued link variables [ e.g. @xmath1105 . for this
, each lattice site must host @xmath83 nearly - degenerate spin states with index @xmath1106 .
using external ( real ) magnetic fields to lift any degeneracies , each state @xmath1107 can then be individually trapped and coupled to other sublevel states @xmath1108 . in particular , the effective tunneling of an atom in a specific internal state @xmath1107 , from site @xmath140 to @xmath890 , could then be induced and controlled individually by external couplings @xcite . in principle , the coupling can be chosen to flip the atomic spin @xmath1109 during the hopping process , resulting in a non - diagonal hopping matrix @xmath1110 .
considering the simplest case @xmath1111 , this scheme can produce synthetic spin - orbit coupling terms with @xmath1112 .
such non - abelian gauge potentials , based on spin - dependent hopping , lead to the spin-1/2 models presented in sect .
[ spinorbitlattice ] , making possible topological insulators in 2d and 3d optical lattices @xcite .
non - trivial hopping operators along both spatial directions , using laser - induced tunneling techniques , generally require checkerboard lattices @xcite . while our list of ingredients for inducing non - trivial link operators @xmath1050 in optical lattices
was based on laser - induced tunneling methods , alternative methods have been proposed and some of them have been experimentally realized .
this section reviews these schemes .
a quite different strategy consists in using off - resonant periodically - driven optical lattices @xcite .
one can generate artificial magnetic flux in such lattices by combining lattices and time - dependent quadrupolar potentials @xcite , by modulating the lattice depth ( i.e. the tunneling amplitude ) in a directional manner @xcite or by shaking optical lattices @xcite .
similarly , spin - orbit couplings could be generated by subjecting an optical lattice to time - dependent magnetic fields @xcite .
a general formalism describing periodically - driven quantum systems and effective gauge structures can be found in refs .
@xcite . motivated by the recent experiments in hamburg @xcite ,
let us consider a scheme based on shaken optical lattices , as already outlined in sect .
[ sub : shaking ] .
the method relies on off - resonant modulations of the lattice potential @xmath1113 , where @xmath1114 is the position vector in the oscillating frame of reference [ see eq . ] .
the displacement vector @xmath1115 is time - periodic , with period @xmath61 , and is given by eq .
( [ eq : linear - circular ] ) for linear or circular harmonic shaking . as presented in sect .
[ sub : shaking ] , the transformation to a non - inertial frame of reference modifies the hamiltonian by adding a spatially homogeneous vector potential @xmath1116 to the momentum , see eq .
( [ eq : h - shaken - frame-1 ] ) : @xmath1117 the full hamiltonian @xmath26 shares the same spatial periodicity as the lattice potential @xmath1118 , hence , the quasimomenta are still good quantum numbers for the instantaneous eigenstates and eigen - energies of the hamiltonian @xmath26 .
since the hamiltonian is also time - periodic , it is appropriate to deal with the quasi - energies and the corresponding eigenstates of the effective ( floquet ) hamiltonian @xmath1119 [ defined below in eq .
( [ eq : u(t ) ] ) ] when analyzing the topological properties and effective gauge structures induced by the driving @xcite .
let us consider a situation where the potential @xmath1118 traps atoms in a deep 1d optical lattice , directed along @xmath510 , which is subjected to a harmonic modulation @xmath1120 . using the tight - binding approximation , the second - quantized lattice hamiltonian reads @xmath1121 with @xmath1122 , where the operator @xmath1123 creates a particle at lattice site @xmath1124 , @xmath779 is the lattice spacing and @xmath871 is the tunneling matrix element between the neighboring sites .
the time - dependent phase factors @xmath1125 accompanying the hopping terms appear by simple application of the peierls substitution @xcite .
it is to be emphasized that these time - dependent peierls phases " oscillate fast in time , and in particular , they should not be mistaken with the standard peierls phases associated with ( synthetic ) magnetic flux discussed in section [ hofsection ] .
the effective ( floquet ) hamiltonian @xmath1119 ruling the time - averaged dynamics can be defined through the evolution operator over a cycle @xcite @xmath1126 where @xmath1127 denotes time - ordering . in the absence of additional confining trap ( e.g. a harmonic trap @xmath1128 ) , and imposing the periodic boundary conditions , one verifies that the 1d tight - binding hamiltonian in eq .
( [ ham_tb_1d ] ) commutes with itself at different times . in this case , the time - ordering can be omitted in eq .
( [ eq : u(t ) ] ) , and the effective hamiltonian @xmath1119 is then simply obtained through the time - average @xmath1129,\quad\xi_{0}=m\kappa a\omega/\hbar\ , , \label{ham_tb_1d_f}\end{aligned}\ ] ] where @xmath1130 denotes the bessel function of the first kind @xcite . thus
, modulating an optical lattice enables one to control the sign and amplitude of the tunneling between the neighboring sites : @xmath1131 .
this can have non - trivial consequences for triangular optical lattices , where a change of sign @xmath1132 leads to staggered synthetic magnetic fluxes and frustrated magnetism @xcite .
the effects of additional potentials , which would require time - ordering in eq .
( [ eq : u(t ) ] ) , can be evaluated by considering a perturbative treatment in @xmath1133 @xcite . in particular
, we note that the cycle - averaged effective hamiltonian in eq
. corresponds to the lowest - order term of the magnus expansion @xcite , which is generally relevant for sufficiently short periods @xmath1134 . to induce complex effective peierls phase factors , @xmath1135 , non - sinusoidal driving is necessarily required , so that certain temporal symmetries should be broken by the forcing @xcite .
the specific protocol implemented by struck et al .
@xcite , consists in a sinusoidal forcing over a period @xmath1136 interrupted by a short periods of rest @xmath1137 , so the lattice perturbation has a total period of @xmath1138 and a zero mean value . under such conditions ,
the effective tunneling operator is @xcite @xmath1139 leading to a constant but complex valued effective peierls phase .
this scheme was realized experimentally in hamburg , for bosons in a 1d optical lattice , where the phase @xmath461 affected the lowest bloch band dispersion through @xmath1140 . in this setup
, the effective peierls phase @xmath461 can therefore be evaluated by measuring the quasimomentum distribution of the bec , whose superfluid ground state is reached at a finite value dictated by @xmath461 @xcite .
recently , jimenez - garcia et al . proposed and implemented an original setup producing periodic potential for cold atoms @xcite , with naturally complex tunneling operators @xmath1147 .
in other words , this scheme produces the lattice and the peierls phases @xmath461 _ simultaneously_. in this method , @xmath393rb atoms in the @xmath341 ground level are concurrently subjected to a radio - frequency ( rf ) magnetic field and to two counter - propagating raman laser beams with wave number @xmath1148 .
the raman and rf fields resonantly coupled the three spin states @xmath1149 , yielding the coupling hamiltonian , see secs .
[ sec : tworaman - general ] and [ sub : a - pair - of ] , h_rf + r ( r ) = ( r ) + constant , [ rfandraman ] where @xmath1150 is the angular momentum operator and @xmath1151 here @xmath1152 and @xmath1153 are the coupling strength , or rabi frequencies , associated with the rf and raman fields , and @xmath352 is the detuning from raman resonance @xcite .
the two non - trivial eigenvalues obtained of are ( x ) = , [ rflattice ] producing a periodic potential along @xmath42 when @xmath1154 .
importantly , eq . describes a spin-1 particle subjected to an effective space - dependent magnetic field @xmath1155 , where @xmath1156 and @xmath1157 are the land factor and bohr magneton , respectively .
now , according to berry @xcite , an eigenstate of hamiltonian acquires a geometric ( berry ) phase @xmath461 when a loop @xmath1158 is performed in the parameter space spanned by the
magnetic " field @xmath1159 . here ,
@xmath1159 is space - dependent , and therefore , the spin will precess and make a loop each time an atom hops from one lattice site to its neighboring sites ( the lattice structure being dictated by eq . ) .
the hopping process is therefore naturally accompanied with a non - trivial peierls " phase @xmath461 , proportional to the solid angle subtended @xmath1160 .
this strategy relies on the fine tuning of both the raman and rf fields , and simultaneously produces a periodic structure intrinsically featuring peierls phases , as demonstrated experimentally at nist @xcite .
extensions of this setup could realize non - trivial magnetic flux configurations in 2d lattices , through space - dependent peierls phases .
again , this requires more complicated but realistic arrangements , for example , combining a state - independent optical lattice along one direction with a state - dependent vector " lattice along the other @xcite .
alternatively , one could make use of the optical flux lattices @xcite considered in sec .
[ sub : flux - latt ] .
+ most proposals based on laser - coupling methods for the alkali atoms , involve two - photon raman transitions coupling different internal atomic states . in this context , a crucial experimental issue concerns the minimization of undesired heating stemming from spontaneous emission , which can not be reduced even using large detuning of the raman lasers , as pointed out at the end of the sec .
[ sec : tworaman - general ] .
this drawback is particularly severe for alkali fermions , e.g. @xmath1161k and @xmath1162 li , where the possible detunings are so small that they necessarily imply large spontaneous emission rates .
therefore alternative methods are required to investigate the physics of fermionic atoms subjected to synthetic gauge potentials .
several solutions have already been invoked above , such as exploiting the coupling to long - lived electronically excited states ( e.g. using ytterbium ) or engineering the gauge potential through lattice - shaking methods .
another solution consists in trapping and coupling different hyperfine levels using the magnetic ( zeeman ) potentials produced by an array of current - carrying wires , therefore avoiding the use of optical state - dependent lattices and raman coupling @xcite .
here we describe how such a setup could be implemented to reproduce the abelian gauge structure of the hofstadter model ( a generalization of this atom - chip method has been described in ref . to produce the su(2 ) hamiltonian ) .
first , consider two atomic states @xmath1163 , @xmath1164 of @xmath1162li , trapped in a primary ( state - independent ) optical lattice potential @xmath1165 along @xmath42 .
importantly , the corresponding pair of laser beams are incident on an atom chip s reflective surface , trapping the atoms about 5 @xmath45 m above the device .
the atom chip is consists of an array of conducting wires , with alternating currents @xmath1166 @xcite , which traps the atoms in a deep 1d zeeman lattice along @xmath43 . since the chosen states @xmath1167
have opposite magnetic moments , the zeeman shifts produce a state - dependent lattice along this direction .
then , additional _ moving _ zeeman lattices @xcite , with space - dependent currents @xmath1168 , act as a time - dependent ( rf ) perturbation directly coupling the internal states ( in direct analogy with ) , and thereby producing the desired space - dependent peierls @xmath1169 .
such an elegant and versatile method , suitable both for bosonic and fermionic species , is currently in development at nist , gaithersburg .
jaksch and klein proposed immersing the optical lattice and the atoms of interest ( @xmath1097-atoms ) into a bec formed by @xmath1098-atoms @xcite .
then , the interaction between the @xmath1097 and @xmath1098 atoms result in phonon excitations , which in turn induce effective interactions between @xmath1097-atoms located at nn sites @xcite . when @xmath1098-bec is made to rotate , the resulting effective interactions imprint space - dependent peierls phases to the tunneling matrix elements of the @xmath1097-atoms .
such a system could potentially lead to the realization of a synthetic magnetic field in the dynamics of the @xmath1097-system .
finally , interesting properties of 2d systems can be captured by 1d systems , through the concept of dimensional reduction @xcite .
this fact , which is particularly relevant for topological systems , offers an interesting route for the exploration of topological order using 1d optical lattices @xcite . also , this strategy suggests an alternative way to reproduce the effects of magnetic fields in a simple optical - lattice environment .
let us illustrate this concept for 2d topological systems , namely , systems exhibiting robust edge states at their 1d boundaries with energies @xmath1170 located within bulk gaps @xcite .
first , suppose that the system is defined on a 2d square lattice and described by a general hamiltonian h = _ j , k c^_j h_j k c_k , [ eq : full2d ] where @xmath1171 creates a particle at lattice site @xmath1172 and @xmath1173 . in this open 2d geometry ,
topological edge states are localized along the single 1d boundary delimiting the large square @xmath1174 @xcite .
now , consider the cylindrical geometry obtained by identifying the opposite edges at @xmath1175 and @xmath1176 , namely , by imposing periodic boundary conditions along the @xmath469 direction only . in this geometry , and
considering bloch s theorem , the system is well described by the spatial coordinate @xmath1177 and the quasi - momentum @xmath1178 , where @xmath1179 . in particular
, the system hamiltonian can be decomposed as a sum h = _ m=1^l _ k_y h ( m ; k_y ) = _ k_y h _
( k_y),[eq : full1d ] which indicates that the 2d system can be partitioned into independent 1d chains @xcite .
the edge states are now located at the two opposite edges of the cylinder , defined at @xmath438 and @xmath1180 , and their dispersion relations are expressed as @xmath1181 .
the general dimensional - reduction strategy can be formulated as follows : the energy spectrum and edge - state structures emanating from the 2d hamiltonian - could be captured by a family of 1d models with hamiltonians @xmath1182 , where the controllable parameter @xmath1183 $ ] should be identified with the quasi - momentum @xmath1184 . in other words ,
engineering a 1d system with tunable hamiltonian @xmath1182 would provide useful informations related to the full 2d system of interest . to be specific ,
let us consider the hofstadter model introduced in sect .
[ hofsection ] and described by the hamiltonian - . using the landau gauge , the single - particle wave function can be written as @xmath1185 , where the function @xmath1186 satisfies a 1d schrdinger equation and where @xmath1184 is the quasi - momentum along @xmath43 .
the schrdinger equation associated with this hofstadter cylinder " takes the form of the harper - aubry - andr equation @xcite @xmath1187,\label{aubry}\ ] ] where @xmath1177 . solving this equation yields the projected bulk bands of the hofstadter model @xmath1188 , which display @xmath1189 bulk gaps for @xmath1190 .
moreover , since the cylinder is a partially opened geometry ( with edges ) , new states with energies @xmath1191 are located within the bulk gaps .
these states are topological edge states , which play an important role in the quantum hall effect @xcite . a typical spectrum , showing the bulk and edge states dispersions , is shown in fig . [ fig_edgestates ] .
originally , the aubry - andr model was studied as a simple model for anderson localization @xcite ; in this 1d model , @xmath1184 is an adjustable parameter ( not related to quasi - momentum ) , and it is therefore treated on the same level as the parameter @xmath905 . recently , this model has been realized with cold atoms @xcite , in a quasi - periodic 1d lattice created by interfering two optical lattices with incommensurate wave numbers @xmath1192 . in this context , the parameters in eq .
, @xmath1193 and @xmath1184 , are tuned by the lasers creating the two lattices @xcite .
this reproduces the dimensional reduction of the hofstadter optical lattice , where the synthetic magnetic flux @xmath905 can be easily adjusted and where the quasi - momentum " @xmath1184 is fixed by the laser phases . considering fermionic atoms in such a 1d setup , and setting the fermi energy inside a bulk gap , it is possible to populate topological edge states for a certain range of the tunable phase @xmath1184 @xcite . extending this scheme to two - component 1d optical lattices , and considering well - designed state - dependent bichromatic optical lattices @xcite , mei et al .
obtained the dimensional reduction of the @xmath1101 topological insulator in eq .
[ goldham ] . by sweeping the tunable phase @xmath1184 ,
one is then able to transfer topological ( helical @xcite ) edge states with opposite spin , from one edge to the other @xcite .
the interacting bosonic version of the 1d bichromatic optical lattice described above has been explored by deng and santos @xcite , where the topological phase diagram has been obtained in terms of the interaction strength , the atomic filling factor and the strength of the auxiliary lattice .
the dimensional reduction strategy therefore allows to access gauge structures and topologically ordered phases in a rather simple manner .
these methods could be extended to access unobserved phenomena , such as the 4d quantum hall effect @xcite .
+ finally , 2d atomic lattice systems could be realized through the concept of _ synthetic dimensions _
@xcite , where the dimensionality of the physical optical lattice @xmath1194 is augmented by a synthetic dimension spanned by the internal states of the atoms .
for instance , a 2d ( semi-)synthetic lattice could be obtained from a @xmath1195 optical lattice with @xmath515 lattice sites ( with spatial coordinates @xmath1196 , where @xmath779 is the physical lattice spacing and @xmath1197 ) , and filled with a @xmath83-component atomic gas : in such a configuration , the 2d synthetic lattice is characterized by @xmath1198 lattice sites located at the coordinates @xmath1199 , where @xmath1200 and @xmath1201 . in synthetic lattices ,
the hopping is natural along the physical direction @xmath510 , while it is assisted along the synthetic ( spin ) direction through atom - light coupling ( e.g. raman transitions ) .
the realization of synthetic magnetic fluxes in synthetic lattices has been described by celi et al . in ref .
@xcite , where it was shown that such setups could be exploited to observe the hofstadter butterfly spectrum and the evolution of chiral ( topological ) edge states , using atoms with @xmath1202 internal states .
the propagation of chiral edge states in ladder systems ( @xmath1111 ) penetrated by synthetic magnetic fields has been studied by hgel and paredes @xcite .
an experimental implementation of of such a ladder scheme ( without involvement of the synthetic dimension ) has been recently reported by atala et al .
[ sect : simulation ] [ quantumsimulation ] the addition of synthetic gauge potentials can modify the properties of the atomic system significantly . by appropriately tuning this gauge field , single- and many - body configurations
can be obtained , leading to a plethora of interesting quantum phases .
numerous cold - atom systems can realize quantum hall physics , where landau - like levels emerge from a synthetic magnetic field , e.g. through rotation @xcite or laser - induced methods @xcite .
similarly , optical - lattice setups subjected to synthetic magnetic fluxes lead to band structures characterized by non - zero chern numbers or wilson loops , offering an alternative route to reach quantum hall ( chern insulating ) phases . cold - atom simulators of quantum hall states are motivated by the fact that these controllable systems could offer an instructive insight on the physics of qh liquids , such as those featuring non - abelian excitations @xcite .
we note that qh photonics systems offer an alternative route towards this goal @xcite .
the full set of non - interacting topological phases of matter have now been classified @xcite .
this periodic table " includes the @xmath1101 topological insulators , which have been experimentally observed in 2d and 3d materials @xcite , but also includes topological insulating and superconducting phases that are not known to exist in materials @xcite .
cold - atom setups subjected to gauge potentials might be tailored to access these phases , including those exhibiting majorana zero modes @xcite ( see also the proposals to realize topological kitaev - like chains using atomic quantum wires @xcite ) .
gauge structures have been experimentally realized in several laboratories ( see previous sections ) , and a first observation of the zak ( geometric ) phase @xcite has been reported for a 1d optical lattice reproducing the rice - mele model @xcite .
however , going beyond this first step , the detection of quantized topological invariants in atomic setups remains an important issue . in this section
, we review methods for identifying topological matter using probes that are available in existing cold - atom laboratories .
while we focus on detecting topological order , synthetic gauge potentials can also lead to other interesting effects , such as pseudo - relativistic band structures [ see the review @xcite and the laboratory measurement of zitterbewegung @xcite ] and vortex physics ( see the review @xcite and also refs . ) .
finally , the quantum simulation of quantum field theories , such as encountered in high - energy physics , is discussed in sect .
[ sect : dynamical ] . in 2d electronics systems ,
the lorentz force from a perpendicular magnetic field deflects electrons perpendicular to a driving electric field , resulting in a non - zero transverse hall conductivity @xcite .
thus the associated transverse current is a natural signature of synthetic magnetic fields in cold - atom systems .
however , transport measurements are relatively complicated to perform in cold - atom experiments , as it requires the challenge of first connecting reservoirs to the system and then detect currents @xcite .
recently , an analogue of a transport experiment in a flattened bec subjected to a synthetic magnetic field was performed in the ac limit
analogous to capacitively contacting an electronic system . in this scheme @xcite ,
currents are induced by modulating the external confining potential , allowing the full reconstruction of the resistivity tensor ( including the hall response ) .
the quantum mechanical problem of a particle in a magnetic field has the well - known landau levels as its eigen - energies , which lie at the root of the quantum hall effect @xcite . for non - interacting fermions in a synthetic magnetic field , the lll - regime " ( where only the lowest landau level ( lll ) is occupied ) could be detected in time - of - flight experiments , by simply observing that the lll states expand much faster than the states in higher landau levels @xcite . in the weakly - interacting regime of cold bosonic gases in a synthetic magnetic field ,
the mean field ground state is characterized by a macroscopic occupation of the lll , giving rise to a triangular vortex lattice in the particle density ( see sect . [
ian : vortex ] and @xcite ) . as the number of vortices increases ,
the atomic analogue of the qh filling factor decreases and the system enters the strongly - correlated regime where a family of fqh - like liquids have been predicted @xcite .
a thoughtful description of the atomic landau problem is given in ref .
@xcite , where observable signatures are discussed both for the weakly and strongly interacting regimes .
the stabilization of fqh atomic states in realistic conditions , e.g. by significantly increasing the incompressibility gap with respect to typical experimental temperatures , still constitutes a fundamental issue @xcite .
once realized in laboratories , the fqh liquids could be distinguished in experiments owing to their incompressible nature , manifested as plateaus in the atomic density distribution @xcite .
additional signatures of these strongly - correlated states are present in transport measurements with fractional transverse ( hall - like ) conductivities @xcite ; through density
density correlation functions @xcite ; or by the response of the atomic cloud to quasihole excitations induced by an external laser beam @xcite .
finally , the topological order associated with these fqh states could be evaluated by directly detecting the topological edge states ( see ref . and sect .
[ detectedge ] below ) .
diverse theoretical proposals to simulate topological phases with cold atoms exist .
these proposals describe techniques for creating : chern insulators @xcite ; @xmath1101 topological insulators in 2d @xcite and 3d lattices @xcite ; ( interaction - induced ) topological mott insulators @xcite ; @xmath1203 ( class aiii ) topological insulators in 1d @xcite ; and topological superconductors @xcite .
recently , the 1d rice - mele model was realized in a 1d optical lattice @xcite , where its non - trivial topological phase has been identified through the experimental determination of the zak phase . in general
, detecting topological properties is subtle .
topological phases are all specified by two holographic " characteristics @xcite : ( 1 ) in a topologically - ordered bulk , there exists a non - zero topological invariant associated with the bulk states @xcite , and ( 2 ) topological edge states are spatially localized on the periphery of the system with energies within the bulk gaps @xcite .
both properties are robust and simultaneously survive as long as the bulk gaps remain open and temperatures are small compared to the gaps . in the following , we consider non - interacting fermionic systems with the fermi energy tuned inside such a topological gap . before reviewing proposals for measuring topological properties in atomic systems ,
let us briefly summarize the state - of - the - art in solid - state systems .
topological invariants have been revealed through transport measurements , e.g. , by measuring the hall conductivity in the integer qh regime @xcite , or in the quantum spin hall ( qsh ) regime of topological insulators @xcite . besides
, edge states have been identified through spectroscopy @xcite and interference methods @xcite . very recently , qsh helical edge states have been directly imaged using a cryogenic microwave impedance microscope @xcite . performing transport measurements in atomic systems
is not as straightforward as in condensed - matter setups .
however , cold - atoms setups offer complimentary techniques that can be exploited to measure topological invariants and identify topological edge states . in the following sections , we sketch several methods allowing to identify the chern number @xmath50 , which classifies chern ( qh ) insulating phases @xcite
. generalizations of these methods to detect other topological classes are also discussed .
[ [ density - plateaus - in - the - hofstadter - optical - lattice - the - streda - formula - and - signatures - in - time - of - flight - experiments ] ] density plateaus in the hofstadter optical lattice , the streda formula and signatures in time - of - flight experiments + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the hofstadter butterfly is the spectrum of a particle on a 2d lattice subjected to a uniform magnetic field @xcite . when the magnetic flux per plaquette is rational , @xmath1190 , this spectrum splits into @xmath581 sub - bands @xmath1204 , where @xmath1205 and where @xmath1206 is the quasi - momentum , see fig .
[ fig_hofstadter]b .
each bulk band @xmath1204 is associated with a chern number @xmath1207 , a topological index which remains constant under external perturbations as long as the bulk gaps do not close @xcite .
the chern number @xmath1208 is an integer given by the integral @xmath1209 over the first brillouin zone ( bz ) of the berry s curvature @xmath1210 associated with the band @xmath1204 . here
, @xmath1211 is the single - particle state in the @xmath1212 band with crystal momentum @xmath1213 .
this topological order strikingly manifest itself when the fermi energy @xmath1214 resides in a bulk energy gap .
the transverse ( hall ) conductivity is then quantized as @xmath1215 where the sum includes the contribution of all occupied bulk bands , and @xmath1216 is the conductivity quantum , see fig .
[ fig_hofstadter]b .
interestingly , this quantized quantity is related to the particle density @xmath1217 , making its detection with cold atoms particularly practical @xcite .
the connection to density is based on the streda formula @xcite , which re - expresses the quantized hall conductivity as the derivative @xmath1218 , where @xmath83 is the number of states lying below the fermi energy and @xmath1098 is the magnetic field . for optical lattice experiments with a smooth confining potential @xmath1219 ,
the spatial density profile @xmath1220 in the local - density approximation is @xcite @xmath1221,\label{lda : chern}\ ] ] where @xmath1222 is the homogeneous - system density of states .
thus , the density @xmath1220 counts the number of states below the local chemical potential " @xmath1223 . in the presence of a uniform synthetic magnetic flux
@xmath1224 , the bulk energy spectrum @xmath1225 associated with the homogeneous system is split into @xmath581 bulk bands . according to eq .
, this splitting will produce @xmath1189 plateaus in the density profile @xmath1220 @xcite .
one can thus associate each density plateau with one of the @xmath1189 bulk gaps characterizing the homogeneous - system spectrum @xmath1225 .
then , by comparing the density plateaus @xmath1226 obtained from two different configurations of the magnetic flux @xmath1227 but corresponding to the opening of the same bulk gap in the bulk spectrum @xmath1225 , one obtains the analogue of the streda formula for the hofstadter optical lattice @xcite . identifying the plateaus @xmath1226 corresponding to the same @xmath1228th bulk gap , one obtains the integer @xmath1229 analogous to the quantized hall conductivity of an electronic system with the fermi energy set within the @xmath1228th gap @xcite .
this method allows access to the sum of chern numbers @xmath1230 , by comparing two measurements of atom density at different values of synthetic magnetic flux , and offers a simple method to identify topological order @xcite and phase transitions @xcite .
+ moreover , it was shown in ref . that the topological quantity @xmath50 could also be revealed in the momentum density @xmath1231 of the same hofstadter optical - lattice . indeed , under specific conditions , i.e. in the limit of large hopping anisotropy or for small synthetic flux per plaquette @xmath905 , the images obtained from time - of - flight experiments should display oscillations whose periodicity can be related to the value of @xmath50 @xcite . [
[ measuring - the - winding - number - of - a - haldane - chern - insulator ] ] measuring the winding number of a haldane - chern insulator + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the haldane model @xcite and its cold - atom generalizations @xcite are all described by two - band hamiltonians of the form @xmath1232 where @xmath1233 is the vector of pauli matrices .
the berry s curvature @xmath1234 associated with the lowest energy band @xmath821 , and the related chern number @xmath50 in eq .
[ chernnumber ] , can be expressed in terms of the normalized vector @xmath1235 @xcite , @xmath1236 from this , we notice that the chern number @xmath50 measures the number of times the vector field @xmath1237 covers the unit sphere as @xmath798 is varied over the entire brillouin zone @xmath1238 . therefore , for a system prepared in a phase @xmath1239 , an experimental measurement of @xmath1237 would depict a skyrmion pattern on a pixelated " brillouin zone , leading to an approximate measure of the chern number @xcite . for the specific haldane - like model introduced by alba et al . and discussed in sect .
[ albasection ] , the vector field @xmath1237 can be reconstructed from spin - resolved momentum densities @xmath1240 associated with the two atomic species present in the lattice @xcite .
this model therefore offers a simple platform to measure the skyrmion patterns and topological index @xmath50 from spin - resolved time - of - flight images .
[ [ semiclassical - dynamics - the - berrys - curvature - and - the - chern - number ] ] semiclassical dynamics , the berry s curvature and the chern number + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the equations of motion of a wave packet evolving on a lattice , centered at position @xmath1241 with crystal - momentum @xmath798 , and driven by an external force @xmath1242 , are given by @xcite @xmath1243 \mathcal{f } ( \bs k ) \nonumber \\
\hbar \dot{\bs k}= \bs f , \label{blochoscil}\end{aligned}\ ] ] where @xmath1244 is the berry s curvature introduced in eq . , and where @xmath1245 is the band structure characterizing the lattice system .
@xcite , price and cooper showed that cold atoms undergoing bloch oscillations @xcite can follow trajectories where the band structure s contribution to the velocity @xmath1246 vanishes . following this protocol ,
a measure of the mean velocity for many trajectories gives the berry s curvature @xmath1244 over a
pixelated " brillouin zone @xcite . by properly adjusting the path undergone by the wave packet
, the chern number can be evaluated . in principle , this method could be applied to any lattice system . +
a similar scheme was recently proposed by abanin et al .
@xcite , where bloch oscillations are combined with interferometry techniques to determine the berry s curvature and the chern number of 2d optical lattice systems .
this method is based on the measure of the zak phase @xcite , which has already been experimentally implemented in a 1d system @xcite . in ref .
@xcite , liu et al . introduced a method to measure the chern number based on spin - resolved bloch oscillations , observing that the topological index can be obtained by measuring the spin - polarization of the atomic gas at specific ( highly symmetric ) points within the brillouin zone .
+ finally , it was shown by dauphin and goldman that the chern number ( or equivalently the hall conductivity ) could be measured by imaging the center - of - mass displacement of a fermi gas subjected to a constant force @xmath1247 . setting the fermi energy within a topological bulk gap
, the contribution from the group velocity naturally vanishes and the displacement along the transverse direction @xmath510 is then directly proportional to the force multiplied by the chern number @xcite .
this simple and direct scheme is robust against perturbations and it could be implemented in any cold - atom setup hosting chern insulating phases to detect non - trivial topological order .
we stress that this method requires high - resolution microscopes to measure the ( integral ) chern number with high precision , i.e. @xmath1248 , the mean displacement being of the order of a few tens of lattice sites after reasonable experimental times @xcite .
[ [ the - hybrid - time - of - flight - measurement ] ] the hybrid - time - of - flight measurement + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + wang et al .
@xcite showed that the chern number could be read out from hybrid time - of - flight ( tof ) images , by detecting atom density after suddenly releasing the external confining potential @xmath1249 along one direction only , for example along @xmath43 . by combining _ in situ _ imaging along @xmath42 and tof imaging along the release direction @xmath43 , such an experiment would give access to the hybrid particle density @xmath1250 , where @xmath1184 is the crystal - momentum along @xmath43 .
the relation between the hybrid density and the chern number are related through dimensional reduction ( sect .
[ quasi2d ] ) . in this picture
, the 2d hofstadter model is viewed as a 1d harper - aubry - andr lattice directed along @xmath42 and described by a hamiltonian @xmath1251 , where @xmath905 ( the flux ) and @xmath1184 ( crystal - momentum ) are both interpreted as parameters ( see eq .
[ aubry ] ) .
the proposal of ref . exploits a relation between the electric polarization of electronic systems to their quantized chern number @xcite .
the chern number measures the charge transported from one boundary of the 1d lattice to the other under the variation of the parameter @xmath1252 , i.e. after a full cycle in the brillouin zone .
in the cold - atom framework , the measured hybrid density @xmath1250 permits a numerical reconstruction of this transport property , thereby providing an efficient way to directly evaluate the chern number of topologically - ordered optical lattices @xcite . [ [ detectedge ] ] detecting topological edge states in atomic chern insulators + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + topological edge states are populated when the fermi energy is located within a topological bulk gap @xcite . in chern insulators , as described by the hofstadter @xcite or haldane @xcite models , for example , this happens when the sum of chern numbers associated with the bulk bands lying below the gap is non - zero @xmath1253 according to the bulk - edge correspondence , the topological index @xmath50 corresponds to the number of edge - modes present within the bulk gap ( these modes are responsible for the quantized hall conductivity in eq . , see ref . ) .
the bulk - edge correspondence is illustrated in fig .
[ fig_edgestates]a . in general
, the number of occupied edge modes within a bulk gap @xmath1254 and below @xmath1214 contains a very small fraction of the total number of particles in the system @xmath1255 . in a circular atomic chern insulator , produced by an external confining potential @xmath1256 , the ratio of edge to total states is @xmath1257 , where @xmath1258 is the fermi radius of the system and @xmath779 is the lattice spacing @xcite . for typical systems with @xmath1259 and @xmath1260 particles
, only a few tens of atoms will occupy edge modes .
this simple observation indicates that the direct detection of topological edge states is a subtle and challenging task .
( from @xmath1261 to @xmath1262 ) , and with periodic boundary conditions along @xmath43 : the spectrum @xmath1263 is shown as a function of the quasi - momentum @xmath1184 for @xmath1264 . the projected bulk bands @xmath1265 , shown in fig .
[ fig_hofstadter]b are plotted in blue .
the red dispersion branches within the bulk gaps correspond to propagating edge modes near @xmath1261 and @xmath1262 .
when the fermi energy is within the first bulk gap , edge modes up to the fermi energy are populated , one each on the system s top and bottom .
the number of edge - state modes within each gap is in agreement with the chern numbers presented in fig .
[ fig_hofstadter ] b. ( b ) an atomic fermi gas with @xmath1214 as in ( a ) , was initially confined by a potential wall @xmath1266 in a small region located in the vicinity of the circular edge . after releasing the wall @xmath1266 , the edge states propagate along the circular edge created by the potential @xmath1267 , making them directly observable in in - situ images of the spatial density @xmath1268 @xcite .
, width=528 ] identifying bulk topological order using the system s edges requires measuring characteristics of topological edge states distinct from those of the many bulk states . in chern insulators ,
all the edge states present within a bulk gap propagate in the same direction
they are chiral and their dispersion relation is approximately linear .
therefore , with the fermi energy in a bulk gap , the dispersion relations of low - energy excitations give a clear signature for the presence ( or absence ) of topological edge states @xcite . in principle , bragg spectroscopy @xcite a technique based on momentum - sensitive light scattering which , for example , found application in measuring bec s collective modes could offer such a probe @xcite .
unfortunately , the number of particles @xmath1269 excited by a bragg probe focused on the cloud s edge ( e.g. using high - order laguerre - gaussian beams @xcite ) would be extremely small compared to the total number of particles @xmath1255 , making the bragg signal undetectable @xcite . to overcome this drawback ,
a shelving method " was proposed in refs .
@xcite . in this scheme , the bragg probe transfers energy and angular momentum to atoms located in the vicinity of the fermi radius and simultaneously changes their internal states .
this completely removes the edge states from the cloud allowing imaging on a dark background unpolluted by the untransferred atoms enabling the detecting of the edge - mode s dispersion relation from the bragg signal @xcite .
another method forces the edge states to propagate in a region that is unoccupied by the bulk states , allowing for a direct imaging of these topological states .
such a method was proposed in ref .
@xcite , where the atomic cloud is initially shaped by large repulsive walls and prepared in a chern insulating phase . after suddenly removing the walls ,
the chiral edge states propagate in a chiral manner along the circular edge of the cloud , while the bulk states tend to fill the initially vacant regions .
this method is particularly efficient starting from a topological flat band , in which case the bulk states remain immobile for long times , allowing for the clear imaging of the propagating edge states , see fig .
[ fig_edgestates]b .
schemes for isolating the edge state signal in the case of dispersive bulk states were also proposed in @xcite , based on their chiral nature . other methods to launch the edge state currents , by quenching the parameters of the microscopic hamiltonian , were proposed in ref .
@xcite .
topological edge states generally survive in smooth confining potentials @xcite
. however , their angular velocity dramatically decreases and their localization length increases as the potential is smoothened .
the use of sharp boundaries is therefore preferable to detect this edge - states physics in experiments @xcite .
the detection of atomic topological edge states has been further studied in refs .
@xcite .
finally , we stress that extending the schemes to probe topological invariants and edge states in static systems to the case of periodically - driven systems , should be handled with care .
indeed , any potential @xmath1270 associated with a probing protocol , e.g. a static force to measure the chern number , or a walking potential to probe edge - state dispersion relations , will potentially alter the effective hamiltonian , and hence , the quasi - energy band structure @xcite . in other words , measuring the topological order associated with an effective hamiltonian may destroy it . adding synthetic spin - orbit couplings to 2d and 3d optical lattices
opens the possibility to simulate and detect the unusual properties of @xmath1101 topological insulators @xcite .
the methods for detecting the chiral edge states of chern atomic insulators discussed above can be directly applied to the case of @xmath1101 insulators exhibiting helical edge states .
for instance , spin - resolved density measurements @xcite could be used to identify the propagation of the different spin species .
using cold atoms subjected to both synthetic spin - orbit couplings and magnetic fields , one could then identify the transition between qh and qsh phases , by studying the nature of the propagating edge states @xcite .
while the bragg spectroscopy scheme of refs .
@xcite could be generalized to identify the edge states of any 2d topological phase , the wall - removal " strategy of ref .
@xcite could be applied to any topological phase exhibiting propagating states ( in 2d , but also 3d systems ) .
moreover , the methods to directly measure the chern number could be extended to detect the topologically invariant spin - chern number of @xmath1101 insulators @xcite .
finally , 3d optical lattices emulating @xmath1101 topological insulators provide a versatile platform for detecting emerging axion electrodynamics : the unusual modifications of maxwell s equations due to the topological axion term @xcite .
a protocol to detect the fractional magnetic capacitor @xcite a signature of axion electrodynamics was described in ref .
+ the beautiful universality of topological band insulators motivated theorists to seek for similar structures in different physical systems .
for example , the bogoliubov - de gennes hamiltonian , describing the excitations of superconductors , can also describe topological phases : topological superconductors .
some topological superconductors those that break time - reversal ( tr ) symmetry host topologically - protected zero modes : majorana fermions with non - abelian exchange statistics @xcite ( akin to the quasiparticle excitations of the @xmath1271 quantum hall state @xcite ) .
several platforms have been envisaged in the quest for these properties , such as ( a ) the interface between a 3d topological insulator and a conventional @xmath780-wave superconductor @xcite , and ( b ) semiconductors with rashba spin - orbit coupling , @xmath780-wave pairing and a tr - breaking perturbation ( e.g. a zeeman coupling ) @xcite .
this second route has been envisaged for 1d spin - orbit coupled semiconducting wires that map onto kitaev s superconducting chain @xcite , but also for 2d spin - orbit coupled semiconductors @xcite .
neutral atoms with synthetic spin - orbit coupling therefore gives a natural and experimentally complementary platform for realizing tr - breaking topological superconductors , where s - wave pairing and tr breaking terms can be easily controlled .
in addition , raman coupling and laser - assisted tunneling methods can also give 2d topological superconductors @xcite and kitaev s superconducting chain with cold atoms trapped in 1d optical lattices @xcite .
these proposals are all motivated by the desire to clearly detect and manipulate the properties of atomic majorana zero modes , with an eye for methods that are not experimentally practical in solid - state systems @xcite .
a first signature might be the anomalous density of an atomic topological superfluid .
a majorana mode located inside a vortex core , should contribute to the total density in a detectable manner @xcite , and tof images would reveal the majorana mode s non - local correlations . by considering a topological atomic chain , kraus et al .
@xcite discussed how the the majorana mode s long - range correlations give modulations in the tof images , and also , that this signal is related to the number of topological modes present in the system .
as stressed above when discussing the detection of chern insulators , it is crucial to reduce the large background stemming from the many bulk states to emphasize the majorana signal , which could be realized by local addressing @xcite .
spectroscopic measurements , similar to the bragg probe discussed above , both gives access to the energy of the majorana states and provides a proof of their localization @xcite . finally , the chern number measurement introduced by alba et al .
@xcite could also be generalized to demonstrate the existence of majorana modes in optical lattices @xcite .
braiding operations , which reveal the anyonic nature of the majorana modes , have been proposed for 2d topological superfluids @xcite , where braiding is realized by externally moving the vortices hosting the zero modes .
more recently , a braiding protocol was proposed for atomic kitaev wires @xcite , where the braiding operations are realized through local addressing by locally switching on / off potentials , hopping and pairing terms .
[ sect : dynamical ] so far we have considered the effects of gauge fields or strictly speaking gauge _ potentials _ that are classical and static ( in the sense that any time - dependence of the field is experimentally specified ) . these gauge potentials are externally controlled , and thus , they typically depend on atom - light coupling parameters , on rotation frequency or on other types of external driving features .
as we have seen , effective magnetic and electric fields arise and give rise to observable effects , however , these _ applied _ fields do not reproduce a complete field - theory picture : they are not dynamical degrees of freedom .
in contrast to real " fields , the synthetic fields considered in the previous sections are not influenced by the matter fields ( i.e. the atoms ) . to be concrete , the synthetic electric and magnetic fields produced in laser - coupled atomic gases ( sect .
[ sect : lightmatter ] ) need not obey maxwell s equations .
this aspect of synthetic gauge potentials emphasizes their artificial origin , and in fact , it could be exploited constructively to simulate exotic situations where electromagnetism is no longer ruled by maxwell s equations .
+ generating _ interacting gauge theories _ with cold atoms , where matter fields and synthetic gauge structures are dynamically related , importantly connects simple experiments with intractable problems in qed and qcd @xcite .
such quantum simulators are a novel tool for gaining physical insight in issues encountered in high - energy physics , such as the fundamental problem of _ confinement _ , which precludes the observation of free quarks in nature @xcite .
a first step towards realizing dynamical gauge fields is introducing back - action where atoms affect a synthetic gauge potential @xmath1272 locally and dynamically .
one possibility is to create a gauge potential that explicitly depends on the atomic density @xmath1273 , as was proposed in refs .
these density - dependent gauge structures , ( described in sect . [ sect : dens - dependent ] ) , do not fully reproduce a conventional gauge field theory in the sense that the emerging fields do not exist in the absence of matter ( i.e. when @xmath1274 ) . however , they give rise to interesting properties , such as anyonic structures and chiral solitons , suggesting novel perspectives in quantum simulation .
schemes that fully realize quantum field theories @xcite require ( a ) quantum matter ( fermionic ) fields @xmath1275 _ and _ gauge ( bosonic ) fields @xmath1276 represented by different atomic species interacting with each other , and ( b ) gauge - invariance conditions usually synthesized by imposing some constraints .
these proposals highlight the birth of a very new and exciting field of research possibly connecting cold - atom and high - energy physicists , and are discussed in sect .
[ sect : qgt ] .
a recent review on the quantum simulation of lattice gauge theories has been recently written by wiese @xcite . finally , we mention the possibility to realize dynamical gauge fields of condensed matter models , such as spin - ice materials , using polar molecules @xcite or rydberg atoms @xcite .
we first consider a non - interacting two component bose gas evolving in space - dependent coupling fields described by the single - particle hamiltonian h = + u = + 0 & e^-i + e^i & 0 , where @xmath464 is the raman rabi frequency ; @xmath462 is the coupling laser s phase ; and for simplicity the detuning @xmath352 from the two - photon raman resonance is set to zero . according to sect .
[ sect : lightmatter ] , a non - trivial gauge structure is generated through the atom - light coupling , when the gradient of the detuning is non - zero @xmath1277 .
the general idea behind the concept of density - dependent synthetic gauge potentials , is that collisions between the atoms can induce an effective _ density - dependent detuning _ at the mean - field level .
reference @xcite considered the mean - field hamiltonian h = + u + v_mf = + g_11|_1|^2 + g_12|_2|^2 & e^-i + e^i & g_22|_2|^2 + g_12|_1|^2 , for an interacting two - level atomic system , where @xmath1278 are the densities associated with the two atomic species , and @xmath1279 are the species - dependent contact - interaction parameters . when the atom - light coupling energy @xmath1280 is much larger than the mean field terms @xmath1281 , the corresponding gauge potential and dressed states @xmath1282 can be obtained perturbatively .
@xmath1283 are the standard unperturbed dressed states for @xmath1284 , and @xmath1285 denote the dressed state densities .
equation highlights the main result : species - dependent collisions , with @xmath1286 , can produce a density - dependent gauge potential @xmath1287 .
this interaction - induced detuning is the simplest scheme realizing a pseudo - dynamical gauge theory with back action between the matter field @xmath809 and the gauge potential @xmath1288 .
the key point is that the parameters in the resulting density - dependent gauge potential @xmath903 are governed by the rabi frequency @xmath464 , the gradient of the phase @xmath1289 and the scattering length difference @xmath1290 .
all these parameters are largely adjustable by tuning the coupling lasers and the scattering lengths , which can be achieved using feshbach resonances @xcite .
the properties of this unusual bose gas can be studied through a generalized gross - pitaevskii ( gp ) equation , which takes into account the presence of the density - dependent gauge potential @xcite @xmath1291\psi,\label{cgp}\ ] ] where @xmath1292 is given by eq .
( [ nonlin ] ) , @xmath1293 , @xmath1294 and a single dressed - state branch has been isolated in the dynamics ( i.e. @xmath1295 , @xmath1296 ) . here
, the nonlinearity of the gp equation manifests itself both through the standard term @xmath1297 , and also through the current @xmath1298.\label{currentgp}\ ] ] thus , this gives rise to rich nonlinear dynamics : the modified gp equation - already leads to exotic properties in 1d , including chiral soliton solutions , as was already pointed out in the context of one - dimensional anyons @xcite . supposing a coupling laser phase in the form @xmath1299 and setting @xmath1300 , the dynamics of the 1d atomic system is described by i _
t = , [ 1dsoliton ] where @xmath1301 is the @xmath890-dependent nonlinearity strength and @xmath1302 .
this equation supports bright or dark chiral solitons , depending on the sign of @xmath1303 , where @xmath1304 is the speed of the soliton . at the critical value @xmath817 , the chiral soliton is destroyed .
hence , in contrast to a conventional soliton , the chiral solitons are strongly altered when reflected from a hard wall @xcite .
other non - trivial effects stemming from the unusual nonlinear equation have been reported in ref .
@xcite , including an asymmetric free expansion of the cloud accompanied with a drift of the centre of mass , and the existence of critical particle numbers for the onset of persistent currents in a ring geometry @xcite . + a density - dependent gauge structure can also be created in optical - lattices . in the lattice framework
( see sect .
[ sect : lattices ] ) , gauge potentials manifest themselves through the peierls - modified hopping matrix element @xmath1305 , describing hopping between lattice sites @xmath140 and @xmath890 . hence , the key idea is to engineer peierls phases @xmath1306 that depend on the occupation number @xmath1307 at each lattice site @xmath140 , e.g. @xmath1308 .
such a scenario was proposed by keilmann et al .
@xcite , giving the generalized bose - hubbard model @xmath1309 this is the lattice version of the hamiltonian leading to eq .
, contains the density operator @xmath1310 at site @xmath140 , the onsite interaction strength @xmath872 , and the @xmath1311-dependent tunneling rate @xmath1312 .
this unusual peierls substitution can be realized using a generalization of the laser - assisted tunneling method presented in sect .
[ experimentallattice ] , as illustrated in fig .
[ keil ] for the case @xmath1313 . in this picture , the two neighboring sites at @xmath140 and @xmath1314 can either host @xmath1175 or @xmath1315 atoms ( omitting the trivial case @xmath828 ) , resulting in a four - dimensional ground state manifold @xmath1316 .
coupling this subsystem to an excited state @xmath1053 , with four different coupling fields , allows to individually address each hopping process , thanks to the energy offsets produced by the onsite interaction @xmath872 and by an additional lattice tilt @xmath425 , see figure [ keil ] . finally , tuning the phases associated with each laser - induced tunneling process ( see sect .
[ experimentallattice ] ) results in the required @xmath1317-dependent tunneling matrix elements between lattice sites @xmath140 and @xmath1314 @xcite . a schematic view of the proposal by keilmann et al . for creating density dependent tunneling matrix elements @xmath1318 , for the case @xmath1313 .
laser assisted tunneling using four detuned ( @xmath352 ) lasers in combination with a tilted lattice given by the off - set @xmath425 induces a phase which depends on the occupation number of the target site . ] the hamiltonian maps onto a theory of anyons in one - dimensional lattice .
the mapping between bosonic ( @xmath1319 ) and anyonic ( @xmath1320 ) operators , @xmath1321 $ ] , gives rise to the anyon - hubbard hamiltonian @xmath1322 where the commutation relation for @xmath1323 is anyonic : @xmath1324 since @xmath1325 when @xmath1326 , particles behave as bosons on - site , and as anyons off - site with statistical angle @xmath461 .
thus , the anyonic nature of eq . stems from the density - dependent peierls phases @xmath1327 .
the hamiltonian in eq . is a lattice version of the continuum hamiltonian leading to the meanfield gross - pitaevskii equation in eq . .
as such it shows unconventional tunneling dynamics due to its nonlinearities ( see eqs .
- ) , which will affect any josephson type dynamics in the lattice @xcite .
reference @xcite showed that density - dependent tunneling gives rise to phase transitions between the standard superfluid phase and exotic mott states , where the particle distributions show plateaus at fractional densities , due to the anionic statistics .
the simulation of dirac fermions with atoms in optical lattices @xcite , together with the possibility of generating background abelian and non - abelian gauge fields @xcite , suggest that cold atoms could be exploited to deepen our understanding of quantum electrodynamics ( qed ) and quantum chromodynamics ( qcd ) : a powerful alternative to numerical lattice - gauge - theory ( lgt ) .
for instance , cold atoms in optical lattices of various spatial dimensions ( @xmath1328 @xcite and beyond @xmath1329 @xcite ) may reveal the mechanisms by which confined phases emerge in various configurations : in abelian and non - abelian gauge theories ; or in @xmath1330 and @xmath1331 space - time dimensions .
these mechanisms could be investigated in a setup where different atomic species encode the matter and the gauge degrees of freedom , and exploit the fact that the interactions between the particles can be tuned ( e.g. by feshbach resonances ) .
such cold - atom quantum simulators are a physical platform where the coupling strength of the gauge theory is externally controllable , and where phase transitions between confined and unconfined phases might be externally driven @xcite .
this quantum - simulation scenario requires that both the matter and the gauge fields be dynamical , and also , that the simulated theory be gauge invariant .
for instance , realizing qed with cold atoms requires that the gauss s law be imposed by a constraint , which generally requires a precise control over the simulated hamiltonian @xcite .
the versatility of cold - atom systems also allows for the simulation of simple toy models , such as the gross - neveu model , which played a major role in the exploration of qcd - like effects @xcite .
the quantum simulation of the gross - neveu model @xcite already captures the general strategy and the main ingredients needed to simulate more elaborate field theories , such as qed and qcd .
the gross - neveu model describes the interaction between a massless dirac fermion @xmath785 and a massive quantized scalar field @xmath1332 in @xmath1333 spatial dimension , with the hamiltonian h= dx , [ grossneveu ] where @xmath270 is a velocity , @xmath802 is the coupling strength , @xmath89 is the mass of the scalar field and @xmath1334 is a dirac matrix ( the color quantum numbers @xmath981 of the dirac field @xmath1335 are implicit in eq . ) . in this simple field theory
the scalar field @xmath1332 has no kinetic term , and the coupling between the dirac @xmath785 and the scalar field @xmath1332 can be formally traced back to the original gross - neveu model ( which describes dirac fermions interacting through the term @xmath1336 in the absence of the scalar field ) @xcite .
this interacting field theory could be implemented with a @xmath83-component fermi gas in an optical lattice featuring spatially - modulated tunneling amplitudes @xmath1337 , see fig .
[ grossfig ] , and along with an independent bec loaded into a separate optical lattice with double spacing , as represented in fig .
[ grossfig ] .
the resulting bose - hubbard hamiltonian is h= - _ j - u ( b_j^)^2 ( b_j)^2 - b_j^ b_j , [ grossoptical ] where @xmath1338 ( resp .
@xmath1339 ) creates a fermion at lattice site @xmath1340 ( resp .
@xmath1341 ) ; @xmath1342 creates a boson at lattice site @xmath140 ; @xmath872 is the onsite interaction strength for bosons ; @xmath45 is the chemical potential ; and @xmath352 denotes the strength of the fermion - boson on - site " interaction , see fig .
[ grossfig ] . according to ref .
@xcite , when the fermi energy is at half - filling , the low - energy excitations exactly reproduce the dynamics dictated by for specific values of the optical - lattice parameters .
since the matter - gauge field coupling strength @xmath802 of the simulated field theory results from a boson - mediated tunneling process of fermions that is proportional to the inter - species interaction strength @xmath1343 @xcite , the coupling constant @xmath802 is tunable .
+ optical - lattice implementation of eq . .
a primary optical lattice ( blue ) traps fermions in the two inequivalent sites @xmath1097 and @xmath1098 , which are distinguished by the alternating tunneling rates @xmath1344 . a superimposed optical lattice ( purple )
traps a bec between the @xmath1097 and @xmath1098 sites of each unit cell , labeled by the index @xmath140 .
the interaction between fermions and bosons within each unit cell @xmath140 results in a boson - mediated tunneling process for the fermions , giving rise to an effective quantum field theory at low energy @xcite . ] realizing more involved field theories , such as lattice qed or non - abelian yang - mills lattice gauge theories , requires additional ingredients and control over inter - species interactions to retain gauge invariance @xcite .
these more complex proposals are still based on the strategy outlined above : different atomic species , fermions and bosons , are trapped in interpenetrating optical lattices , forming a many - body lattice system , whose effective ( low - energy ) description is that of an interacting quantum field theory .
these elaborate cold - atom schemes are largely based on available technologies taken to the extreme suggesting that experimental realizations of lattice gauge theories are within reach .
very recently , a scheme has been proposed where the gauss s law and gauge invariance emerge naturally ( i.e. they are not imposed by an external constraint that modifies the effective low - energy description ) , based on the fundamental symmetries of the atomic system @xcite .
optical - lattice implementations of lattice gauge theories offer unique measurement tools .
experimenters might observe the mass dynamically acquired by fermions in the gross - neveu model @xcite , the electric flux tube " at the origin of confinement in @xmath1345 abelian lgt @xcite , the wilson - loop area law @xcite , cosmological inflation processes @xcite and the nonperturbative string breaking effect present in quantum link models and qcd @xcite .
[ sect : conclusion ] artificial gauge potentials in cold gases are a new laboratory tool that allow access to new physical phenomena . a handful of techniques for inducing the gauge potentials have been proven in the lab ; these range from optically addressing the internal level structure of the atoms in combination with orbital motion , to mechanical shaking of optical lattices . in particular , synthetic uniform magnetic fields
were realized leading to quantized vortices in bec @xcite , and non - trivial lattice gauge structures have been produced in optical lattices , leading to staggered @xcite and uniform @xcite magnetic flux configurations .
additionally , the ( one - dimensional ) spin - orbit coupling has now been created and its manifestations have been explored for atomic bosons and fermions @xcite .
examples of recent applications of artificial gauge potentials also include thermometry of cold atoms in optical lattices @xcite , mapping the berry curvature and chern numbers of bloch bands @xcite , efimov states with soc @xcite and gauge potentials in rydberg gases @xcite .
artificial electromagnetism has begun to mature .
it can be applied with or without an optical lattice present , leading to a broad spectrum of applications . at the single particle level , a surge of theoretical work using optical lattices
propose to emulate exotic condensed matter systems .
prominent examples include topological insulators , quasi - relativistic systems , integer quantum hall physics , and non - trivial spin dynamics from spin - orbit coupling . including collisional interactions enables a whole new field of research .
interacting bosonic and fermionic gases show intriguing new effects when gauge fields are present . in the abelian case , fractional quantum hall ( fqh ) physics
could be realized . to generate and stabilize such fqh liquids , it is crucial to develop schemes leading to topological flat bands @xcite , namely , dispersionless bands characterized by non - zero chern numbers , which are well separated with respect to higher energy bands [ see fig . [ fig_edgestates ] for such a configuration offered by the hofstadter model at flux @xmath1264 ] .
note that such a topological - flat - band configuration has been realized experimentally in munich @xcite , where a chern number @xmath1347 has been measured .
moreover , the rich set of spin - orbit coupled systems enable a range of novel effects such as unconventional superfluid properties in bcs - bec crossover regimes , and finite momentum ground states for fermi gases .
interacting spin - orbit coupled gases also offer an interesting route towards interacting topological phases in 3d , namely , the high - dimensional cousins of fractional quantum hall states .
the prospect of creating non - trivial topological states of matter also opens up the possibility to address fundamental questions in quantum information .
the availability of orbital magnetism provides a novel tool for atomtronics situations , i.e. , the atomic version of spintronics .
the non - trivial topological states together with the concept of non - abelian dynamics may also provide the route towards the mercurial and highly anticipated topological quantum computer @xcite , which promises to be the holy grail of fault tolerant quantum computing .
quantum simulators would in general benefit from artificial orbital magnetism .
the gauge fields might enable the simulation of more exotic gauge theories that are computationally intractable .
a prominent example linked to the ability to create dynamical gauge fields is the quantum chromodynamics description of the strong force between elementary particles .
while dynamical gauge fields are still in their infancy and unrealized in the lab , recent advances and theoretical proposals using optical lattices show that it is in principle possible to engineer fully dynamical gauge fields .
it is intriguing to speculate on the future direction for artificial gauge fields .
first of all there are challenges both theoretically and experimentally which need to be addressed . from a theoretical point of view
perhaps the most pressing question is to what extent the gauge fields can be combined with strong collisional interactions and whether there are any fundamental or practical limitations for doing so and still retain the dynamics governed by the effective magnetic fields .
an interesting and promising direction in this respect is the inclusion of cavity qed where also strong interaction between the light and matter can be achieved which may provide a mechanism for creating a back - action between the matter field and the effective gauge fields @xcite , and by doing so also address the question of creating dynamical gauge fields but now possibly in the quantum regime .
experimentally the challenges are also many .
the level of controllability of key parameters such as stable laser frequencies , collisional properties , maintaining ultralow temperatures , avoiding numerous sources of heating and decay of the prepared quantum superpositions , are certainly taxing , and will require a constant development and refining of experimental techniques and technologies .
this will have implications for reaching strongly correlated regimes in experiments .
all techniques for creating artificial magnetic fields to date could in principle reach the strongly correlated regime such as quantum hall liquids .
each one of them have however their own experimental drawbacks .
for instance the raman based proposals including also flux lattices , will have problems with spontaneous emission which will cause heating .
the mechanical shaking of a lattice will also induce heating eventually .
finally the rotational techniques have the disadvantage that there one would have to work with a very small number of particles .
these limitations are all of a technical character , and we are likely to see a development of all them in parallel in the future . on a more speculative note artificial electromagnetism may provide a route to completely new physical scenarios which can not be be found in conventional solid state systems . emulating solid - state phenomena with the hope to shed some light on new and not so well understood phenomena , is the most natural application of the artificial gauge fields .
it is however not inconceivable to envisage a situation where the effective gauge field allows for the preparation of highly exotic quantum states which captures not only the solid state and condensed matter phenomena , but also addresses fundamental question regarding the fabric of space - time . for instance , gauge fields in the presence of analog gravity in bose - einstein condensates @xcite could address important questions , such as the role of quantum entanglement and unitarity near black holes .
the inclusion of gauge fields , and in particular dynamical gauge fields , into the picture of analog gravity , is largely uncharted territory and may provide some insight into the interplay between the quantum world and gravity .
we acknowledge helpful discussions with m. aidelsburger , b. anderson , e. anisimovas , l. aycock , m. a. baranov , r. barnett , g. barnich , a. bermudez , j. beugnon , i. bloch , n. r. cooper , j. dalibard , m. dalmonte , a. dauphin , a. eckardt , m. edmonds , t. esslinger , v. galitski , p. gaspard , f. gerbier , p. hauke , w. ketterle , c. v. kraus , z. lan , j. larson , m. lewenstein , d. maldonado - mundo , s. nascimbene , m. rechtsman , j. ruostekoski , j. ruseckas , l. santos , m. valiente , h. zhai and p. zoller .
n. g. thanks the frs - fnrs and the ulb for financial support .
g. j. acknowledges the financial support by the lithuanian research council project no .
p. . acknowledges support from the uk epsrc grant no .
ep / j001392/1 and the carnegie trust for the universities of scotland .
i. b. s. acknowledges the financial support by the nsf through the physics frontier center at jqi , and the aro with funds from both the atomtronics muri and darpa s ole program .
kundhikanjana w , wang j , calvo m r , lian b , yang y , lai k , baenninger m , knig m , ames c , brne c , buhmann h , leubner p , tang q , zhang k , molenkamp l w , zhang s c , goldhaber - gordon d , kelly m a and shen z x 2012 _ arxiv:1212.6441 _ roscilde t 2014 _ phys . rev . lett . _
* 112 * , 110403 duca l , li t , reitter m , bloch i , schleier - smith m , and schneider u 2014 _ arxiv:1407.5635 _ xiaoling cui and wei yi 2014 _ phys . rev . x _ * 4 * 031026 zhe - yu shi , xiaoling cui , and hui zhai 2014 _ phys
_ * 112 * 013201 zygelman b 2013 _ j. phys .
* 134011 kiffner m , li w , and jaksch d 2013 _ phys . rev . lett .
_ * 110 * , 170402 grusdt f and fleischhauer m 2013 _ phys . rev .
a _ * 87 * , 043628 cesa a and martin j 2013 _ phys .
a _ * 88 * , 062703
|
gauge fields are central in our modern understanding of physics at all scales . at the highest energy scales known , the microscopic universe is governed by particles interacting with each other through the exchange of gauge bosons . at the largest length scales , our universe is ruled by gravity , whose gauge structure suggests the existence of a particle the graviton that mediates the gravitational force . at the mesoscopic scale ,
solid - state systems are subjected to gauge fields of different nature : materials can be immersed in external electromagnetic fields , but they can also feature emerging gauge fields in their low - energy description .
+ in this review , we focus on another kind of gauge field : those _ engineered _ in systems of ultracold neutral atoms . in these setups , atoms
are suitably coupled to laser fields that generate effective gauge potentials in their description .
neutral atoms feeling " laser - induced gauge potentials can potentially mimic the behavior of an electron gas subjected to a magnetic field , but also , the interaction of elementary particles with non - abelian gauge fields . here , we review different realized and proposed techniques for creating gauge potentials both abelian and non - abelian in atomic systems and discuss their implication in the context of quantum simulation . while most of these setups concern the realization of background and classical gauge potentials , we conclude with more exotic proposals where these synthetic fields might be made dynamical , in view of simulating interacting gauge theories with cold atoms .
|
we consider principal component analysis ( pca ) in gaussian graphical models .
pca is a classical dimensionality reduction method which is frequently used in statistics and machine learning @xcite .
the first principal components of a multivariate are its orthogonal linear combinations which preserve most of the variance . in the gaussian case
, pca has special properties which make it especially favorable : it is the best linear approximation of the data and it provides independent components . on the other hand , gaussian graphical models , also known as covariance selection models , provide a graphical representation of the conditional independence structure within the gaussian distribution @xcite . exploiting the extensive knowledge and literature on graph theory , graphical models allow for efficient distributed implementation of statistical inference algorithms , e.g. , the well known belief propagation method and the junction tree algorithm @xcite . in particular , decomposable graphs , also known as chordal or triangulated graphs ,
provide simple and intuitive inference methods due to their appealing structure .
our main contribution is the application of decomposable graphical models to pca which we nickname dpca , where d denotes both _
decomposable _ and _
distributed_. the main motivation for distributed pca is decentralized dimensionality reduction .
it plays a leading role in distributed estimation and compression theory in wireless sensor networks @xcite , and decentralized data mining techniques @xcite .
it is also used in anomaly detection in computer networks @xcite . in particular
, @xcite proposed to approximate the global pca using a sequence of conditional local pca solutions .
alternatively , an approximate solution which allows a tradeoff between performance and communication requirements was proposed in @xcite using eigenvalue perturbation theory .
dpca is an efficient implementation of distributed pca based on a prior graphical model . unlike the above references it does not try to approximate pca , but yields an exact solution up to on any given tolerance .
on the other hand , it assumes additional prior knowledge in the form of a graphical model which previous works did not take into account .
although , it is interesting to note that the gauss markov source example in @xcite is probably the most celebrated decomposable graphical model .
therefore , we now address the availability of such prior information .
in general , practical applications do not necessarily satisfy any obvious conditional independence structure . in such scenarios , dpca can be interpreted as an approximate pca method that allows a tradeoff between accuracy and decentralization by introducing sparsity . in other problems
it is reasonable to assume that an unknown structure exists and can be learned from the observed data using existing methods such as @xcite .
alternatively , a graphical model can be derived from non - statistical prior knowledge on the specific application .
an intuitive example is distributed networks in which the topology of the network suggests a statistical graph as exploited in @xcite .
finally , we emphasize that even if a prior graphical model is available , it does not necessarily satisfy a decomposable form . in this case
, a decomposable approximation can be obtained using classical graph theory algorithms @xcite .
pca can be interpreted as maximum likelihood ( ml ) estimation of the covariance using the available data followed by its eigenvalue decomposition .
when a prior graphical model is available , pca can still be easily obtained by adjusting the ml estimation phase to incorporate the prior conditional independence structure using existing methods @xcite , and then computing the eigenvalue decomposition ( evd ) .
the drawback to this approach is that it does not exploit the structure of the graph in the evd phase .
this disadvantage is the primary motivation to dpca which is specifically designed to utilize the structure of gaussian graphical models .
decomposable covariance selection models result in sparse concentration ( inverse covariance ) matrices which can be estimated in a decentralized manner .
therefore , we propose to reformulate dpca in the concentration domain and solve the global evd using a sequence of local evd problems in each of the cliques of the decomposable graph with a small amount of message passing .
this allows for distributed implementation according to the topology of the graph and reduces the need to collect all the observed data in a centralized processing unit .
when the algorithm terminates , each clique obtains its own local version of the principal components . to illustrate dpca we apply it to distributed anomaly detection in computer networks @xcite . in this context
, dpca learns a low dimensional model of the normal traffic behavior and allows for simple outlier detection .
this application is natural since the network s topology provides a physical basis for constructing an approximate a graphical model .
for example , consider two nodes which are geographically distant and linked only through a long path of nodes .
it is reasonable to believe that these two sensors are independent conditioned on the path , but a theoretical justification of this assertion is difficult and depends on the specific problem formulation .
we examine the validity of this claim in the context of anomaly detection in the abilene network using a real - world dataset .
we propose an approximate decomposition of the abilene network , enable the use of dpca and obtain a fully distributed anomaly detection method .
the outline of the paper is as follows .
decomposable graphs are easy to explain using a special graph of two cliques which is their main building block .
therefore , we begin in section [ sec_2dpca ] by introducing the problem formulation and solution to dpca in this simple case .
the generalization to decomposable graphs is presented in section [ sec_dpca ] which consists of their technical definitions followed by a recursive application of the two cliques solution .
we demonstrate the use of dpca using two numerical examples .
first , in section [ sec_num ] we simulate our proposed algorithm in a synthetic tracking scenario .
second , in section [ sec_abilene ] we illustrate its application to anomaly detection using a real - world dataset from the abilene backbone network .
finally , in section [ sec_conc ] we provide concluding remarks and address future work .
the following notation is used .
boldface upper case letters denote matrices , boldface lower case letters denote column vectors , and standard lower case letters denote scalars .
the superscripts @xmath0 and @xmath1 denote the transpose and matrix inverse , respectively .
the cardinality of a set @xmath2 is denoted by @xmath3 .
the matrix @xmath4 denotes the identity , @xmath5 is the minimum eigenvalue of square symmetric matrix @xmath6 , @xmath7 is a null vector of @xmath6 , @xmath8 is the maximum eigenvalue of @xmath6 , and @xmath9 means that @xmath6 is positive definite . finally , we use indices in the subscript @xmath10}_a$ ] or @xmath11}_{a , b}$ ] to denote sub - vectors or sub - matrices , respectively , and @xmath11}_{a,:}$ ] denotes the sub - matrix formed by the @xmath2th rows in @xmath6 .
where possible , we omit the brackets and use @xmath12 or @xmath13 instead .
in this section , we introduce dpca for a simple case which will be the building block for the general algorithm . and @xmath14 are conditionally independent given @xmath15 .
[ fig_twoclique],scaledwidth=40.0% ] let @xmath16}^t$ ] be a length @xmath17 , zero mean gaussian random vector in which @xmath10}_a$ ] and @xmath10}_b$ ] are independent conditionally on @xmath10}_c$ ] where @xmath2 , @xmath15 and @xmath14 are disjoint subsets of indices . for later use , we use graph terminology and define two cliques of indices @xmath18 and @xmath19 coupled through the separator @xmath20 ( see fig . [ fig_twoclique ] ) .
we assume that the covariance matrix of @xmath21 is unknown , but the conditional independence structure ( defined through index sets @xmath22 and @xmath23 ) is known .
the input to dpca is a set of @xmath24 independent and identically distributed realizations of @xmath21 denoted by @xmath25 for @xmath26 .
more specifically , this input is distributed in the sense that the first clique has access to @xmath27}_{c_1}$ ] for @xmath26 , whereas the second clique has access only to @xmath27}_{c_2}$ ] for @xmath26 . using this data and minimal message passing between the two cliques , dpca searches for the linear combination @xmath28 having maximal variance .
when the algorithm terminates , each of the cliques obtains its own local version of @xmath29 , i.e. , the sub - vectors @xmath30}_{c_1}$ ] and @xmath30}_{c_2}$ ] .
the following subsections present the proposed solution to dpca .
it involves two main stages : covariance estimation and principal components computation .
first , the covariance matrix of @xmath21 is estimated using the maximum likelihood ( ml ) technique .
due to the known conditional independence structure , the ml estimate has a simple closed form solution which can be computed in a distributed manner ( more details about this procedure can be found in @xcite ) . each clique and
the separator computes their own local sample covariance matrices @xmath31}_{c_1}{\left[}{\mathbf{x}}_i{\right]}_{c_1}^t\\
\tilde{\mathbf{s}}^{c_2,c_2}&=&\frac{1}{n}\sum_{i=1}^n{\left[}{\mathbf{x}}_i{\right]}_{c_2}{\left[}{\mathbf{x}}_i{\right]}_{c_2}^t\\ \tilde{\mathbf{s}}^{s , s}&=&\frac{1}{n}\sum_{i=1}^n{\left[}{\mathbf{x}}_i{\right]}_{s}{\left[}{\mathbf{x}}_i{\right]}_{s}^t,\end{aligned}\ ] ] where the tilde and the superscripts are used to emphasize that these are local estimates .
similarly , the local concentration matrices , also known as precision matrices , are defined as @xmath32 where it is assumed that the matrices are non - singular ( otherwise , the ml estimate does not exist ) .
next , the global ml concentration matrix @xmath33 is obtained by requiring @xmath34 due to the conditional independence of @xmath12 and @xmath35 given @xmath36 .
the global solution is @xmath37}+{\left[}\begin{array}{cc } { \mathbf{0 } } & \begin{array}{cc } { \mathbf{0 } } & { \mathbf{0}}\end{array } \\
\begin{array}{c } { \mathbf{0}}\\ { \mathbf{0}}\end{array } & \tilde{\mathbf{k}}^{c_2,c_2 } \end{array } { \right]}-{\left[}\begin{array}{ccc } { \mathbf{0 } } & { \mathbf{0 } } & { \mathbf{0}}\\ { \mathbf{0 } } & \tilde{\mathbf{k}}^{s , s } & { \mathbf{0}}\\ { \mathbf{0 } } & { \mathbf{0 } } & { \mathbf{0}}\end{array } { \right]}.\end{aligned}\ ] ] it is easy to see that the sub - matrices associated with the cliques are perturbations of to their local versions : @xmath38}_{c_1,c_1}&=&\tilde{\mathbf{k}}^{c_1,c_1}+{\left[}\begin{array}{cc } { \mathbf{0 } } & { \mathbf{0}}\\ { \mathbf{0 } } & { \mathbf{m}}_{b } \end{array } { \right]}\\ \label{ksubblocks2 } { \left[}{\mathbf{k}}{\right]}_{c_2,c_2}&=&\tilde{\mathbf{k}}^{c_2,c_2}+{\left[}\begin{array}{cc } { \mathbf{m}}_{a } & { \mathbf{0}}\\ { \mathbf{0 } } & { \mathbf{0}}\end{array}{\right]}\end{aligned}\ ] ] and require only message passing via @xmath39 and @xmath40 : @xmath41}_{s , s}-\tilde{\mathbf{k}}^{s , s}\\ { \mathbf{m}}_{a}&=&{\left[}\tilde{\mathbf{k}}^{c_1,c_1}{\right]}_{s , s}-\tilde{\mathbf{k}}^{s , s}.\end{aligned}\ ] ] the dimension of these messages is equal to @xmath42 which is presumably small .
thus , the global ml concentration matrix can be easily found in a distributed manner .
the global covariance estimate is simply defined as the inverse of its concentration @xmath43 .
it is consistent with the local estimates of its sub - matrices : @xmath44}_{c_1,c_1}&=&\tilde{\mathbf{s}}^{c_1,c_2}\\ { \left[}{\mathbf{s}}{\right]}_{c_2,c_2}&=&\tilde{\mathbf{s}}^{c_2,c_2},\end{aligned}\ ] ] but there is no special intuition regarding its @xmath45}_{a , b}$ ] and @xmath45}_{b , a}$ ] sub - blocks .
given the global ml covariance estimate @xmath46 , the pca objective function is estimated as @xmath47 which is maximized subject to a norm constraint to yield @xmath48 this optimization gives both the maximal eigenvalue of @xmath46 and the its eigenvector @xmath29 .
the drawback to the above solution is that the evd computation requires centralized processing and does not exploit the structure of @xmath33 .
each clique needs to send its local covariance to a central processing unit which constructs @xmath46 and computes its maximal eigenvalue and eigenvector .
we will now provide an alternative distributed dpca algorithm in which each clique uses only local information along with minimal message passing in order to calculate its local version of @xmath49 and @xmath29 .
our first observation is that dpca can be equivalently solved in the concentration domain instead of the covariance domain .
indeed , it is well known that @xmath50 when the inverse @xmath51 exists .
the corresponding eigenvectors are also identical .
the advantage of working with @xmath33 instead of @xmath46 is that we can directly exploit @xmath33 s sparsity as expressed in ( [ kab ] ) . before continuing
it is important to address the question of singularity of @xmath46 .
one may claim that working in the concentration domain is problematic since @xmath46 may be singular .
this is indeed true but is not a critical disadvantage since graphical models allow for well conditioned estimates under small sample sizes .
for example , classical ml exists only if @xmath52 , whereas the ml described above requires the less stringent condition @xmath53 @xcite .
in fact , the ml covariance is defined as the inverse of its concentration , and thus the issue of singularity is an intrinsic problem of ml estimation rather than the dpca solution .
we now return to the problem of finding @xmath54 in a distributed manner .
we begin by expressing @xmath55 as a trivial line - search problem : @xmath56 and note that the objective is linear and the constraint set is convex .
it can be solved using any standard line - search algorithm , e.g. bisection . at first
, this representation seems useless as we still need to evaluate @xmath57 which was our original goal .
however , the following proposition shows that checking the feasibility of a given @xmath58 can be done in a distributed manner .
let @xmath33 be a symmetric matrix with @xmath59 .
then , the constraint @xmath60}{\right)}\end{aligned}\ ] ] is equivalent to the following pair of constraints @xmath61}{\right)}\end{aligned}\ ] ] with the _ message matrix _ defined as @xmath62 the proof is obtained by rewriting ( [ eigasopt1 ] ) as a linear matrix inequality @xmath63}-t{\mathbf{i}}\succ{\mathbf{0}}\end{aligned}\ ] ] and decoupling this inequality using the following lemma : [ schur ] let @xmath6 be a symmetric matrix partitioned as @xmath64}.\end{aligned}\ ] ] then , @xmath9 if and only if @xmath65 and @xmath66 . applying schur s lemma to ( [ eig2lmi ] ) with @xmath67 and rearranging yields @xmath68}.\end{aligned}\ ] ] finally , ( [ prop2 ] ) and ( [ prop1 ] ) are obtained by rewriting ( [ tikbb ] ) and ( [ tikac ] ) as eigenvalue inequalities , respectively .
proposition 1 provides an intuitive distributed solution to ( [ eigasopt ] ) . for
any given @xmath58 we can check the feasibility by solving local eigenvalue problems and message passing via @xmath69 whose dimension is equal to the cardinality of the separator .
the optimal global eigenvalue is then defined as the maximal globally feasible @xmath58 .
we note that the solution in proposition 1 is asymmetric with respect to the cliques .
the global constraint is replaced by two local constraints regarding clique @xmath18 and the remainder @xmath70 .
alternatively , we can exchange the order and partition the indices into @xmath71 and @xmath19 .
this asymmetry will become important in the next section when we extend the results to general decomposable graphs .
after we obtain the minimal eigenvalue @xmath55 , we can easily recover its corresponding eigenvector @xmath29 .
for this purpose , we define @xmath72 and obtain @xmath73 .
the matrix @xmath74 follows the same block sparse structure as @xmath33 , and the linear set of equations @xmath75 can be solved in a distributed manner .
there are two possible solutions .
usually , @xmath76 is non - singular in which case the solution is @xmath77}_{c_1}&=&{\mathbf{u}}_{\rm{null}}{\left(}{\mathbf{q}}_{c_1,c_1}-{\left[}\begin{array}{cc } { \mathbf{0 } } & { \mathbf{0}}\\ { \mathbf{0 } } & { \mathbf{m}}\end{array } { \right]}{\right)}\\ \label{ub } { \left[}{\mathbf{u}}{\right]}_b&=&-{\mathbf{q}}_{b , b}^{-1}{\mathbf{q}}_{b , c}{\left[}{\mathbf{u}}{\right]}_c,\end{aligned}\ ] ] where the _ message _ @xmath78 is defined as @xmath79 otherwise , if @xmath80 is singular then the solution is simply @xmath81}_{c_1}&=&{\mathbf{0}}\\ { \left[}{\mathbf{u}}{\right]}_b&=&{\mathbf{u}}_{\rm{null}}{\left(}{\mathbf{k}}_{b , b}{\right)}.\end{aligned}\ ] ] this singular case is highly unlikely as the probability of ( [ ac0 ] ) in continuous models is zero .
however , it should be checked for completeness . in practice ,
dimensionality reduction involves the projection of the data into the subspace of a few of the first principal components .
we now show that the algorithm in [ sec_eigval ] can be extended to provide higher order components .
the @xmath82th principal component is defined as the linear transformation which is orthogonal to the preceding components and preserves maximal variance .
similarly to the first component it is given by @xmath83 where @xmath84 is the @xmath82th principal eigenvector of @xmath46 . in the concentration domain , @xmath84 is the eigenvector associated with @xmath85 , the @xmath82th smallest eigenvalue of @xmath33 . in order to distribute the computation of @xmath85 ,
we adjust ( [ eigasopt ] ) using the following lemma : let @xmath33 be a symmetric matrix with eigenvalues @xmath86 and eigenvectors @xmath87 . then , @xmath88 the optimal @xmath89 are any values which satisfy @xmath90 for @xmath91 . the proof is based on the recursive variational characterization of of the @xmath82th smallest eigenvalue : @xmath92 where @xmath93 are the preceding eigenvectors , and the optimal solution @xmath94 is the eigenvector associated with @xmath85 .
a dual representation can be obtained using lagrange duality .
we rewrite the orthogonality restrictions as quadratic constraints @xmath95 and eliminate them using lagrange multipliers : @xmath96}{\mathbf{u}}\end{aligned}\ ] ] where the inequality is due to the weak duality @xcite .
the inner minimization is unbounded unless @xmath97 therefore , @xmath98 lagrange duality does not guaranty an equality in ( [ duallambdak ] ) since ( [ dualuk ] ) is not convex .
however , it is easy to see that the inequality is tight and can be attained by choosing @xmath99 .
finally , ( [ suptv ] ) is obtained by replacing the maximum with a supremum and relaxing the constraint .
lemma 2 allows us to find @xmath85 in a distributed manner .
we replace @xmath33 with @xmath100 where @xmath101 is a @xmath102 matrix with the preceding eigenvectors as its columns and @xmath103 is a @xmath104 diagonal matrix with sufficiently high constants on its diagonal , and search for its principal component . the matrix @xmath105 does not necessarily satisfy the sparse block structure of @xmath33 so we can not use the solution in proposition 1 directly .
fortunately , it can be easily adjusted since the modification to @xmath33 is of low rank .
let @xmath33 be a symmetric matrix with @xmath59 .
then , the constraint @xmath106}+{\mathbf{u}}{\mathbf{d}}{\mathbf{u}}^t{\right)}\end{aligned}\ ] ] is equivalent to the following pair of constraints @xmath107}_{b,:}{\mathbf{d}}{\left[}{\mathbf{u}}{\right]}_{b,:}^t{\right)}\\ \label{prop1u } t&<&{\rm{eig}}_{\min}{\left(}{\mathbf{k}}_{c_1,c_1}+{\left[}\overline{\mathbf{u}}{\right]}_{c_1,:}\overline{\mathbf{d}}{\left[}\overline{\mathbf{u}}{\right]}_{c_1,:}^t{\right)}\end{aligned}\ ] ] where @xmath108}_{c_1,:}&=&{\left[}\begin{array}{cc } \begin{array}{c } { \mathbf{0}}\\ { \mathbf{i}}\end{array } & { \left[}{\mathbf{u}}{\right]}_{c_1 } \\ \end{array } { \right]}\\ \overline{{\mathbf{d}}}&=&{\left[}\begin{array}{cc } { \mathbf{0 } } & { \mathbf{0}}\\ { \mathbf{0 } } & { \mathbf{d}}\end{array } { \right]}-{\mathbf{m}}_{\mathbf{u}}{\left(}t{\right)}\end{aligned}\ ] ] and the _ message matrix _ @xmath109 is defined as @xmath110}_{b,:}^t \end{array } { \right]}{\left(}{\mathbf{k}}_{b , b}+{\left[}{\mathbf{u}}{\right]}_{b,:}{\mathbf{d}}{\left[}{\mathbf{u}}{\right]}_{b,:}^t - t{\mathbf{i}}{\right)}^{-1}{\left[}\begin{array}{cc } { \mathbf{k}}_{b , c } & { \left[}{\mathbf{u}}{\right]}_{b,:}{\mathbf{d}}\end{array } { \right]}. \end{aligned}\ ] ] the proof is similar to that of proposition 1 and therefore omitted .
thus , the solution to the @xmath82th largest eigenvalue is similar to the method in section [ sec_eigval ] .
the only difference is that the messages are slightly larger .
each message is a matrix of size @xmath111 . in practice ,
dimensionality reduction involves only a few principal components and this method is efficient when @xmath112 is considerably less than @xmath17 ( the size of the messages in a fully centralized protocol ) .
the higher order components can therefore be found in a distributed manner as detailed in section [ sec_eigvec ] above .
we now proceed to the general problem of dpca in decomposable graphs . in the previous section
, we showed that dpca can be computed in a distributed manner if it is a priori known that @xmath12 and @xmath35 are conditionally independent given @xmath36 .
graphical models are intuitive characterizations of such conditional independence structures . in particular , decomposable models are graphs that can be recursively subdivided into the two cliques graph in fig .
1 . therefore , this section consists of numerous technical definitions taken from @xcite followed by a recursive application of the previous results .
an undirected graph @xmath113 is a set of nodes connected by undirected edges .
a random vector @xmath21 satisfies the markov property with respect to @xmath113 , if for any pair of non - adjacent nodes the corresponding pair of random variables are conditionally independent on the rest of the elements in @xmath21 . in the gaussian distribution ,
this definition results in sparsity in the concentration domain . if @xmath33 is the concentration matrix of a jointly gaussian multivariate @xmath21 that satisfies @xmath113 , then @xmath114}_{i , j}=0 $ ] for any pair @xmath115 of non - adjacent nodes .
decomposable graphs are a specific type of graph which possess an appealing structure .
a graph is decomposable if it can be recursively be subdivided into disjoint sets of nodes @xmath2 , @xmath14 and @xmath15 , where @xmath15 separates @xmath2 and @xmath14 , and @xmath15 is complete , i.e. , there are no edges between @xmath2 and @xmath14 and all the nodes within @xmath15 are connected by an edge .
clearly , the simplest non - trivial decomposable graph is the two cliques graph in fig .
[ fig_twoclique ] .
a clique is a maximal subset of nodes which is fully connected .
it is convenient to represent a decomposable graph using a sequence of cliques @xmath116 which satisfy a _
perfect elimination order_. an important property of this order is that @xmath117 separates @xmath118 from @xmath119 where @xmath120 note that this perfect elimination order induces an inherent asymmetry between the cliques which will be used in our recursive solution below .
the two cliques graph in fig .
[ fig_twoclique ] is a simple example of a decomposable graph with @xmath18 , @xmath19 , @xmath121 , @xmath122 , @xmath123 and @xmath124 .
accordingly , @xmath121 separates @xmath125 from @xmath126 .
similarly to the previous section , global ml estimation of the concentration matrix in decomposable gaussian graphical model has a simple closed form .
it can be computed in a distributed manner : @xmath127}^0-\sum_{k=2}^k{\left[}\tilde{\mathbf{k}}^{s_k , s_k}{\right]}^0\end{aligned}\ ] ] where the local estimates are defined as : @xmath128 and @xmath129}_{c_k}{\left[}{\mathbf{x}}_i{\right]}_{c_k}^t,\quad k=1,\cdots , k\\ \label{decomp_ml4 } \tilde{\mathbf{s}}^{s_k , s_k}=\frac{1}{n}\sum_{i=1}^n{\left[}{\mathbf{x}}_i{\right]}_{s_k}{\left[}{\mathbf{x}}_i{\right]}_{s_k}^t,\quad k=2,\cdots , k.\end{aligned}\ ] ] the zero fill - in operator @xmath130}^0 $ ] in ( [ decomp_ml ] ) outputs a matrix of the same dimension as @xmath33 where the argument occupies the appropriate sub - block and the rest of the matrix has zero valued elements ( see ( [ kabc ] ) for a two clique example , and @xcite for the exact definition of this operator ) .
dpca can be recursively implemented by using the previous two clique solution .
indeed , proposition 1 shows that the eigenvalue inequality @xmath131 is equivalent to two adjusted local eigenvalue inequalities @xmath132 where @xmath133}^0.\end{aligned}\ ] ] where @xmath134 is a message as in ( [ message ] ) and @xmath130}^0 $ ] is the zero fill - in operator .
next , we can apply schur s lemma again and replace ( [ eigminhk1 ] ) with two additional inequalities : @xmath135 where @xmath136 and @xmath137 are similarly defined .
we continue in an iterative fashion until we obtain @xmath138 decoupled eigenvalue inequalities .
thus , the feasibility of a given @xmath58 can be checked in a distributed manner with minimal message passing between the cliques , and any line - search can efficiently solve dpca .
specifically , in algorithm 1 displayed below we provide a pseudo code for dpca that solves for @xmath58 using the bisection method .
given initial bounds @xmath139 algorithm 1 is guaranteed to find the minimal eigenvalue up to any required tolerance @xmath140 within @xmath141 iterations .
each iteration consists of up to @xmath142 messages through the matrices @xmath134 whose dimensions are equal to the cardinalities of @xmath143 for @xmath144 .
a simple choice for the bounds is @xmath145 since @xmath33 is positive definite , and @xmath146 as proved in the appendix .
given a principal eigenvalue @xmath55 , its corresponding eigenvector can be computed by solving @xmath75 where @xmath72 as detailed in section [ sec_eigvec ] . beginning with @xmath147 we partition @xmath148 into @xmath149 and @xmath150 and test the singularity of @xmath151 .
if it is singular , then @xmath55 is associated with @xmath149 .
otherwise , we send the message @xmath152 to @xmath153 and repartition it .
we continue until we find the associated remainder @xmath149 or reach the first clique .
then , we compute the corresponding local null vector and begin propagating it to the higher remainders as expressed in ( [ ub ] ) .
a pseudo code of this method is provided in algorithm 2 below .
@xmath154 @xmath155 @xmath147 @xmath156 algorithm 1 can be easily extended to compute higher order eigenvalues through application of proposition 2 . for this purpose ,
note that the inequality in ( [ prop1u ] ) has the same structure as ( [ eigasopt2 ] ) and therefore can be recursively partitioned again .
the only difference is that the rank of the modification is increased at each clique and requires larger message matrices .
thus , the algorithm is efficient as long as the size of the separators ( @xmath157 ) , the number of cliques ( @xmath138 ) and the number of required eigenvalues ( @xmath82 ) are all relatively small in comparison to @xmath17 . given any eigenvalue ( first or high order ) , algorithm 2 finds the associated eigenvector in a distributed and efficient manner .
we now illustrate the performance of dpca using a synthetic numerical example .
specifically , we use dpca to track the first principle component in a slowly time varying setting .
we define a simple graphical model with @xmath158 nodes representing three fully connected networks with only 5 coupling nodes , i.e. , @xmath159 , @xmath160 , and @xmath161 .
we generate @xmath162 length @xmath163 vectors @xmath25 of zero mean , unit variance and independent gaussian random variables . at each time point
, we define @xmath33 through ( [ decomp_ml])-([decomp_ml4 ] ) using a sliding window of @xmath164 realizations with @xmath165 samples overlap .
next , we run dpca using algorithm 1 .
due to slow time variation , we define the lower ( @xmath166 ) and upper ( @xmath167 ) bounds as the value of the previous time point minus and plus 0.1 , respectively .
we define the tolerance as @xmath168 corresponding to 8 iterations .
figure [ timevarying ] shows the exact value of the minimal eigenvalue as a function of time along with its dpca estimates at the @xmath169th , @xmath170th and @xmath171th iterations .
it is easy to see that a few iterations suffice for tracking the maximal eigenvalue at high accuracy .
each iteration involves three evds of approximately @xmath172 matrices and communication through two messages of size @xmath173 . for comparison , a centralized solution would require sending a set of @xmath174 length @xmath158 vectors to a central processing unit which computes an evd of a matrix of size @xmath175 . , scaledwidth=40.0% ]
a promising application for dpca is distributed anomaly detection in computer networks . in this context , pca is used for learning a low dimensional model for normal behavior of the traffic in the network .
the samples are projected into the subspace associated with the first principal components .
anomalies are then easily detected by examining the residual norm .
our hypothesis is that the connectivity map of the network is related to its statistical graphical model .
the intuition is that two distant links in the network are ( approximately ) independent conditioned on the links connecting them and therefore define a graphical model .
we do not rigorously support this claim but rather apply it in a heuristic manner in order to illustrate dpca .
following @xcite , we consider a real world dataset of abilene , the internet2 backbone network .
this network carries traffic from universities in the united states .
figure [ abilene_map ] shows its connectivity map consisting of 11 routers and 41 links ( each edge corresponds to two links and there are additional links from each of the nodes to itself ) . examining the network
it is easy to see that the links on the east and west sides of the map are separated through six coupling links : dnvr - kscy , snva - kscy and losa - hstn .
thus , our first approximated decomposable graph , denoted by @xmath176 , consists of two cliques : an eastern clique and a western clique coupled by these six links .
graph @xmath176 corresponds to a decomposable concentration matrix with a sparsity level of @xmath177 .
our second decomposable graph denoted by @xmath178 is obtained by redividing the eastern clique again into two cliques separated through the four coupling links : ipls - chin and atla - wash .
its corresponding concentration matrix has a sparsity level of @xmath179 .
finally , for comparison we randomly generate an arbitrary graph @xmath180 over the abilene nodes , with an identical structure as @xmath178 ( three cliques of the same cardinalities ) , which is not associated with the topology of the abilene network . in our experiments ,
we learn the @xmath181 covariance matrix from a @xmath182 data matrix representing 1008 samples of the load on each of the 41 abilene links during april 7 - 13 , 2003 . we compute pca and project each of the @xmath183 samples of dimension @xmath184 into the null space of the first four principal components .
the norm of these residual samples is plotted in the top plot of fig .
[ abilene_fig1 ] .
it is easy to see the spikes putatively associated with anomalies .
next , we examine the residuals using dpca with @xmath176 , @xmath178 and @xmath180 .
the norms of the residuals are plotted in the three lower plots of fig .
[ abilene_fig1 ] .
, respectively .
as expected , the topology based plots are quite similar with spikes occurring at the times of these anomalies .
thus , we conclude that the decomposable graphical model for abilene is a good approximation and does not cause substantial loss of information ( at least for the purpose of anomaly detection ) . on the other hand , the residual norm using the random graph is a poor approximation as it does not preserve the anomalies detected by the full non - distributed pca .
these conclusions are supported in fig .
[ abilene_fig2 ] where we show the absolute errors of dpca with respect to pca using the different graphical models .
it is easy to see that @xmath176 results in minimal error , @xmath178 provides a reasonable tradeoff between performance and computational complexity ( through its increased sparsity level ) , while graph @xmath180 is clearly a mismatched graphical model and results in significant increase in error .
, scaledwidth=40.0% ] , scaledwidth=40.0% ] , scaledwidth=40.0% ]
in this paper , we introduced dpca and derived a decentralized method for its computation .
we proposed distributed anomaly detection in communication networks as a motivating application for dpca and investigated possible graphical models for such settings .
future work should examine the statistical properties of dpca . from a statistical perspective
, dpca is an extension of classical pca to incorporate additional prior information .
thus , it would be interesting to analyze the distribution of its components and quantify their significance , both under the true graphical model and under mismatched models .
in addition , dpca is based on the intimate relation between the inverse covariance and the conditional gaussian distribution .
therefore , it is also important to assess its sensitivity to non - gaussian sources .
finally , alternative methods to ml in singular and ill conditioned scenarios should be considered .
the authors would like to thank clayton scott for providing the abilene data , and arnau tibau puig for stimulating discussions and his help with the abilene dataset . in this appendix , we prove that the minimal eigenvalue of a @xmath185 symmetric matrix @xmath33 is less than or equal to the minimal eigenvalue of any of its sub - matrices , say @xmath186 for some set of indices @xmath2 . for simplicity
, we assume that @xmath187 for some integer @xmath188 . the proof is a simple application of the rayleigh quotient characterization of the minimal eigenvalues : @xmath189}{\mathbf{k}}{\left[}\begin{array}{c } { \mathbf{v}}\\ { \mathbf{0}}\end{array } { \right]}}{{\left[}\begin{array}{cc } { \mathbf{v}}^t & { \mathbf{0}}^t \end{array } { \right]}{\left[}\begin{array}{c } { \mathbf{v}}\\ { \mathbf{0}}\end{array } { \right]}}\\ & = & \frac{{\mathbf{v}}^t{\mathbf{k}}_{a , a}{\mathbf{v}}}{{\mathbf{v}}^t{\mathbf{v}}}\\ \label{mineigv } & = & \min_{\mathbf{u}}\frac{{\mathbf{u}}^t{\mathbf{k}}_{a , a}{\mathbf{u}}}{{\mathbf{u}}^t{\mathbf{u}}}\\ & = & { \rm{eig}}_{\min}\{{\mathbf{k}}_{a , a}\}\end{aligned}\ ] ] where @xmath190 is the optimal solution to ( [ mineigv ] ) .
y. qu , g. ostrouchovz , n. samatovaz , and a. geist . principal component analysis for dimensions reduction in massive distributed data sets . in _ ieee international conference on data mining ( icdm )
|
we consider principal component analysis ( pca ) in decomposable gaussian graphical models . we exploit the prior information in these models in order to distribute its computation . for this purpose ,
we reformulate the problem in the sparse inverse covariance ( concentration ) domain and solve the global eigenvalue problem using a sequence of local eigenvalue problems in each of the cliques of the decomposable graph .
we demonstrate the application of our methodology in the context of decentralized anomaly detection in the abilene backbone network .
based on the topology of the network , we propose an approximate statistical graphical model and distribute the computation of pca .
|
event shapes and jet - rates are infrared and collinear ( irc ) safe observables , which describe the energy and momentum flow of the final state . they constitute an ideal compromise between simplicity and sensitivity to properties of qcd radiation .
they provide then a wealth of information , e.g. in measurements of the coupling @xmath0 and its renormalisation group running , in cross checks / measurements of the values of the colour factors of qcd and , most importantly , in studies of the connection between parton - level ( the perturbative ( pt ) description of quarks and gluons ) and hadron - level ( the real ) , for a review see @xcite .
irc - safety ensures that event - shape distributions can be computed within perturbation theory , however in the more exclusive phase space region where perturbative radiation is suppressed ( conventionally associated to almost vanishing values of the observable , @xmath1 ) large logarithmic corrections need to be resummed to all orders .
more specifically , given an event - shape @xmath2 , a function of all secondary final state momenta @xmath3 and of the set of recoiling born momenta @xmath4 , the probability of `` constrained events '' i. e. @xmath5 has a _ divergent _ pt expansion for @xmath6 @xmath7 i. e. there is a soft and collinear divergence [ @xmath8 for each emitted gluon . today
s state - of - the art accuracy accounts for all leading ( ll ) and next - to - leading logarithms ( nll ) as follows : @xmath9 furthermore resummations are matched to fixed order results at nlo .
the main ingredient in resummations is factorisation .
usually two steps are needed to obtain a resummed prediction .
the first task is to exploit angular ordering to reduce @xmath10-parton matrix elements to a qed - like factorised form , i. e. to the product of independent emissions from the hard scattering partons only .
schematically , e.g. for @xmath11 this can be written as @xmath12 with corrections which contribute beyond nll to @xmath13 .
the second step requires some analytical understanding of the observable s behaviour in the presence of soft - collinear emissions in order to factorise its definition via mellin transforms , i. g. for the thrust @xmath14 in @xmath152 jets one has ( @xmath16 ) @xmath17 by combining eq .
( [ eq : mfact ] ) and eq .
( [ eq : vfact ] ) one obtains the resummed answer @xmath18\>,\ ] ] where the effects due to the running of the coupling have been included . despite the simplicity of the ideas underlying resummations ,
some technicalities can not be avoided .
indeed , at nll accuracy , care is needed to treat properly the emission of hard collinear or soft large - angle gluons and the inclusive gluon splittings .
furthermore , it is known that for some observables , when seeking nll accuracy , the multi - parton matrix - element can not be factorised as in eq .
( [ eq : mfact ] ) , notably this is the case for non - global observables @xcite . in a similar way some observables , such as jet rates in jade - like algorithms @xcite , do not exponentiate , so that it is not possible to write the distribution in the form in eq .
( [ eq : sigma ] ) .
also , the observable s factorisation itself can be non trivial especially for multi - jet observables like the @xmath19-jet limit of thrust minor , which involves 5 integral transforms @xcite .
the observation that the origin of logarithmic enhancement in jet - shape distributions is always the same ( i. e. it is due to the radiation of soft - collinear gluons ) and that the approximations which lead to resummed expressions such as eq . ( [ eq : tres ] ) are very similar for different observables , it was natural to investigate the possibility of developing a general and automated approach to resummation , which does not require analytical treatment of observable specific multiple emission effects via mellin / fourier transforms . such an approach is feasible , and has been implemented in the computer code caesar(computer automated expert semi - analytical resummer ) .
@xcite this program differs from usual monte carlos whose task is generally the numerical evaluation of integrals via generation of random events .
caesarinstead is based on a generic master resummation formula , presented and derived in detail in @xcite .
this formula applies to a well - defined class of observables , whose requirements can be summarised as follows .
* given a born event , when just one soft emission @xmath20 is radiated collinear to the born parton @xmath21 , the observable should behave as @xmath22 where , for each hard leg @xmath23 , @xmath24 are some numbers and @xmath25 is a regular function parameterising the azimuthal dependence .
* the observable should be _ continuously global _ , meaning that it should be _ sensitive to emissions everywhere _ and the _ transverse momentum dependence should be uniform ( @xmath26)_. * the observable should be _ recursive infrared and collinear ( r - irc ) safe _ , meaning that _ the addition of emissions which are much softer or more collinear should not drastically change the value of the observable_. this new concept was introduced and illustrated in detail @xcite . given the general master formula , together with it s applicability conditions , caesarworks as an expert system , which in a first step establishes whether the observable is within its scope .
it s next task is to determine the inputs needed for the evaluation of the master formula , basically the coefficients and functions @xmath27 together with a nll correction function @xmath28 which accounts for effects due to multiple emissions @xcite .
finally as a last trivial step , caesarevaluates the master formula ( integrating over different born configurations when necessary ) .
notice that the first two steps are critical : they require high precision arithmetic @xcite to take asymptotic ( soft & collinear ) limits and they follow methods of `` experimental mathematics '' @xcite to validate or falsify hypothesis .
currently implemented processes are 2 & 3 jets in @xmath29 , [ 1 + 1 ] & [ 1 + 2 ] jets in @xmath30 , drell - yan + 1 jet , hadron - hadron dijet events
. results as obtained with caesarhave been tested against all known nll global event shape resummations in @xmath31 , dis and dy.in hadronic dijet events no event - shape resummations had ever been carried out , so there was no standard definitions of event shapes , as in @xmath31 and dis .
in the following we will therefore show how to define observables at hadronic colliders which reconcile the seemingly conflicting theoretical requirement of globalness and the limited experimental reach of detectors . indeed in hadron - hadron collisions
the presence of radiation from the initial state , together with the limited coverage close to the beam region ( usually expressed in terms of a maximum rapidity @xmath32 ) makes this a critical issue .
measurements for this class of observables include ideally all particles in the event , i.e. the maximum rapidity @xmath32 is taken as large as experimentally possible .
one then defines variants of the usual @xmath31 observables , e. g. * the transverse thrust @xmath33 where @xmath34 are the transverse momenta with respect to the beam axis and @xmath35 defines the transverse thrust axis . *
the thrust minor @xmath36 here @xmath37 are the momentum - components out of the event - plane ( i. e. that containing the beam axis and the @xmath35-axis ) .
nll resummations are then valid as long as @xmath38 @xcite .
explicit results for the differential distributions show that most of the data lie in this region @xcite . the drawback of observables presented in sec .
[ sec : dirglob ] is that they require measurements quite far in the forward region . while this is possible at the tevatron ( where @xmath39 ) , only calorimeter information is available in forward regions , whose resolution is quite low . a way to circumvent
this is to define measurements which explicitly select a restricted , central region and to modify the observable s definition so as to make it indirectly sensitive to emissions in the unobserved phase space region .
this sensitivity is typically achieved exploiting recoil effects .
the idea is essentially that momentum conservation ensures that the _ vectorial _ sum of transverse momenta in the central region , exactly balances the vectorial sum of transverse momenta in the unobserved forward regions .
the definition of this type of observables proceeds then as follows .
one first selects a central region @xmath40 ( e. g. taking all particles with @xmath41 ) , one then defines a central , non - global observable , e. g. a central thrust minor @xmath42 and adds to it a recoil term , e. g. @xmath43 the recoil enhanced thrust minor is then given by @xmath44 these observables are often named indirectly global observables , since they are indirectly sensitive of emissions in the unobserved region through recoil effects .
the drawback of these observables is that in the region of extremely small @xmath2 the nll function @xmath28 , parameterising multiple emission effects @xcite , has a divergence .
the presence of this divergence is well understood , and is not specific of indirectly global observables .
a divergence occurs whenever , due to cancellation between contributions from different emissions , an observable can be very small in the presence of radiation .
the origin of the divergence can be understood with simple arguments : the master resummation formula assumes that the mechanism responsible for keeping the value of the observable small is a ll sudakov effect .
however , in the phase space region of very small @xmath2 , it is more likely that @xmath2 is small because of bi - dimensional cancellations in @xmath45 .
the nll function @xmath28 can not compensate a wrong ll behaviour and manifests this with a divergence . despite the presence of the divergence ,
explicit results show that most of the distribution is in a region which is under control of nll resummations @xcite .
what makes these observables challenging experimentally , is the accuracy with which the recoil term can be measured , indeed @xmath45 is subject to big cancellations between almost equal and opposite large transverse momenta .
a variant of directly global observables is to define a class of observables with exponentially suppressed forward terms .
one introduces the mean transverse - energy weighted rapidity @xmath46 and total transverse momentum of a given central region @xmath40 @xmath47 and with particle in the forward region one defines an exponentially suppressed forward term @xmath48 then e. g. the thrust minor with exponential forward suppression is @xmath49 these observables are intended to have some of the better features of both purely global and indirectly global observables .
theoretical predictions are not affected by divergences and from an experimental side these observables are optimal in that there is no need for a fine resolution in rapidity and azimuth in the forward region , and calorimeter information should be enough to determine the forward term @xmath50 .
despite the fact that observables with exponentially suppressed forward terms seem to be more suitable in many respects , we point out that observables in different classes have complementary sensitivities to perturbative and non - perturbative ( np ) radiation , so that the _ simultaneous study _ of a number of observables within each class is a powerful tool to investigate properties of qcd radiation , e. g. for * _ studies of underlying event _ , since the forward sensitivity ( to beam - fragmentation ) can be specifically tuned , suppressing it for purely pt studies , or deliberately enhancing it when studies np effects .
this allows one to test quantitative , as well as qualitative features of existing models , e. g. the sensitivity of the underlying event to the partonic channel ( i.e. @xmath51 , @xmath52 or @xmath53 ) . *
_ studies of hadronization corrections in multi - jet events _ , in particular for tests of power - corrections beyond the `` feynman tube model '' @xcite . * _ studies of the non - trivial quantum evolution of colour _ , indeed the novel perturbative qcd colour evolution structure that arises in events with 4-jet topology @xcite has never been investigated experimentally before . given the much more challenging environment in hadronic collisions , compared to @xmath31 and dis , we believe that only an automated approach can make such a programme feasible .
a number of other possible measurements are given in @xcite , while results for the automated analysis and resummation of a large number of observables are available from http://qcd-caesar.org .
i wish to thank the organisers of ismd2004 for the invitation .
this work was carried out together with andrea banfi and gavin salam .
99 m. dasgupta and g. p. salam , _ j. phys .
* g 30 * ( 2004 ) r143 _ [ hep - ph/0312283 ] .
m. dasgupta and g. p. salam , _ phys .
* b 512 * ( 2001 ) 323 [ hep - ph/0104277 ] .
n. brown and w. j. stirling , _ phys .
* b 252 * ( 1990 ) 657 .
a. banfi , g. p. salam and g. zanderighi , _ j. high energy phys . _ * 08 * ( 2004 ) 062 [ hep - ph/0407287 ] .
r. d. field and r. p. feynman , _ phys .
d 15 * ( 1977 ) 2590 _ ; _ nucl
b 136 * ( 1978 ) 1 _ ; r. p. feynman , ` photon - hadron interactions ' , w. a. benjamin , new york ( 1972 ) . g. p. korchemsky and g. sterman , _ nucl
b 437 * ( 1995 ) 415 _ [ hep - ph/9411211];e .
gardi and j. rathsman , _ nucl . phys .
* b 609 * ( 2001 ) 123 _ [ hep - ph/0103217 ] . j. botts and g. sterman , _ nucl .
* b 325 * ( 1989 ) 62_.
|
we present recently defined jet - observables for hadron - hadron dijet production , which are designed to reconcile the seemingly conflicting theoretical requirement of globalness , which makes it possible to resum them ( automatically ) at nll accuracy and the limited experimental reach of detectors , so that they are measurable at the tevatron and at the lhc .
|
recently , there has been a surge of interest in the properties of pyrochlore compound dy@xmath3ti@xmath3o@xmath4 , which is considered to be a model system of `` spin ice '' materials .
`` spin ice '' materials governed by the same statistical mechanics of so - called `` ice rule '' as the hydrogen atoms in the ground state of ordinary hexagonal ice i@xmath5 have macroscopically degenerate ground states down to almost zero temperature.nature99,prb04,nature01,apa02 , prb03,jpcm01,prl01,jpsj04,prb02 experimentally , the observed value of ( 1/2)rln(3/2 ) through specific heat measurement@xcite is consistent with what is expected by pauling s theory , while the spin entropy only freezes out below about 4 k.@xcite from magnetic susceptibility studies , a strongly frequency dependent cooperative spin freezing is observed at about 16 k , which is associated with a very narrow distribution of spin relaxation times and a sharp drop at about 2 k.@xcite neutron scattering studies performed by fennell et al.@xcite also well demonstrate the spin ice state and the coexistence of long range ferromagnetic and short range antiferromagnetic order in a magnetic field applied along the [ 110 ] axis of dy@xmath3ti@xmath3o@xmath6 . in theoretical aspect ,
melko and coworkers@xcite report numerical results on the low temperature properties of the dipolar spin ice model by the multicanonical monte carlo ( mc ) method and they find a first order transition to a long - range ordered phase .
other researchersjpsj04 also confirm the existence of the transition under a magnetic field along the [ 110 ] axis with mc simulation .
dy@xmath3ti@xmath3o@xmath4 has a typical a@xmath3b@xmath3o@xmath4 structure with the space group ( fd3 m , o@xmath5@xmath7 ) , no.227 .
b cation is sixfold coordinated and locates at the center of the distorted octahedron formed by corner o ions .
the a - site ions , i.e. , dy@xmath8(a magnetic rare - earth ion with effective spin s=1/2 ) resides on a lattice of corner - sharing tetrahedral , as shown in figure 1 of reference 3 . in the dy@xmath8 sublattice topology the spin configuration with two spins pointing directly towards while two spins pointing directly away from the center of the tetrahedral corresponds to the analogous proton disorder in real ice .
then a ferromagnetic and dipolar nearest - neighbor spin - spin interaction leads to a strong geometrical frustration , preventing the system from long range ordering . as a result
, high degeneracy of spin states will occur at low temperature .
owing to the ising anisotropy arising from crystal - field effect in the pyrochlore lattice , the macroscopically degenerate states when the field is along the [ 111 ] direction are different from those for other field directions .
for the [ 111 ] direction , the frustration structure changes from that of a three - dimensional pyrochlore to that of a two - dimensional kagome - like lattice with constraint , leading to different values of the zero - point entropy.@xcite the combination of ferromagnetic coupling and ising anisotropy may be a reason for the strong frustration . but how the spins interact with each other is not well understood in the spin ice materials .
up to now , there has been no experimental report on the infrared optical properties of dy@xmath3ti@xmath3o@xmath4 .
ir spectroscopy can give insight to the dynamical processes related to phonon , charge carrier , and spin .
motivated by this situation , we studied the infrared reflectivity of dy@xmath3ti@xmath3o@xmath4 single crystal on the [ 111 ] plane from room temperature down to 10 k. the temperature dependence of phonon modes is obtained , and the role played by spin - phonon coupling on the phonon modes is being discussed .
dy@xmath3ti@xmath3o@xmath4 single crystal was prepared by the floating zone method , using an infrared furnace equipped with two elliptical mirrors . before the single crystal growth
, the polycrystalline rods was prepared by a standard solid state reaction .
stoichiometric mixture of dy@xmath3o@xmath9 ( cerac , 99.99% ) and tio@xmath3 ( cerac , 99.9% ) was heated in air at 1250@xmath10c for five days with intermediate regrinding to ensure complete reaction .
the typical growth condition was 4.5 mm / h for both feed and growth speeds . to avoid oxygen deficiency , the single crystal was grown in an o@xmath11 atmosphere of 0.3 mpa .
the single crystal obtained was translucent yellow .
powder x - ray diffraction measurements on the crystal confirmed that the product was a single phase with cubic pyrochlore structure and the parameter of the unit cell was a=10.1122 .
the principal axes were determined using laue diffraction pattern .
the crystal was cleaved and one - side polished with the surface parallel to the [ 111 ] plane .
the reflectivity spectra r(@xmath12 ) at different temperatures were measured at near - normal incidence of about 8@xmath13 on a bomen da8 fourier transform infrared spectrometer , in the range from 50 to 6500 @xmath2 . in the far infrared region , a he cooled bolometer and 6 mm mylar beamsplitter were used . in the middle infrared region ,
a liquid nitrogen cooled mercury cadmium telluride detector and kbr beamsplitter were utilized . to obtain the absolute reflectivity ,
an evaporated golden mirror was served as a reference .
spectra were collected with a resolution of 4 @xmath14 .
the samples were mounted in a continuous helium flow cryostat in which the temperature could be varied between 300 and 10 k.
the temperature dependent reflectivity of the dy@xmath3ti@xmath3o@xmath4 single crystal shown in figure 1 is typical that of a nonmetallic system .
the sharp features in the reflectivity spectra are due to the unscreened infrared active optical phonon modes , above the highest observed lattice vibration frequency , the reflectivity is flat and featureless up to the highest measured frequency .
it can be seen that the reflectivity within the reststrahlen bands increases with decreasing temperature . at high temperature
seven modes can easily be identified .
the mode frequencies are obtained from a damped harmonic oscillator fit of reflectivity spectra with a complex dielectric function @xmath15 in the factorized form ( generalized lyddane - sachs - teller relation ) infrared , prb95 : @xmath16 where @xmath17 represents the high frequency dielectric constant ( at frequencies large compared with the lattice vibration frequencies but small compared with the electronic transition frequencies ) , @xmath18 and @xmath19 are the longitudinal and transverse eigenfrequencies of the jth optical phonon mode , @xmath20 and @xmath21 represent their longitudinal and transverse damping constants . using this dielectric function , all the reflectivity spectra in our measurements
were fitted with the well known fresnel formula for reflectivity of a half space medium in vacuum@xcite : @xmath22 based on equation ( 1 ) and ( 2 ) , a fit of r(w ) to the observed reflectivity spectra can be obtained with a proper choice of the model parameters @xmath23 , @xmath19 , @xmath24 , @xmath21 and @xmath25 .
adjustment of the parameters is made by trial and error fitting of formula ( 2 ) to the experimental spectra .
this method yields not only @xmath26 but also the model parameters which characterize the infrared active phonons . from a group
theoretical analysis,@xcite a compound with the pyrochlore structure symmetry should have the vibrational phonon modes at the brillouin zone center @xmath27 point with@xmath28=8 f@xmath29 + 4 f@xmath30 + 2 f@xmath31
+ 4 f@xmath32 + 3 e@xmath33 + e@xmath34 + a@xmath31 + 3 [email protected] these 26 normal modes , only a@xmath31 , e@xmath34 , 4 f@xmath32 are raman active , 7f@xmath29 are infrared active , and one f@xmath29 acoustical .
consequently , the observed phonons in the spectra of dy@xmath3ti@xmath3o@xmath4 at low temperature except that at about 610 @xmath2 were denoted sequentially from low to high frequencies by f@xmath29@xmath35 to f@xmath29@xmath7 .
the number is in good agreement with the group theoretical analysis .
the mode at 610 @xmath2 , denoted by f@xmath29@xmath36 , is weak in intensity and does not change with the temperature .
it is probably originated from two phonon absorption because of its low intensity.@xcite the fitting results of the oscillator parameters for all the phonon modes of dy@xmath3ti@xmath3o@xmath4 are summarized in table [ tab : table1 ] .
the high frequency dielectric constant is adopted as 5.0 .
figure 2 shows the experimental and fitting curves at two representative temperatures of 300 k and 10 k. the fitting results with the lorentz oscillators could explain the global features of the phonon spectra reasonably well , even though some difference between the experimental and the fitted data remains .
a small deviation of the calculated curves from the experimental one is seen on the low frequency edge of f@xmath29@xmath37 mode in the high temperature .
a possible reason is the reflection from the rear surface of the sample due to its transparency at low frequencies as well as the cutoff of the working frequency of the detector .
however , the deviation has few impact on other phonon fitting parameters in other frequency range . the real part of the optical conductivity @xmath38 can be extracted from @xmath39 , where @xmath40=@xmath41 , here @xmath12 is in unit of @xmath2 and @xmath42 in unit of @xmath43 @xmath2 , the corresponding result is shown in figure 3.@xcite the conductivity
is dominated by seven peak structures due to optical phonon absorption , no free carrier contribution can be observed . for normal materials , with decreasing t , the anharmonic thermal motion would decrease , which results in a decrease in the lattice constant .
then the phonons should shift to higher frequencies and their linewidth should become narrower .
the f@xmath29@xmath44 mode has clearly demonstrated this effect .
its center frequency shifts from 439 @xmath2 to 452 @xmath2 as the temperature decreases from 300 k down to 10 k. moreover , the phonon peak is more and more prominent and separated from the nearby mode f@xmath29@xmath45 .
but it is puzzling that other modes do not exhibit the same behavior , the three most prominent phonons , f@xmath29@xmath37 , f@xmath29@xmath46 and f@xmath29@xmath47 , show discernible redshifts .
coupling of phonons to magnetic excitations in solids may result in a change of phonon self - energy , i.e. , the frequency due to spin - phonon interaction .
therefore , spin - phonon coupling@xcite closely correlated with the dy@xmath8 ( its effective spin s=1/2 ) could be a possible origin for the abnormal temperature dependence of the phonon frequency . as we mentioned in the introduction
, dy@xmath8 cations can have appreciable magnetic moments .
consequently , ferromagnetic nearest neighbor dipole - dipole interactions will present in spin ice dy@xmath3ti@xmath3o@xmath6 in which large dipole interactions have been suggested to be responsible for the spin ice behavior.@xcite these effects have been quantitatively confirmed by both experimental and theoretical approaches.@xcite at the nearest neighboring sites , the exchange interactions between the magnetic atoms become the strongest .
subsequently , the crystal potential u can be described as following : @xmath48 where x is the atomic displacement from the equilibrium position in the oscillator model . in the second term , j@xmath49 is the exchange energy constant which is a function of the structural parameters , such as the bond lengths between dy@xmath8cations and the corresponding bond angles mediated by the o ions .
s@xmath50 represents the spin of dy@xmath8cation at the ith site .
@xmath51 denotes a statistical average over the adjacent spins .
the harmonic force constant derived from its second derivative formula reads : @xmath52 note that the second term represents the spin - phonon coupling , which suggests that the phonon frequency should have an additional contribution . as for the spin - phonon coupling coefficient
, @xmath53 can be different for each phonon and can have either a positive or a negative sign .
furthermore , different phonon frequencies will have redshift or blueshift in the optical conductivity spectra .
as the temperature decreases , spin fluctuation becomes weaker and spin - phonon coupling stronger .
it is important to notice that although ferromagnetic correlations are short ranged without long range order in spin ice compounds , from neutron scattering measurements harris et al.@xcite have shown that the range of ferromagnetic order increases at lower temperature . as a result ,
spin - phonon coupling range is wider so that the phonon frequency shows redshift or blueshift , which is in agreement with what is observed in our optical conductivity spectra .
now we focus on the remarkable change of the spectra weight which is proportional to the area under the optical conductivity peak .
it is well known that , by decreasing the temperature , thermal fluctuation reduces , which may result in narrowing in the linewidth and enhancement in the oscillator strength without any changes in the bonding or coordination .
almost all phonon modes exhibit the theoretically predicted behavior in our optical conductivity spectra .
but the spectral weight should not change . the anomalous increase in oscillator strength of the low - frequency modes
must be taken sufficient care of .
optical sum rules provide a powerful tool with which to analyze the behavior of free carriers and bound excitations.handbook the partial conductivity sum rules corresponding primarily to a single class of absorption such as excitation of phonons , conduction , valence , or core electrons have been developed . for oscillator states , the partial conductivity sum rule can be expressed as @xcite @xmath54 where @xmath55 , @xmath56 and @xmath57 ( in unit of @xmath2 ) are the integral lower and upper limits and effective plasma frequency associated with the isolated phonon absorption , respectively . in the above equation , @xmath58 is in unit of @xmath59 .
the integral region of the jth oscillator should be chosen to cover the full spectral weight .
the dramatic increase in the spectra weight has implications for the distribution of charge and the change in local strength of the binding charge
. the effective charge of dy , ti and o ions in the unit cell with k atoms can be determined through the following equation ( 6)(in gauss unit system)@xcite @xmath60 where v@xmath61 is the unit cell volume , j and k index the lattice modes and the atoms with mass m@xmath62 , respectively . for the effective charge
, there is a general expression @xmath63=0 . in dy@xmath3ti@xmath3o@xmath4 , oxygen is the lightest element , therefore , in the right side of equation ( 6 ) , the summation is dominated by o ion item and the terms for dy ion and ti ion may be neglected .
the change in the effective charge is associated mainly with the oxygen ( i.e. , @xmath64@xmath65@xmath66 ) . combining equation ( 5 ) and equation ( 6 )
, we obtain the value for @xmath66 as shown in figure 4 .
the absolute value of @xmath66 is of less significance , what is important is the temperature dependence of the deduced value for @xmath66 , which increases with decreasing temperature .
when the incident light couples to the induced dipole moments created by the atomic displacements associated with a normal mode , if the born effective charge per oxygen atom @xmath67 is increasing , then the size of the induced dipole moment and the optical absorption will also increase , which well explain the above temperature dependence of the optical conductivity . on the other hand ,
the increase in @xmath66 means a change in the bond length and bond angle between o ion and cations so that the bond lengths between dy@xmath8 cations and the corresponding bond angles mediated by the o ions will change .
subsequently , j@xmath49 and @xmath51 have different changes associated with phonon central frequency shift as we observed above . in summary , the change in the effective charge arising from electrical charge localization strengthening is the fundamental reason why the spectra weight increases and phonon frequency shifts . from our experimental results , within the framework of the ferromagnetic spin ice model , we consider that the increase in @xmath67 is closely related with nearest - neighbor ferromagnetic interaction strengthening and/or widening in the correlation range . in this simple physical picture , the increasing fm exchange between dy@xmath8 cations will lead to a decrease of the distance between mediated o ion and dy@xmath8 , which enhance the increase in @xmath66 , in agreement with our experimental observation .
lst relationship also provides an effective examination on static optical permittivities
@xmath68 in the approximation of zero phonon frequency . when @xmath69=0 , the equation evolutes into the following form : @xmath70 from equation ( 7 ) @xmath68 of dy@xmath3ti@xmath3o@xmath4 single crystal at different temperatures are also obtained , as shown in the inset of figure 4 .
it exhibits the same temperature dependence as @xmath66 , which suggests that the temperature dependence of @xmath68 can be originated from that of @xmath66 and essentially , charge localization arising from fm exchange plays an important role in the temperature variation of both @xmath68 and @xmath66 .
in conclusion , fir response of dy@xmath3ti@xmath3o@xmath4 single crystals has been studied under different temperatures .
seven infrared active phonons have been observed , which agrees well with a group analysis .
all the spectra are fitted with the oscillator model and an excellent agreement between experimental and calculated spectra is obtained . of all phonons ,
one shows discernible blueshift and three obvious redshifts with decreasing temperature .
the shifts are attributed to the spin - phonon coupling in a geometrically frustrated configuration in the spin ice material .
the oscillation strength of the low frequency modes increase dramatically at low temperature , indicating that the born effective charges are increasing in the unit cell .
we propose that the similar temperature dependence of the static optical permittivity with born effective charge may originate from the intrinsic charge localization resulted from the nearest - neighbor fm interaction .
this work is supported by national science foundation of china ( grants no .
10474128 and no .
10474074 ) .
99 ramirez a p , hayashi a , cava r j , siddharthan r and shastry b s 1999 _ nature ( london ) _ * 399 * 333 figure 2 representative experimental and fitted reflectivity spectra of dy@xmath72ti@xmath0o@xmath1 at ( a ) 300 k and ( b ) 10 k. solid lines represent experimental data and dot lines are the fitting results .
figure 4 the temperature dependence of born effective charge per oxygen atom in dy@xmath0ti@xmath0o@xmath1 .
the dots represent the deduced values at various temperatures between 300 and 10 k. inset : the temperature dependence of the static optical permittivities .
lccccccccc @xmath73&&&&&@xmath74&@xmath75 & + t=300 k & @xmath76&@xmath77 & 137 & 226 & 261 & 374 & 440 & 545 & 608 + @xmath78&@xmath79 & @xmath77 & 23 & 77 & 44 & 62 & 26 & 36 & 18 + @xmath78&@xmath80 & @xmath77 & 143 & 253 & 320 & 437 & 537 & 608 & 745 + @xmath78&@xmath81 & @xmath77 & 12 & 31 & 33 & 26 & 38 & 19 & 36 + t=250 k&@xmath76 & @xmath77 & 133 & 224 & 261 & 372 & 442 & 544 & 611 + @xmath78&@xmath79 & @xmath77 & 18 & 65 & 40 & 60 & 22 & 31 & 18 + @xmath78&@xmath80 & @xmath77 & 142 & 253 & 320 & 440 & 537 & 611 & 742 + @xmath78&@xmath81 & @xmath77 & 13 & 29 & 31 & 22 & 33 & 19 & 26 + t=200 k & @xmath76&93 & 135 & 204 & 259 & 370 & 444.5 & 543 & 608 + @xmath78&@xmath79 & 15 & 23 & 62 & 26 & 42 & 16 & 23 & 17 + @xmath78&@xmath80 & 93.5 & 144 & 254 & 322 & 443 & 538 & 608 & 746 + @xmath78&@xmath81 & 12 & 12 & 22 & 29 & 16 & 24 & 18 & 20 + t=150 k & @xmath76&91.5 & 130 & 199 & 259 & 370 & 447 & 544 & 608 + @xmath78&@xmath79 & 18 & 18 & 45 & 45 & 35 & 16 & 23 & 16.5 + @xmath78&@xmath80 & 92 & 142 & 254 & 321 & 445 & 538 & 608 & 749 + @xmath78&@xmath81 & 14 & 13 & 20 & 29 & 16 & 25 & 17 & 28 + t=100 k & @xmath76&88 & 127 & 198 & 258 & 370 & 450 & 546 & 607 + @xmath78&@xmath79 & 18 & 12 & 36 & 20 & 33 & 19 & 29 & 20 + @xmath78&@xmath80 & 90 & 140 & 252 & 320 & 447.5 & 537 & 607 & 748 + @xmath78&@xmath81 & 23 & 12 & 20 & 30 & 20 & 35 & 21 & 45 + t=50 k & @xmath76&86 & 127 & 197 & 259 & 370 & 453 & 545 & 612 + @xmath78&@xmath79 & 15 & 10 & 29 & 18 & 31 & 19 & 30 & 20 + @xmath78&@xmath80 & 89 & 141 & 252 & 320 & 450 & 536 & 612 & 748 + @xmath78&@xmath81 & 29 & 15 & 18 & 28 & 20 & 34 & 21 & 45 + t=10 k & @xmath76&85 & 126 & 197 & 260 & 370 & 454 & 545 & 612 + @xmath78&@xmath79 & 15 & 7 & 26 & 17 & 27 & 18 & 31 & 22 + @xmath78&@xmath80 & 90 & 139 & 253 & 317 & 451 & 535 & 612 & 748 + @xmath78&@xmath81 & 27 & 17 & 18 & 26 & 20 & 34 & 23 & 43 +
|
near normal incident far - infrared reflectivity spectra of [ 111 ] dysprosium titanate ( dy@xmath0ti@xmath0o@xmath1 ) single crystal have been measured at different temperatures .
seven phonon modes ( eight at low temperature ) are identified at frequency below 1000 @xmath2 .
optical conductivity spectra are obtained by fitting all the reflectivity spectra with the factorized form of the dielectric function .
both the born effective charges and the static optical primitivity are found to increase with decreasing temperature .
moreover , phonon linewidth narrowering and phonon modes shift with decreasing temperature are also observed , which may result from enhanced charge localization .
the redshift of several low frequency modes is attributed to the spin - phonon coupling .
all observed optical properties can be explained within the framework of nearest neighbor ferromagnetic(fm ) spin ice model .
|
the detailed comparison of quasar absorption spectra with the predictions of cosmological hydrodynamic simulations has established that the ly@xmath8 absorption ` forest ' is caused by a smoothly - fluctuating neutral component of the intergalactic medium ( igm ) ( e.g. , cen et al .
1994 ; zhang , anninos , & norman 1995 ; hernquist et al .
the simulations also reveal a strong correlation between absorber hi column density and gas overdensity @xmath9 , so that the study of @xmath10 absorbers gives information about the igm for @xmath11 at @xmath5 ( e.g. zhang et al . 1998 ; dav et al .
furthermore , studies of civ , siiv and ovi have revealed that the low - density igm is not as was long expected chemically pristine , but has been enriched to @xmath12 ( with a scatter of @xmath13dex ) down to @xmath14 ( @xmath15 ) ( e.g. , meyer & york 1987 ; songaila & cowie 1996 ) , and perhaps to even lower overdensities ( e.g. , cowie & songaila 1998 ; ellison et al . 2000 ; schaye et al .
2000 ) .
this enrichment may have occurred during a population iii star - formation epoch at @xmath16 ( e.g. carr , bond & arnett 1984 ; ostriker & gnedin 1996 ; haiman & loeb 1997 ; abel et al .
1998 ) , later through the disruption of small ` protogalaxies ' at @xmath17 ( e.g. gnedin & ostriker 1997 ; gnedin 1998 ; madau , ferrara & rees 2001 ) , or even more recently by the ejection of metals from fairly massive galaxies at @xmath18 .
earlier enrichment might allow more uniform igm pollution ( depending on how strongly biased the star formation is ) , but is constrained by the amount of metal that can be generated at early times ; conversely , sufficient metal is easily available at late times , but the larger inter - galaxy spacing may preclude a homogeneous metal distribution ( which observations may or may not indicate ) . in this paper , we address enrichment of the igm by supernova - driven winds from galaxies of ( gas+star ) masses @xmath19 at @xmath20 .
our calculations employ a smoothed - particle hydrodynamics ( sph ) simulation with @xmath21 dark matter particles and @xmath21 sph particles in a @xmath22 box , adopting a cosmology with @xmath23 , @xmath24 and @xmath25 .
the simulation is described in more detail in aguirre et al .
( 2001 ) and in weinberg et al .
these simulations do _ not _ include winds , but we use a parameterized prescription , described in [ sec - method ] , to add metals in a way that models their ejection and distribution by supernova - driven galactic outflows .
this post - processing method allows us to easily vary our assumptions and parameters , and we give results for a number of representative models .
we then compare the resulting intergalactic metal distribution to the quasar absorption line observations and discuss the implications for wind enrichment of the igm .
the method by which we calculate igm enrichment is discussed in detail in aguirre et al .
briefly , our method employs a limited number of outputs from already - completed sph simulations that include star formation .
each unit of forming stellar mass instantaneously generates @xmath26 units of metal mass ( @xmath27 is assumed for all models ) .
we then deposit this metal mass in nearby gas particles as follows : 1 .
particles are grouped into bound objects ( ` galaxies ' ) .
the mass of star formation is tallied for a given galaxy , for which we also compute the mass , star - formation rate ( sfr ) , and an area @xmath28 based on an empirical relation between disk area and mass ( gavazzi , peirini & boselli 1996 ) .
galaxies with sfr / a @xmath29 sfr@xmath30 are assumed to coherently drive steady - state , mass - conserving winds with velocity @xmath31 at the center of star formation and with a mass outflow rate of @xmath32 times the sfr .
here @xmath33 parameterizes the fraction of supernova kinetic energy in the wind , if we assume @xmath34ergs are released per 100@xmath35 of star formation .
based on observations that generally @xmath36 ( heckman et al . 2000 ) , we use @xmath37 .
we allow galaxies with sfr / a @xmath38 sfr@xmath39 to drive winds also , but with the wind energy attenuated by @xmath40 ^ 4 $ ] .
we divide the volume about the center of star formation into @xmath41 regions of equal solid angle about directions @xmath42 , with @xmath43 galactic gas particles per region ( thus @xmath41 increases with galaxy mass ) .
4 . for each angle
@xmath42 we integrate the equations of motion of a ` test shell ' beginning at radius @xmath44 containing 10% of the galaxy s mass .
the shell at radius @xmath45 feels accelerations due to gravity , the ram pressure of the wind ( attenuated by the potential difference between @xmath45 and @xmath44 ) , the ambient thermal pressure , and the sweeping up of a fraction @xmath46 of the ambient medium ( matter could be left behind by the shell either if the ambient medium is inhomogeneous , or as the shell fragments and leaves part of itself with lower velocity .
we take a fiducial value of @xmath47 ) .
all accelerations are computed using the simulation particles contained in the solid angle about @xmath42 ( see aguirre et al .
2001 for details ) .
eventually , the shell stalls ( i.e. has a negligible velocity with respect to the local medium ) at a radius @xmath48 .
we also stop the shell if / when its propagation time exceeds the time between its launch and the redshift at which we quote our results .
5 . in each angular region ,
metals are distributed among gas particles within @xmath48 such that the metal mass in a shell at @xmath45 of thickness @xmath49 is @xmath50 we generally take @xmath51 , but the resulting enrichment is rather insensitive to @xmath8 for @xmath52 ( aguirre et al .
2001 ) . 6 . if @xmath53 for some angle , the corresponding metal mass is not distributed as per the wind prescription , but is deposited ` locally ' over the 32 gas particles nearest to its progenitor star , using the sph smoothing kernel ( see , e.g. , hernquist & katz 1989 ) . the process is repeated for each galaxy , at each time step .
new stars are formed with the metallicity of the gas from which they form .
metals accumulate in gas particles ( which move between time steps ) , and the distribution of metals in stars and in galactic and intergalactic gas can be assessed at each time step .
we first give results for six illustrative models .
the first four have all their parameters fixed ( to the ` fiducial ' values given above ) except for the wind outflow velocity @xmath31 , which we take to be @xmath54 , @xmath55 , @xmath56 , or @xmath57 ( uniformly distributed in the given range for each model ) , independent of galaxy mass ( as suggested by heckman et al .
next , we fix @xmath58 and vary the entrainment fraction @xmath46 from 1% to 100% . the key results of the calculations are shown in fig .
[ fig - windres ] .
the top two panels give a sparse sampling of individual particle metallicities , versus the gas overdensity @xmath59 .
the stellar yield @xmath26 is uncertain by perhaps a factor of two , and all of the curves could be scaled vertically for a higher assumed value .
the metallicity at @xmath60 could also be ( roughly ) scaled down by a factor @xmath61 if a fraction @xmath62 of the metal produced is assumed not to be incorporated into the wind .
panel a shows that in the @xmath63 model , the enrichment is fairly uniform .
a significant fraction of particles ( shown in a bar at the bottom with an artificial 1 dex of scatter added ) have zero metallicity , but the rest have a metallicity spread at a given overdensity of @xmath641 - 2 dex .
in contrast , the @xmath65 model ( panel b ) has _ some _ metal rich particles at low densities , but many more pristine regions that the winds have not reached .
panels c and d show the median and mean gas metallicities vs. overdensity for all six models .
the shaded box indicates the rough range of metallicity found in most absorption systems for which metal lines are detected .
the area with vertical lines indicates the region where the metallicity is agreed upon by independent groups ( e.g. , rauch , haehnelt , & steinmetzet 1997 ; dav et al .
the horizontal lines indicate yet smaller overdensities to which this metallicity may extend ( e.g. , ellison et al . 2000 ; schaye et al .
2000 ) . even if the observational results were complete , neither the mean nor the median particle metallicities we quote are directly comparable to the observational results , but should bracket the quantity that is comparable
. the mean would be too high because ( especially in cases with inhomogeneous distributions ) it tends to be dominated by a small number of high - metallicity regions ; the median gives information about the _ number _ of lines that would be metal rich , but is an underestimate because an absorption line would generally probe more than one simulation gas particle , decreasing the scatter .
in panel e the gas metallicity is multiplied by the fraction of gas at the given overdensity , showing what fraction of all cosmic metal lies at a given density . for the models that can plausibly account for the observed metal lines in the igm ,
@xmath66 of all metal is extragalactic ( eg ) , i.e. is not in cooled gas or stars in galaxies the curves show that the assumed wind velocity at inception is quite important in the resulting igm enrichment .
slow winds tend to be confined , as can be seen from the small fraction ( @xmath67 ) of all metals outside of galaxies in the @xmath68 model ( see panel e ) .
those that escape tend to stall at relatively small distances where the density is still high , and fail to effectively pollute the low - density igm .
increasing @xmath31 both increases the total eg metal density ( raising the eg fraction to 50% for @xmath69 ) and shifts the metals to lower density regions ( increasing the @xmath70 mean enrichment by a factor of several hundred ) .
changes in the entrainment fraction also have a fairly large effect .
if ( instead of the fiducial 10% ) only 1% of the ambient gas were entrained , 20% of all metals would be outside of galaxies in our fiducial 300km / s model ; while only 4% would be extragalactic if all ambient gas were entrained .
changing our other model parameters within a range of reasonable values has a smaller but still substantial effect ; doubling sfr@xmath39 or changing @xmath33 by a factor of two changes the ejection fraction by @xmath71 and changes the mean enrichment at @xmath70 by @xmath72 ( lowering sfr@xmath39 would have little effect since most galaxies at @xmath73 are already driving winds for our assumed value . )
two methodological points concerning our results merit discussion .
first , resolution tests reveal that galaxies with @xmath74 in baryons are well - resolved .
galaxies below this limit exist ( and are used in the analysis ) but will be under - represented both in their number and in their star - formation rates ( weinberg et al .
studies of dwarf galaxies at high @xmath75 have concluded that at @xmath76 , galaxies of baryon mass @xmath77 are suppressed by photo - evaporation ( e.g. , ferrara & tolstoy 1999 ) , leaving a ( redshift - dependent ) ` gap ' in mass where galaxies may exist but are not resolved by our simulations .
second , our method necessarily neglects the effect of the outflows on the galaxies .
galaxy formation studies with ineffective feedback generally predict more low - mass galaxies than are observed , and the simulations used here may over - estimate the cosmic sfr and stellar mass for this reason ( see weinberg et al .
1999 and aguirre et al .
while we can not quantify exactly the effect these two ( opposing ) inaccuracies will have on our results , comparison of the mass functions for different numerical resolutions indicates that we can probably hope to estimate the enrichment by @xmath76 galaxies to within a factor of a few .
because igm enrichment is so sensitive to @xmath31 , we can set a lower limit on that quantity if the observed enrichment is primarily due to winds of the type discussed in this paper .
to achieve the _
mean _ metallicity of @xmath78 observed in @xmath79 absorbers , @xmath80 is necessary , and @xmath81 would be needed to enrich absorbers down to @xmath82 with a similar metallicity ( though this may _
over_-enrich higher column density systems ) .
a higher velocity is probably necessary if the observed enrichment is fairly uniform ; for a _
median _ metallicity of @xmath83 down to @xmath84 , a wind speed of @xmath81 is probably required , though a more detailed analysis is necessary for a robust comparison .
these conclusions depend on the assumed entrainment fraction ( and more weakly on the other parameters ) , but changing these assumptions can not increase the enrichment very much because enrichment is always limited by the available time between shell launch and @xmath85 thus if the observed enrichment turns out to be uniform at column densities @xmath86 , enrichment from smaller and/or higher redshift galaxies would be required .
in addition to providing estimates of intergalactic enrichment , our calculations also shed light on the physics of wind escape .
figure [ fig - wqz3 ] gives information about the outflows propagating between @xmath87 and @xmath88 from individual galaxies in the @xmath89 model .
panel a shows the shell stalling radius for each shell about a sample of 500 simulation galaxies ; each point corresponds to one propagation direction and each vertical ` stripe ' to a galaxy . at @xmath90 about half of all galaxies are driving winds that reach @xmath91kpc . the variation in @xmath92 for a given galaxy is due to anisotropies in the galaxy and igm ; variations in @xmath92 among galaxies of a given mass are due to combined variations in the outflow velocity , the sfr and the properties of the nearby igm . since we have assumed that wind velocity is independent of galaxy mass , the most massive galaxies drive winds ineffectively because of their relatively large potential wells .
the cutoff at @xmath93kpc is due to the finite time between @xmath94 and @xmath88 ( where we stop each shell s integration ) .
this is shown in panel b , which displays the stalling time for the shell segments ( one point for each angle ) .
as shown in panel c , at @xmath95 almost all of the galaxies exceed the assumed critical sfr/(area ) .
but as long as this threshold is satisfied , the areal sfr does not strongly influence how well winds escape .
panel d allows a comparison of our results for wind escape with two other investigations .
the points in the figure indicate the supernova energy injection rate for each galaxy in units of @xmath96ergs , converted from star formation rates by assuming that one supernova releasing @xmath97ergs forms per per @xmath98 of star formation .
the vertically shaded region indicates the range of sfrs measured in bright , wind - driving lyman - break galaxies by pettini et al .
( 2001 ) , converted in the same way .
the solid line is a fit that extrapolates the simulation results to lower - mass galaxies , and the dotted line represents our fiducial critical sfr / area ( i.e. we assume that galaxies above the dotted line are driving winds coherently though the winds may ultimately be retained as shown in panel a ) .
mac low & ferrara ( 1999 ; see also ferrara & tolstoy 2000 ) have studied galactic winds in dwarf galaxies using numerical simulations , for the range of galaxy masses and supernova energy injection rates shown by the shaded box .
they find that only the lowest - mass galaxies ( @xmath99 ) can ` blow - away ' the bulk of their ism , though higher mass galaxies can allow ` blowout ' where hot , metal rich gas can escape .
their results agree roughly with the estimates of silich & tenorio - tagle ( 2001 ) , who find that winds can only escape spherical galaxies ( corresponding to ` blowout ' ) for energy injection rates above the dashed line of panel d ; winds can escape flattened galaxies for rates above the dot - dashed line . according to our calculations , winds can escape the simulation galaxies ( which are not highly flattened ) , even for entrainment fraction unity , from galaxies with mass @xmath100 .
this would seem to conflict with the findings of mac low & ferrara ( 1999 ) , but the disagreement is probably due to the rather small energy injection rates they assume for their largest galaxies @xmath101 times smaller than the rate indicated by the cosmological simulations .
the simulated galaxy sfrs at high @xmath75 are high enough to fulfill the ( extrapolated ) ` blow - away ' criterion of silich & tenorio - tagle for all galaxy masses , as are the sfrs of the lyman - break galaxies ( unless the galaxies have masses @xmath102 . ) in figure [ fig - forces ] we show plots of both igm enrichment and stalling radius vs. mass in the fiducial model with each effect on shell propagation turned off in turn .
first ( panel d ) we remove the time constraint .
this shows that some shell fragments from @xmath95 galaxies would ultimately propagate to @xmath103mpc ( the cutoff comes from the distance @xmath104 at which @xmath105 ) .
panel e shows the results if the wind is turned off @xmath106years after the shell launch .
this is roughly the minimal interval @xmath107 over which a starburst ` event ' could drive the shell , since the energy deposition rate is roughly constant for @xmath108yr after a coeval starburst ( tenorio - tagle et al .
1999 ) , and the wind ( or thermal pressure ) will still drive the shell for the time @xmath109 it takes the wind ( or sound ) to cross the shell radius . if the shell speed is comparable to the wind speed , @xmath110 so @xmath111 .
the effect of truncating the wind is noticeable but slight . in panel
f we have turned off gravity , and essentially all of the shells escape , showing that in this case gravity is the dominant effect in confining winds . of the shells that still do not escape ,
about half are retained due to thermal pressure , and about half due to the sweeping up of matter .
panel g shows the effect of very low mass entrainment : winds escape to slightly larger radii on average , but can actually increase the escape radius in some cases because higher entrainment adds more momentum to the shell .
entrainment actually causes as much deceleration on the shell as gravity ; but since gravity also attenuates the wind ( decreasing the acceleration of the shell ) , it has a larger overall effect .
panel h shows that taking into account infall from the igm is not very important , but tends to slightly decrease the stalling radius . in summary , we find that only the hot wind , the entrainment of ambient material and gravity are important in determining shell propagation in most cases at @xmath88 . more hot gas ( especially in clusters ) and stronger infall ,
however , might make thermal pressure and ig infall more important at low @xmath75 .
the most important factor limiting the enrichment of very low density regions is the time constraint as shown in fig .
[ fig - times ] .
if we assume that the winds can propagate ` instantaneously ' into the igm , they reach quite low - density regions , and the enrichment at @xmath88 is only slightly higher and more uniform than at @xmath95 .
but if the time limit is respected , the enrichment spreads to significantly lower - density regions between @xmath95 and @xmath88 due to the extra gyr of propagation time .
metal enrichment of the @xmath112 igm has also been investigated by gnedin & ostriker ( 1997 ) , gnedin ( 1998 ) and cen & ostriker ( 1999 ) using simulations that include chemical evolution .
gnedin finds that supernova blow - apart of protogalaxies is ineffective , but that dynamical removal of metals can explain the metals in low - density regions though this requires a metal formation rate at @xmath113 that is about 10 - 50 times that observationally inferred at @xmath114 ( e.g. , steidel et al . 1999 ) and implies more efficient dynamical removal of metals than found by aguirre et al .
( 2001 ) .
cen & ostriker ( 1999 ) assume a fixed metal ejection fraction into regions of @xmath115 ( the resolution of their grid - based hydrodynamical treatment ; see cen & ostriker 2000 ) .
their results for mean metallicity are similar to our @xmath116 model , though their metallicity distribution at fixed @xmath59 shows almost no scatter at @xmath88 ( but becomes more inhomogeneous with time ) .
in contrast , our calculations find a highly inhomogeneous metallicity distribution , which becomes somewhat more uniform with time .
this may be due to the rather low spatial resolution of their simulations or may indicate that stars are forming ( and expelling metals into the igm ) in very low density regions in cen & ostriker s simulations , in contrast to ours .
comparison of quasar absorption spectra to simulated spectra shows that @xmath117 dex of scatter in gas metallicity ( among absorbers in which metals are detected ) may be necessary to explain the variation in observed civ / hi ratios ( rauch et al . 1997 ; hellsten et al .
1997 ; dav et al .
1998 ) , so this point merits further investigation , but will probably require the generation of synthetic spectra from our simulations .
an important physical effect that is not addressed by our treatment is that of the outflowing winds on the igm : if metals are transported directly from galaxies to very low density regions by movement of large quantities of gas , then this kinetic energy must be absorbed by the igm . yet
the widths of low - column density ly@xmath8 forest lines are typically @xmath118 , and line ratios indicate photoionization , not collisional ionization ( e.g. songaila & cowie 1996 )
. our method does not allow us to assess the effects of winds on the igm ( see theuns , mo & schaye [ 2000 ] for some discussion ) , but we can make some general comments . if expanding shells of swept - up gas maintained their integrity and propagated to roughly the inter - galaxy spacing , they would completely re - make the structure of the ly@xmath8 forest , except perhaps in deep voids . but this is not the expected scenario : upon leaving a galaxy , the wind and ` shell ' will likely mix into a complex , multi - phase outflow consisting of fragmented shell(s ) , the hot wind , dense clouds of entrained ( but not swept - up ) ambient medium , etc .
small , dense blobs might maintain fairly low temperature through efficient cooling , and could reach large radii before stalling and mixing with the igm ( and by definition when the wind stalls it has small local velocity and will not greatly heat the medium ) .
in addition , the hot wind itself will not be observable in absorption , and will flow more more easily in low - density directions .
hence it may not seriously disrupt the relatively dense filaments that give rise to much of the ly@xmath8 absorption .
these issues merit further study because our results , combined with observations of high-@xmath75 galaxies , indicate that significant outflow from @xmath5 galaxies seems inevitable .
essentially all of the simulated galaxies at @xmath5 display specific time - averaged sfrs exceeding the observed threshold for driving superwinds in local ( starburst ) galaxies and stochastic star formation should result in much higher rates for short periods .
moreover , available observations of @xmath0 galaxies reveal that they _ are _ driving winds , with characteristic velocities exceeding those of our fiducial model : pettini et al .
( 2001 ) find that lyman break galaxies at @xmath119 with sfrs of @xmath120/yr drive winds at @xmath121 ( as measured by absorption lines ) or @xmath122 ( in ly@xmath8 emission ) .
these galaxies correspond to the highest ( @xmath123 ) mass simulated galaxies ( see fig .
[ fig - wqz3 ] and dav et al . 1999 ) .
the outflow signatures in the spectra are quite similar to those in the nearby wind - driving starbursts upon which we base our method , so the observed velocities should roughly correspond to our assumed initial outflow velocities . in this case , as shown in fig .
[ fig - lybr ] , the outflows tend to escape the galaxies even for an entrainment fraction unity , and propagate to large distances
none of the forces we include in our calculations would be capable of containing the outflow .
thus we predict that winds from lyman - break galaxies do escape those galaxies potentials and enrich the igm , population . ] and expect that this would hold for lower mass galaxies as well .
we have simulated the enrichment of the @xmath88 igm by galactic winds from @xmath124 galaxies , assuming that most drive winds with properties similar to those observed in local starbursts and in lyman - break galaxies .
we find that for winds with outflow velocity @xmath125 enrichment is significant , and that if @xmath126 , the winds can ( roughly ) account for the metallicity of the low - column density ( @xmath127 ) ly@xmath8 forest .
higher velocities could enrich yet lower - density components of the igm .
our treatment of wind - driven outflows includes forces due to gravity , thermal pressure , ram pressure of the ambient medium , and the hot wind itself .
we find that at @xmath112 thermal pressure and the the infall velocity of the igm are generally negligible , but that outflows can be confined by gravity or ambient material for sufficiently high mass galaxies and/or low velocity winds .
our results suggest that the winds observed in lyman break galaxies should escape and travel to fairly large distances .
if the lyman break galaxies represent high - mass , non - starburst galaxies , then the lower - mass @xmath88 galaxies should drive outflows to large distances as well , unless outflow velocity depends strongly on galaxy mass .
the effect of such winds on the physical and thermal structure of the igm is essentially unknown .
the good agreement between simulations like these and the observed statistical properties of the ly@xmath8 forest suggest that the effect can not be extremely large .
we find that the main effect preventing the enrichment of very low - density regions of the igm is the time available for metal - rich material to propagate from galaxies ( which are concentrated in filaments ) into the voids .
this leads to a picture in which a given set of galaxies enriches a progressively larger cosmic volume with decreasing redshift , and we predict that if low - density regions are enriched by winds from @xmath124 galaxies at @xmath76 , the median metallicity of low - density regions should be a very strong function of redshift .
the scenario described here , enrichment of the igm by massive ( @xmath124 ) galaxies at modest ( @xmath76 ) redshift , is but one of several interpretations for the origin of metal in the ly@xmath8 forest .
other possibilities , distinguished by epoch and mass scale , include enrichment by population iii stars at @xmath128 , or the disruption of low - mass protogalaxies at @xmath129 .
how are we to decide which , if any , of these three scenarios is most relevant to the real universe ? based on our modeling , we anticipate that these mechanisms will predict differences in the metal distribution of the igm that can be tested observationally .
in addition to the mean enrichment , the dispersion in metallicity at a given overdensity and the trend of metallicity with overdensity can both be measured .
it seems natural to expect that metal deposition at relatively higher redshift will lead to a rather different distribution of metals than that inferred from our analysis .
thus , future modeling , combined with observational samples of increasingly large size , may discriminate between the various scenarios . to the extent that this goal can be realized , the metals discovered in the @xmath112 ly@xmath8
forest may provide a direct probe of the conditions responsible for the production of the first metals in the universe .
we thank t. heckman for useful comments and information on superwinds .
this work was supported by nasa astrophysical theory grants nag5 - 3922 , nag5 - 3820 , and nag5 - 3111 , by nasa long - term space astrophysics grant nag5 - 3525 , and by the nsf under grants asc93 - 18185 , aci96 - 19019 , and ast-9802568 .
jg was supported by nasa grant ngt5 - 50078 for the duration of this work , and aa was supported in part by the national science foundation grant no .
phy-9507695 and by a grant in aid from the w.m .
keck foundation .
the simulations were performed at the san diego supercomputer center .
weinberg , d. h. , dav , r. , gardner , j. p. , hernquist , l. , & katz , n. 1999 , in photometric redshifts and high redshift galaxies , eds .
r. weymann , l. storrie - lombard , m. sawicki , & r. brunner , asp conference series 191 , san francisco , p. 341 ; astro - ph/9908133
|
studies of quasar absorption lines reveal that the low density intergalactic medium at @xmath0 is enriched to @xmath1 solar metallicity .
this enrichment may have occurred in an early generation of population iii stars at redshift @xmath2 , by protogalaxies at @xmath3 , or by larger galaxies at @xmath4 .
this paper addresses the third possibility by calculating the enrichment of the igm at @xmath5 by galaxies of baryonic mass @xmath6 .
we use already completed cosmological simulations to which we add a prescription for chemical evolution and metal ejection by winds , assuming that the winds have properties similar to those observed in local starbursts and lyman - break galaxies .
results are given for a number of representative models , and we also examine the properties of the galaxies responsible for the enrichment as well as the physical effects responsible for wind escape and propagation . we find that winds of velocity @xmath7 are capable of enriching the igm to the mean level observed , though many low - density regions would remain metal free .
calibrated by observations of lyman - break galaxies , our calculations suggest that most galaxies at @xmath5 should drive winds that can escape and propagate to large radii .
the primary effect limiting the enrichment of low - density ig gas in our scenario is then the travel time from high- to low - density regions , implying that the metallicity of low - density gas is a strong function of redshift .
|
most stars are expected to form in young star clusters ( yscs , @xcite ; @xcite ; @xcite ) . like globular clusters ( gcs ) ,
the densest yscs are collisional systems : their two - body relaxation timescale is shorter than their lifetime , and they undergo intense dynamical evolution . on the other hand ,
yscs are considerably different from gcs : the former have generally lower mass ( @xmath3 m@xmath4 ) and smaller size ( half - mass radius @xmath5 pc ) than the latter ( see e.g. @xcite , for a recent review ) .
this explains why the central relaxation time of yscs is @xmath6 myr , orders of magnitude shorter than that of gcs ( e.g. @xcite ) .
yscs populate the disc of late - type galaxies , while gcs are spherically distributed in the host - galaxy halo .
finally , gcs are old ( @xmath7 gyr ) and long - lived systems , whereas yscs are young and short lived : most of them dissolve in the disc of the host galaxy in @xmath8 yr ( e.g. @xcite ) .
thus , the stellar content of dissolved yscs is expected to build up a considerable fraction of the field population of the host galaxy .
this must be taken into account when modelling the evolution of binary stellar systems in the galactic field : a large fraction of these binaries likely formed in yscs , and then evolved through intense dynamical interactions , before being ejected into the field after the disruption of the parent ysc .
this scenario is important for the study of stellar black hole ( bh ) binaries . in (
* hereafter paper i ) , we studied the formation and the dynamical evolution of accreting bh binaries in yscs .
we found that dynamical interactions in yscs have a significant impact on the expected population of x - ray sources powered by bhs . in the current paper
, we study the formation and the dynamical evolution of black hole - black hole ( bh - bh ) binaries in yscs . for the sake of completeness
, we will compare the evolution of bh - bh binaries with that of neutron star - neutron star ( ns - ns ) binaries and with that of binaries composed of a bh and a neutron star ( ns ) in yscs .
bh - bh , ns - ns and ns - bh binaries are among the most promising sources of gravitational waves ( gws ) detectable by ground - based detectors ( e.g. @xcite ; @xcite ) .
understanding the demographics of such double compact object binaries ( dcobs ) is particularly important in light of the forthcoming second - generation ground - based gw detectors , advanced ligo and virgo @xcite .
the dynamics of yscs can influence the formation and evolution of bh - bh binaries in three different ways : ( i ) dynamical friction causes the bhs ( which are more massive than most stars ) to sink to the denser ysc core , where they have a higher probability to interact with other bhs ( e.g. @xcite ) ; ( ii ) three - body encounters ( i.e. close encounters between a binary and a single star ) change the binary orbital properties : if the binary is hard ( i.e. if its binding energy is higher than the average kinetic energy of a star in the cluster ] where @xmath9 is the gravitational constant , @xmath10 and @xmath11 are the mass of the primary member and the mass of the secondary member of the binary , respectively , while @xmath12 and @xmath13 are the average mass and velocity dispersion of a star in the star cluster . ] ) , three - body encounters tend to shrink the binary semi - major axis @xcite ; ( iii ) dynamical exchanges ( i.e. three - body interactions in which one of the members of the binary is replaced by the single star ) enhance the formation of bh - bh binaries .
in fact , the probability for a single star with mass @xmath14 to replace a binary member is higher if @xmath15 or @xmath16 ( where @xmath10 and @xmath11 are the masses of the former binary members , see @xcite and @xcite ) .
as bhs are more massive than most stars , they efficiently acquire companions through dynamical exchanges .
previous studies investigated the formation and evolution of dcobs either in gcs , via monte carlo codes ( e.g. @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) , or in the field , using population synthesis simulations of isolated binaries ( e.g. @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) .
our study provides a new perspective on this subject : we study the formation of bh - bh binaries in yscs , by using direct n - body simulations coupled with up - to - date stellar and binary evolution recipes .
the paper is organized as follows . in section [ sec : methodsandsimulations ] , we briefly describe our simulations . in section [ sec : results ] , we present our results .
section [ sec : discussion ] is devoted to discuss the results and to compare them with previous work .
our conclusions are presented in section [ sec : conclusions ] .
.summary of initial ysc properties [ cols="<,<",options="header " , ] the timescale for coalescence @xcite is defined as @xmath17 where @xmath18 is the speed of light and @xmath9 the gravitational constant .
@xmath19 is the timescale for a binary to merge by gw emission .
it scales as @xmath20 , and it is shorter for high eccentricity .
gw emission affects the coalescence timescale by shrinking the semi - major axis and circularizing the binary orbit .
[ fig : vpeterstimesmap ] shows @xmath19 as a function of semi - major axis , eccentricity and metallicity of the simulated systems .
most of the systems with @xmath21 ( where @xmath22 gyr is the hubble time ) have eccentricity close to zero , as a consequence of circularization by gw emission .
however , we found an outlier ( with eccentricity @xmath23 , see table [ tab : dbhpeterstable ] ) produced by dynamical exchange .
this is interesting not only because its coalescence timescale is short , due to the high value of the eccentricity , but also because it suggests that the use of templates which include eccentric effects in the ligo and virgo searches could be important @xcite .
unfortunately , this binary is destroyed by a new dynamical exchange before it merges . on the other hand
, we expect to find other systems like this with a larger simulation sample , and we can not exclude that some of them can evolve ( without being destroyed by further exchanges ) till they merge
. such systems would be very important for gw detection @xcite .
all bh - bh binaries with @xmath21 are at low metallicity ( @xmath24 z@xmath1 and @xmath25 z@xmath1 ) , while we find none at solar metallicity .
the bottom panels in fig .
[ fig : vpeterstimesmap ] show the coalescence timescale for ns - ns binaries .
the total number of ns - ns binaries is much smaller than that of bh - bh binaries but they are much harder . as a consequence ,
their coalescence timescales are generally shorter .
the minimum coalescence timescale for bh - bh binaries in our simulations is @xmath26 gyr , while that for ns - ns binaries is @xmath27 gyr .
we also found that 11 ns - ns binaries actually merged before 100 myr . in the bottom panel of fig .
[ fig : vpeterstimesmap ] , we also show the coalescence timescale for ns - bh binaries .
no ns - bh mergers are expected in less than a hubble time , because ns - bh binaries are much less numerous than bh - bh binaries and they are not favoured by dynamical encounters .
tables [ tab : dbhpeterstable ] and [ tab : dnspeters&table ] list the shortest coalescence timescales for bh - bh binaries and ns - ns binaries , respectively .
it has been debated ( e.g. @xcite ) whether @xmath28 is a reliable indicator of the merger timescale in star clusters .
in fact , dynamical interactions in star clusters may affect the evolution of a dcob and delay or anticipate the merger with respect to the expected @xmath28 . in our simulations
there is good agreement between the coalescence timescales and the actual mergers , thus , we can conclude that in most cases dynamics does not affect the actual merger timescale of the simulated ns - ns binaries .
since most stars form in yscs , the mass density of yscs in the universe is expected to scale as the star formation rate ( sfr ) density ( @xcite ) .
thus , from the results discussed in section [ sec : coalescencetimescale ] and using a drake - like equation , the merger rate of bh - bh binaries can be estimated as @xmath29\,{}\left(\frac{\rho_{\rm sf}}{1.5\times{}10^{-2}\,{}{\rm m}_\odot{}\,{}{\rm yr}^{-1}\,{}{\rm mpc}^{-3}}\right)\nonumber\\\left(\frac{t_{\rm life}}{10 ^ 8{\rm yr}}\right)\,{}\left(\frac{f_{\rm sf}}{0.8}\right),\end{aligned}\ ] ] where @xmath30 is the cosmological density of sfr at redshift zero ( @xmath31 from @xcite ) , @xmath32 is the average lifetime of a ysc , @xmath33 is the fraction of star formation ( sf ) that occurs in yscs ( we take @xmath34 from @xcite ) , and @xmath35 is the number of bh - bh binary mergers per solar mass per year , as estimated from our simulations ( see table [ tab : dbhpeterstable ] ) . in equation [ eq : mrgrdbhb ] , we assume that @xmath36 does not change significantly with time . this approximation is reasonable for the distance range of advanced ligo and virgo ( see the short discussion at the end of this section ) .
equation [ eq : mrgrdbhb ] has been derived following the same approach as explained in mapelli et al .
( 2010a , see also @xcite ) .
the main differences between equation [ eq : mrgrdbhb ] of this paper and equations 2 and 3 of @xcite are the following : ( i ) in equation [ eq : mrgrdbhb ] we just estimate the merger rate , while @xcite estimate the detection rate for different interferometers ; ( ii ) in equation [ eq : mrgrdbhb ] we derive @xmath35 directly from our simulations , while in @xcite we used the results of a toy model for intermediate - mass bhs .
in particular , we estimate @xmath35 as @xmath37 where @xmath38 is the number of simulated yscs , @xmath39 is the average mass of a single ysc m@xmath4 , equation [ eq : mrgrdbhb2 ] suffers from the approximation that we do not consider a mass spectrum for the simulated yscs . on the other hand , yscs with @xmath40 m@xmath4
are among the most diffuse yscs in the local universe ( @xcite ) . in a forthcoming paper
, we will consider a mass spectrum for the yscs .
] , and @xmath41 is the number of bh - bh binaries that are expected to merge within a time @xmath42 . for example , at @xmath25 z@xmath1 , we find that 3 bh - bh binaries are expected to merge within @xmath43 gyr ( see table [ tab : dbhpeterstable ] ) . at @xmath24 z@xmath1
, we find 4 bh - bh binaries are expected to merge within @xmath44 gyr , while at @xmath45 z@xmath1 we do not find any bh - bh binaries that merge within @xmath46 .
thus , we find that @xmath47 depending on the metallicity .
the resulting values of the merger rate are @xmath48 , @xmath49 , and @xmath50 , if we assume that all yscs in the local universe have metallicity @xmath51 and 0.01 z@xmath4 , respectively .
thus , the merger rate of bh - bh binaries is @xmath52 if we assume that all yscs in the local universe formed at low metallicity ( @xmath53 ) , and is @xmath54 if we assume that all yscs in the local universe formed at high metallicity ( @xmath55 ) , since in our simulations we did not find any bh - bh binary at @xmath55 with coalescence timescale shorter than the hubble time .
even if the statistics is low , this result is important , as we can conclude that bh - bh binaries are enhanced at low metallicity , where more massive bhs can form . as a first - order approximation
, we can assume that the merger rate of bh - bh binaries in the local universe is included in this range of values , i.e. @xmath56 . for a more realistic assumption about the metallicity of yscs in the local universe , see the discussion at the end of this section .
similarly , the merger rate of ns - ns binaries can be estimated as @xmath57\,{}\left(\frac{\rho_{\rm sf}}{1.5\times{}10^{-2}\,{}{\rm m}_\odot{}\,{}{\rm yr}^{-1}\,{}{\rm mpc}^{-3}}\right)\nonumber\\\left(\frac{t_{\rm life}}{10
^ 8{\rm yr}}\right)\,{}\left(\frac{f_{\rm sf}}{0.8}\right),\end{aligned}\ ] ] where @xmath58 is the number of ns - ns binary mergers per solar mass per year and can be derived as @xmath59 where @xmath60 is the number of ns - ns binaries that actually merged during our simulations and @xmath61 myr is the assumed ysc life ( and the duration of the simulation ) . in the case of ns - ns binaries
we use the number of merged binaries ( rather than the number of expected mergers , as in the case of bh - bh binaries ) , because we have sufficient statistics to do so . at @xmath62 and 0.01
z@xmath1 @xmath63 , respectively .
thus , the merger rate of ns - ns binaries is @xmath64 if we assume that all yscs in the local universe formed at high metallicity ( @xmath55 , see table [ tab : dnspeters&table ] ) , and is @xmath65 if we assume that all yscs in the local universe formed low metallicity ( @xmath66 ) .
this is another important results of our simulations , as it implies that ns - ns mergers are suppressed at low metallicity .
as a first - order approximation , we can assume that the merger rate of ns - ns binaries in the local universe is included in this range of values , i.e. @xmath67 . in equation [ eq : mrgrdnsb ] , we assume that @xmath68 does not change significantly with time .
this approximation is reasonable for the distance range of advanced ligo and virgo ( see the short discussion at the end of this section ) .
finally , the merger rate of ns - bh binaries is @xmath69 for all considered metallicities , as we found no simulated systems with coalescence timescale shorter than the hubble time . in our simulations ,
ns - bh systems are much less common than bh - bh binaries , since the latter are favoured by dynamical exchanges with respect to the former .
our estimates of the merger rate show that there is a possible trend with metallicity : the mergers of ns - ns binaries are favoured at high metallicity ( @xmath70 ) , while the mergers of bh - bh binaries are more frequent at low metallicity ( @xmath71 z@xmath4 ) .
we recall that @xmath24 z@xmath4 is the typical metallicity of gcs in the milky way ( e.g. @xcite ) , @xmath25 z@xmath4 is the metallicity of many irregular galaxies and dwarf galaxies in the local universe ( e.g. @xcite ) , while a metallicity close to solar is normally found in the bulges of giant spiral galaxies and elliptical galaxies ( e.g. @xcite ) .
furthermore , a metallicity gradient ( with @xmath72 decreasing at larger distance from the centre ) has been found in most local late - type galaxies ( @xcite ) .
thus , the metallicity of the local universe is quite patchy , with a preference for higher metallicity at the centre of the most massive galaxies and for lower metallicity in the outskirts of massive galaxies as well as in dwarf and irregular galaxies .
furthermore , the sloan digital sky survey shows that the sf in the last gyr has a bimodal distribution : about half of it occurs at solar metallicity , while the remaining half takes place at @xmath73 z@xmath4 ( @xcite ) .
therefore , we expect that about half of the yscs that formed in the last gyr have @xmath74 z@xmath1 , while the remaining half have @xmath73 z@xmath1 .
in contrast , a negligible fraction of yscs formed at @xmath24 z@xmath4 in the last gyr .
if we assume ( as suggested by @xcite ) that half of the yscs that formed in the last gyr have @xmath74 z@xmath1 , while the remaining half have @xmath73 z@xmath1 , the rate of mergers we expect today from our simulated yscs ( using equations [ eq : mrgrdnsb ] and [ eq : mrgrdbhb ] ) is @xmath75 and @xmath76 , for ns - ns and bh - bh binaries , respectively .
the aforementioned values of @xmath68 and @xmath36 have been derived from the typical properties of yscs in the local universe and assuming a metallicity mixture valid for the last gyr ( i.e. up to redshift @xmath77 ) .
are they valid over the entire distance range of advanced ligo and virgo ? according to @xcite ,
the distance range of advanced ligo and virgo will be @xmath78 mpc ( @xmath79 ) and 1 gpc ( @xmath80 ) for ns - ns and bh - bh mergers , respectively .
thus , we can conclude that our estimated merger rates are fairly uniform ( within the uncertainties ) across the range of advanced ligo and virgo , especially in the case of ns - ns mergers .
we recall that the dcobs that form in yscs will be ejected to the field as a consequence of evaporation , natal kicks and three - body encounters , and because of the disruption of the parent yscs by the tidal field of the host galaxy .
thus , the merger rate we estimate in this section represents the expected merger rate for the field .
this is very important , as previous studies estimated the merger rate either for for long - lived gcs ( e.g. @xcite ; @xcite ; @xcite ) or for the field ( e.g. @xcite ; @xcite ; @xcite ) .
in previous work , the effect of dynamics has been included only in the estimate of the merger rate within gcs , while field binaries have been assumed to form and evolve in isolation ( through population synthesis codes ) .
on the other hand , it is well known that most stars form in yscs and evolve dynamically via three - body encounters , before being ejected into the field .
our results show that the estimate of the merger rate in the field should account for dynamical evolution .
[ fig : mgrcomparison ] compares our predictions of the merger rates with some of the most representative estimates available in the literature . from this figure
, it is apparent that our prediction of @xmath68 is fairly consistent with the estimate derived from short gamma - ray bursts ( @xcite ) .
furthermore , our results for @xmath68 and @xmath36 are consistent with the estimates provided in @xcite . in contrast , our results for @xmath81 are significantly lower than predicted by @xcite .
we recall that the value of @xmath68 reported by @xcite is derived from the observed rate of ns - ns binaries in the milky way ( @xcite ) , while the values of @xmath81 and @xmath36 are obtained from population synthesis codes ( @xcite and @xcite , respectively ) and are only indirectly constrained by the sn rate . the main differences between the approach presented in @xcite and ours are the following .
( i ) the estimates presented in @xcite are based on population synthesis simulations of isolated binaries and do not account for the fact that most stars form in yscs ; ( ii ) the mass spectrum of bhs is significantly different ; ( iii ) @xcite assume that most galaxies in the local universe are milky way analogues , while in this paper we adopt the cosmic sfr by @xcite .
the fact that we account for the dynamical evolution of yscs and include more massive bhs than @xcite affects the results significantly , as the formation of bh - bh binaries is enhanced with respect to that of ns - bh systems .
in general , our simulated dcobs can not evolve in isolation but frequently undergo three - body encounters that perturb their orbits , while the results of @xcite are obtained assuming that all binaries evolve in isolation .
recent studies by @xcite , @xcite and @xcite adopt a bh mass spectrum much more similar to ours and investigate the dependence of the merger rate on metallicity , even if they do not include three - body encounters . as a consequence ,
the distribution of bh - bh binary chirp masses in the three aforementioned papers is very similar to our distribution .
the main difference is the absence of massive bhs that come from a merger in the papers by @xcite , @xcite and @xcite , because they do not allow merged binaries to acquire a new companion dynamically . in their standard model ,
@xcite find an estimate of @xmath68 that is fairly consistent with ours , while their prediction for @xmath81 and @xmath36 are about a factor of ten higher . in addiction ,
@xcite , @xcite and @xcite present an alternative model in which common envelope ( ce ) phases on the hertzsprung gap are allowed ( i.e. the binary is not assumed to merge when one of the two members reaches the hertzsprung gap ) .
the merger rates obtained with this assumption are a factor of @xmath82 higher than our results .
this discrepancy is consistent with our expectations , as our simulations adopt the same recipes for the ce phase as in the standard model of @xcite to model the ce phase ( see @xcite for a definition ) , and we assume that all binaries that enter a ce phase when at least one of the two members is in the hertzsprung gap merge . ] .
@xcite study the merger rate of dcobs in gcs and in the field by means of monte carlo simulations and population synthesis models , respectively .
they find that ns - ns binaries and ns - bh binaries should dominate the dcob population in the field , whereas bh - bh binaries are the main merger candidates in gcs .
we confirm their result , in the sense that the formation of bh - bh binaries is enhanced by dynamics in star clusters .
our results agree with those of @xcite also for the importance of dynamical exchanges : @xcite find that 6 per cent ( 94 per cent ) of bh - bh binary merger candidates come from primordial binaries ( dynamical exchanges ) , while we find that 1.7 per cent of our bh - bh binaries come from primordial binaries .
on the other hand , @xcite neglect the fact that many of the merger candidates in the field have been ejected from yscs ( by dynamical ejection , natal kick or ysc disruption ) .
accounting for field dcobs that were ejected from yscs increases the relative importance of bh - bh binaries in the field , especially at low metallicity .
furthermore , @xcite find a merger rate @xmath83 mpc@xmath84 myr @xmath85 in dense star clusters , substantially higher than our result ( @xmath86 mpc@xmath84 myr @xmath85 ) , because they assume that the bhs remain in dynamical equilibrium with the rest of the cluster .
this suppresses the dynamical ejection of bhs .
other recent papers ( @xcite ; @xcite ; @xcite ; @xcite ) focus on dcob merger in dense stellar systems and gcs . in particular , @xcite perform monte carlo simulations of gcs in which they assume that the bh population is concentrated in the core and dynamically decoupled from the rest of the cluster , because of the spitzer instability ( @xcite ) .
@xcite find that most bh - bh binaries are ejected from the parent star cluster and that the resulting merger rate is @xmath87 mpc@xmath84 myr @xmath85 , much lower than in @xcite , because of the assumed spitzer instability .
the merger rate estimated by @xcite is very similar to our result .
@xcite and @xcite perform monte carlo simulations of gcs .
they ( i ) include a treatment of metallicity that is close to ours ( even if their maximum bh mass is generally lower than ours , as they use the same distribution as in @xcite ) , ( ii ) assume neither spitzer instability nor rigid equilibrium between the bhs and the rest of the cluster _ a priori_. @xcite find that the bhs strongly mass segregate and evolve similarly to what assumed by @xcite .
@xcite find an even lower merger rate than the one derived by @xcite and by our paper , but they admit that this may be due to their approximate treatment of three - body encounters . on the other hand ,
the distribution of orbital periods in the simulations by @xcite is similar to ours ( see fig . [
fig : vorbprophist ] ) .
furthermore , both this paper and @xcite find that most bh - bh binaries form dynamically , through exchanges .
finally , @xcite find that bh - bh binaries form earlier and are more stable at low metallicity , because bhs are more massive , in agreement with our results ( see figures [ fig : vcountintime ] and [ fig : lifetimes ] ) . in conclusion ,
our results confirm that most bh - bh binaries in star clusters come from dynamical exchanges , in agreement with the findings of monte carlo simulations of dense star clusters ( @xcite ; @xcite ; @xcite ) . on the other hand ,
our simulated star clusters are a factor of @xmath88 less massive and a factor of @xmath89 smaller than those studied in previous work ( e.g. @xcite ) .
thus , they are expected to be much more numerous in the local universe than those considered by previous work ( since the mass function of yscs scales as @xmath90 , @xcite ) .
furthermore , the dynamical evolution timescale of our simulated yscs is much shorter , as @xmath91 .
thus , most dcobs that form in our simulated yscs will be ejected to the field ( by ysc evaporation , three - body encounters or tidal fields ) , over a timescale much shorter than found in previous work .
therefore , our yscs can be considered the building blocks of the galaxy disc , and the merger rate we have estimated represents the expected merger rate of the field population .
we studied the impact of metallicity and dynamics on the formation and evolution of dcobs .
to this purpose , we have run 600 n - body realizations of yscs chosen to match the properties of the most common yscs in our galaxy
. we simulated yscs , because most stars form in yscs .
thus , we can not study the formation and evolution of dcobs without accounting for the fact that most of them originate in yscs . for our simulations , we used an upgraded version of the public code starlab , which includes recipes for metallicity - dependent stellar evolution and winds , and which allows stars with final mass larger than @xmath92 to directly collapse to a bh .
direct collapse leads to the formation of massive stellar bhs ( @xmath93 ) at low metallicity .
we found that , while the number of nss is about four times larger than the number of bhs , the number of bh - bh binaries is about ten times higher than the number of ns - ns binaries .
the reason is that dynamical interactions enhance the formation of bh - bh binaries with respect to ns - ns binaries .
heavier bhs sink to the centre of the ysc , where they are more likely to interact with other bhs : bhs can acquire companions through three - body exchanges . since the probability of a dynamical exchange is higher when the single star is more massive than one of the members of the binary and since bhs are among the most massive objects in a ysc , exchanges favour the formation of bh - bh binaries .
bh - bh binaries form earlier at low metallicity , because bhs are more massive in metal - poor yscs . furthermore ,
bh - bh binaries formed at low metallicity are more stable : they live longer than bh - bh binaries in metal - rich yscs .
the simulated bh - bh binaries have very large chirp masses ( @xmath94 m@xmath1 ) , because of the direct collapse at low metallicity and because mergers between stars and bhs are allowed .
bh - bh binaries span a wide range in periods ( @xmath2 yr ) .
in contrast , most ns - ns binaries have periods @xmath95 yr . as a consequence , the coalescence timescale is generally longer for bh - bh binaries than for ns - ns binaries .
the minimum coalescence timescale for bh - bh binaries and ns - ns binaries is @xmath26 gyr and @xmath27 gyr , respectively .
only 7 bh - bh binaries are expected to merge within a hubble time .
moreover , no bh - bh binaries merge during our simulations , while 11 ns - ns binaries do . from our simulations , we can estimate the merger rate of dcobs in the local universe .
we find a merger rate @xmath96 , @xmath69 and @xmath97 for bh - bh , ns - bh and ns - ns binaries , respectively .
the merger rate of ns - ns binaries is fairly consistent with the estimates based on both the observed galactic ns - ns binaries ( @xcite ) and the observed rate of short gamma - ray bursts ( @xcite ) .
the merger rate of bh - bh binaries is consistent with recent monte carlo simulations of dense star clusters ( e.g. @xcite ; @xcite ) .
the merger rate of ns - bh binaries is quite low with respect to previous estimates based on population synthesis codes ( e.g. @xcite ) .
this can be explained with the fact that the formation of ns - bh binaries is less favoured by dynamical exchanges than the formation of bh - bh binaries .
our merger rates are still affected by a number of assumptions that will be improved in forthcoming studies .
first , in our study we assume that the lifetime of the simulated yscs is 100 myr , but we do not take into account the presence of a realistic galactic tidal field .
second , we explore only a limited portion of the parameter space . in forthcoming studies
, we will consider yscs with different concentration , half - mass radius , total mass and binary fraction .
our simulated yscs are expected to dissolve in the galactic disc in @xmath98 myr , that is much shorter than the coalescence timescale of all bh - bh binaries and of some ns - ns binaries .
the dcobs that form within the simulated yscs are ejected in the field ( due to three - body interactions or because of the disruption of the parent ysc ) . once in the field
, the dcobs will not undergo more dynamical interactions and will continue their evolution in isolation , until they merge .
thus , the mergers of ( most ) our simulated dcobs are expected to take place in the field .
accounting for the fact that most dcobs form in yscs and evolve through dynamical interactions is a crucial step towards obtaining a realistic description of the demographics of dcobs , in light of the forthcoming advanced ligo and virgo scientific runs .
we thank the anonymous reviewer for the careful reading and the constructive comments which helped us to improve the manuscript .
we also thank alessandra mastrobuono battisti and roberto soria for useful discussions .
we made use of the public software package starlab ( version 4.4.4 ) and of the sapporo library ( @xcite ) to run starlab on graphics processing units ( gpus ) .
we acknowledge all the developers of starlab , and especially its primary authors : piet hut , steve mcmillan , jun makino , and simon portegies zwart .
we thank the authors of sapporo , and in particular e. gaburov , s. harfst and s. portegies zwart .
bmz , mm and mb acknowledge financial support from the italian ministry of education , university and research ( miur ) through grant firb 2012 rbfr12pm1f . bmz was supported by a phd fellowship from padova university through the strategic research project aacse ( algorithms and architectures for computational science and engineering ) .
mm acknowledges financial support from inaf through grant prin-2011 - 1 and from conacyt through grant 169554 .
we acknowledge the cineca award n. hp10b3bjew , hp10cli3bx , hp10cxb7o8 , hp10c894x7 , hp10cgubv0 , hp10cp6xso and hp10c3anjy for the availability of high performance computing resources and support .
+ 99 aasi et al . , 2013 , phys rev d , 87 , 022002 abadie et al , 2010 , cqg , 27 , 173001 abadie et al , 2012a , phys .
082002 abadie et al , 2012b , phys . rev .
d 85 , 102004 abramovici a. , althouse w. e. , drever r. w. p. , gursel y. , kawamura s. , raab f. j. , shoemaker d. , sievers l. , spero r. e. , thorne k. s. , 1992 , science , 256 , 325a accadia , t. et al .
virgo document vir-0128a-12 , 2012 , https://tds.ego-gw.it/ql/?c=8940 acernese , f. et al . virgo technical report vir-0027a-09 , 2009 , https://tds.ego-gw .
it / ql/?c=6589 belczynski k. , kalogera v. , bulik t. , 2002 , apj , 572 , 407 belczynski k. , sadowski s. , rasio f. a. , bulik t. , 2006 , apj , 650 , 303 belczynski k. , taam r. e. , kalogera v. , rasio f. a. , bulik t. , 2007 , apj , 662 , 504 belczynski k. , dominik m. , bulik t. , oshaughnessy r. , fryer c. , holz d.e . , 2010 , apjl 715 : l318-l141 belczynski k. , bulik t. , fryer c. l. , ruiter a. , valsecchi f. , vink j. s. , hurley j. r. , 2010 , apj , 714 , 1217 brown d. a. , zimmerman p. j. , 2010 , phys . rev .
d 81 , 2 , 4007 coward d.m . , howell e.j . , piran t. , stratta g. , branchesi m. , bromberg o. , gendre b. , burman r.r . , guetta d. , 2012 , mnras , 425 , 2668 carpenter j. m. , 2000 , aj , 120 , 3139c clausen d. , sigurdsson s. , chernoff d. f. , 2013 , mnras , 428 , 3618c davis p. j. , kolb u. , knigge c. , 2012 , mnras , 419 , 287 dewi j. d. m. , podsiadlowski ph . , sena a. , 2006 , mnras , 368 , 1742 dominik m. , belczynski k. , fryer c. , holz d. e. , berti e. , bulik t. , mandel i. , oshaughnessy r. , 2012 , apj , 759 , 52 dominik m. , belczynski k. , fryer c. , holz d. e. , berti e. , bulik t. , mandel i. , oshaughnessy r. , 2013 , apj , 779 , 72 downing j. m. b. , benacquista m. j. , giersz m. , spurzem r. , 2010 , mnras , 407 , 1946d downing j. m. , benacquista m. j. , giersz m. , spurzem r. , 2011 , mnras , 416 , 133d duquennoy a. , mayor m. , 1991 , a&a , 248 , 485 fryer c. l. , 1999 , apj , 522 , 413f fryer c. l. , kalogera v. , 2001 , apj , 554 , 548f fryer c. l. , belczynski k. , wiktorowicz g. , dominik m. , kalogera v. , holz d. e. , 2012 , apj , 749 , 91f gaburov e. , harfst s. , portegies zwart s. 2009 , new astronomy , 14 , 630 gieles m. , portegies zwart s. f. , 2011 , mnras , 410 , l6 goddard q. e. , bastian n. , kennicutt r. c. , 2010 , mnras , 405 , 857 harris w.e .
1996 , aj , 112 , 1487 harry g. m. for the ligo scientific collaboration , 2010 , classical and quantum gravity , 27 , 084006 hartman j. w. , 1997 , a&a , 322 , 127 heggie d. c. , 1975 , , 173 , 729 hills j. g. 1989 , aj , 97 , 222 hills j. g. 1992 , aj , 103 , 1955 hopkins a. m. , beacom j. f. 2006 , apj , 651 , 142 hurley j. r. , pols o. r. , tout c. a. , 2000 , mnras , 315 , 543h kalogera v. et al . , 2004 , apj , 614 , l137 kalogera v. , belczynski k. , kim c. , oshaughnessy r. , willems b. 2007 , phr , 442 , 75 king i. r. , 1966 , aj , 71 , 64 kraicheva z. t. , popova e. i. , tutukov a. v. , iungelson l. r. , 1978 , astronomicheskii zhurnal , 55 , 1176 kroupa p. , 2001
, , 322 , 231 kruijssen j. m. d. , pelupessy f. i. , lamers h. j. g. l. m. , portegies zwart s. f. , icke v. , 2011 , mnras , 414 , 1339k lada c. j. , lada e. a. , 2003 , ara&a , 41 , 57l maggiore m. , 2008 , gravitational waves : volume 1 : theory and experiments , oxford university press mapelli m. , colpi m. , zampieri l. , 2009 , mnras , 395l , 71 m mapelli m. , huwyler c. , mayer l. , jetzer ph . , vecchio a. , 2010a , apj , 719 , 987 mapelli m. , ripamonti e. , zampieri l. , colpi m. , bressan a. , 2010b , mnras , 408 , 234 mapelli m. , ripamonti e. , vecchio a. , graham a. w. , gualandris a. , 2012 , a&a , 542a , 102 mapelli m. , zampieri l. , ripamonti e. , bressan a. , 2013 , , 429 , 2298 - 2314 mapelli m. , bressan a. , 2013 , , 430 , 3120 - 3127 oleary r. m. , rasio f. a. , fregeau j. m. , ivanova n. , oshaughnessy r. , 2006 , apj , 637 , 937o oshaughnessy r. , kim c. , kalogera v. , belczynski k. , 2008 , apj , 672 , 479o zel f. , psaltis d. , narayan r. , mcclintock j. e. , 2010 , apj , 725 , 1918o panter b. , jimenez r. , heavens a. f. , charlot s. , 2008 , mnras , 391 , 1117 peters p. c. , 1964 , phys .
, 136 , 1224 pfahl e. , podsiadlowski p. , rappaport s. , 2005 , apj , 628 , 343 pilyugin l. s. , vlchez j. m. , contini t. , 2004 , a&a , 425 , 849 porras a. , christopher m. , allen l. , di francesco j. , megeath s. t. , myers p. c. , 2003 , aj , 126 , 1916p portegies zwart s. f. , mcmillan s. l. w. , hut p. , makino j. , 2001 , mnras , 321 , 199p portegies zwart s. f. , verbunt f. , 1996 , a&a , 309 , 179p portegies zwart s. f. , mcmillan s. l. w. , 2002 , , 576 , 899 portegies zwart s. f. , 2004 , arxiv : astro - ph/0406550 portegies zwart s. f. , mcmillan s. l. w. , gieles m. , 2010 , ara&a , 48 , 431p quinlan g. d. , new astronomy , 1 , 35q sadowski a. , belczynski k. , bulik t. , ivanova n. , 2008 , , 676 , 1162 - 1169 samsing j. , macleod m. , ramirez - ruiz e. , 2014 , , 784 , 71 - 94 siellez k. , bor m. , gendre b. , 2014 , mnras 437 , 649 sigurdsson s. , hernquist l. , 1993 , nature , 364 , 423s silva - villa e. , larsen s. s. 2010 , a&a , 516 , 10 spitzer l. jr . , 1969 , apj , 158 , l139 vink j. s. , de koter a. , lamers h. j. g. l. m. , 2001 , a&a , 369 , 574v vink j. s. , de koter a. , 2005 , a&a , 442 , 587 voss r. , tauris t. m. , 2003 , mnras , 342 , 1169
as we mentioned in section [ sec : dbhpopulation ] , in our paper a binary system is defined as a bound pair , i.e. the most general possible definition . on the other hand ,
it is reasonable to expect that a portion of these binaries are extremely loose systems , which remain bound only for one ( or few ) time - steps ( see the discussion in section [ sec : orbitalproperties ] ) . in this appendix
, we discuss how our results are influenced by our definition of binary systems .
in particular , we will compare the main properties of stable and unstable dcobs .
starlab defines as stable binaries those bound pairs with periastron distance @xmath99 ( see @xcite ) , where @xmath100 is defined as @xmath101 then , unstable binaries are binaries with periastron @xmath102 . in the following ,
we consider stable and unstable binaries separately .
[ fig : appendixdistro ] is the same as fig .
[ fig : meandcobperz_and_distro ] , but it has been derived considering stable and unstable binaries separately ( in the top and bottom panel , respectively ) . the inset of fig .
[ fig : appendixdistro ] shows the average number of bh - bh , ns - ns and ns - bh per ysc as a function of the metallicity .
it is remarkable that bh - bh binaries are at least ten times more numerous than ns - ns and ns - bh binaries , when considering both the stable binary sample and the unstable binary sample .
this shows that dynamics has a strong impact on the population of dcobs , regardless of the distinction between stable and unstable binaries .
it is also worth noting that we have found no unstable ns - ns binaries .
this confirms that only hard ( stable ) ns - ns binaries can survive ( without being disrupted ) the two sn explosions of the two progenitors and the dynamical evolution of the binary .
the main panel of fig .
[ fig : appendixdistro ] shows the distribution of bh - bh binaries per ysc ( integrated over the simulation time ) . here , the difference between stable and unstable binaries is quite marked : a single ysc can host up to @xmath103 unstable binaries , but only up to @xmath104 stable binaries . fig .
[ fig : appendixtime ] compares the average number of bh - bh binaries per ysc as a function of time for stable ( top ) and unstable ( bottom ) binaries .
it is worth noting that unstable binaries peak at @xmath105 , i.e. immediately after the core collapse : it is reasonable to expect that the formation of loose binaries is triggered by the increase of the central density due to the core collapse phase ( see @xcite ) .
in contrast , the number of stable binaries steadily increases with time ( because they tend to survive for a longer time , after their formation ) .
the differences among metallicities that we discussed in section [ sec : dbhpopulation ] still hold , when considering stable and unstable binaries separately .
[ fig : appendixsma ] shows the distribution of semi - major axes of bh - bh , ns - ns and ns - bh binaries at @xmath25 z@xmath4 , distinguishing between stable ( top ) and unstable ( bottom ) binaries . as it is reasonable to expect , most unstable ( stable ) binaries have semi - major axes @xmath106 au ( @xmath107 au ) . however , there are also some unstable binaries with @xmath108 smaller than that of stable binaries .
the reason is that the stability criterion depends not only on the separation of the two objects , but also on their mass ( in this sense , it is a hardness criterion ) and eccentricity . in the bottom panel of fig .
[ fig : appendixsma ] , the most loose unstable binaries have semi - major axes as large as @xmath109 au , that is @xmath110 pc ( similar to the initial ysc tidal radius ) , with periods comparable to the initial central two - body relaxation time ( @xmath111 10 myr , see also fig .
[ fig : vorbprophist ] ) .
these extremely loose bound pairs are very short - lived : it is reasonable to expect that they would completely disappear , if a galactic tidal field would be included in our simulations .
on the other hand , these highly unstable systems are completely negligible from the point of view of gw sources .
[ fig : appendixgw ] confirms that unstable dcobs are completely negligible from the point of view of gw emission : their coalescence timescale is by orders of magnitude longer than the hubble time .
thus , it is sufficient to consider stable binaries alone , when we are interested in possible gw sources .
|
in this paper , we study the formation and dynamical evolution of black hole - black hole ( bh - bh ) binaries in young star clusters ( yscs ) , by means of n - body simulations .
the simulations include metallicity - dependent recipes for stellar evolution and stellar winds , and have been run for three different metallicities ( @xmath0 ) .
following recent theoretical models of wind mass - loss and core - collapse supernovae , we assume that the mass of the stellar remnants depends on the metallicity of the progenitor stars .
we find that bh - bh binaries form efficiently because of dynamical exchanges : in our simulations , we find about 10 times more bh - bh binaries than double neutron star binaries .
the simulated bh - bh binaries form earlier in metal - poor yscs , which host more massive black holes ( bhs ) than in metal - rich yscs .
the simulated bh - bh binaries have very large chirp masses ( up to 80 m@xmath1 ) , because the bh mass is assumed to depend on metallicity , and because bhs can grow in mass due to the merger with stars .
the simulated bh - bh binaries span a wide range of orbital periods ( @xmath2 yr ) , and only a small fraction of them ( 0.3 per cent ) is expected to merge within a hubble time .
we discuss the estimated merger rate from our simulations and the implications for advanced virgo and ligo .
[ firstpage ] black hole physics
methods : numerical gravitational waves galaxies : star clusters : general binaries : general
|
electrons in a t - shaped structure with three terminals opening , i.e. , three terminals extending to infinity , are classically extended . however , numerical study for the t - shaped structure by lin _
et al_. showed the existence of a localized state trapped at the intersection.@xcite after that , openov presented an analytical solution of this localized state in one - dimensional t - shaped quantum wires.@xcite moreover , he also showed the existence of the localized state trapped at the intersection in a four - terminal - opening cross - shaped structure .
the existence of such localized state essentially shows the confinement effect of the geometry in quantum region .
it is noted that the three - terminal - opening structure here is very different from the previously studied t - junction system,@xcite where only two terminals are open and the structure confinement is more like a kind of cavity confinement.@xcite very recently , xu _
et al_. investigated the localized state in the three - terminal - opening t - shaped graphene nanoribbons.@xcite as reported , the existence of the localized state due to the t - shaped confinement provides the discrete channel to interfere with the directly propagating channels since the localized state embeds in the continuum .
this typical interference , known as the fano - type interference,@xcite leads to a characteristic asymmetric line shape in the transmission . furthermore ,
if one connects two infinity terminals in the cross - shaped structure , to shape a ring geometry , due to the topological similarity there should exist at least one localized state trapped at the intersection of ring and the attached wire .
however , this localized state has not yet been studied in the literature .
ring geometries , thanks to its special topological property , have attracted intensive attention.@xcite for example , electrons confined to a closed path pierced by a magnetic flux , manifest a phase coherence phenomenon : the aharonov - bohm ( ab ) effect,@xcite which has been experimentally observed both in metallic@xcite and semiconducting rings.@xcite in addition , in an analogous effect of the ab effect : the aharonov - casher effect,@xcite the phase coherence of electrons traveling through a closed path with spin - orbit interaction,@xcite has been experimentally demonstrated through a single@xcite and through an array of mesoscopic rings.@xcite topics , such as berry phase,@xcite spin - related conductance modulation,@xcite spin filters@xcite and detectors,@xcite spin rotation@xcite and spin switching mechanisms,@xcite have also been exploited intensively in ring geometries . however , up till now , the above mentioned localized state in a ring geometry attached to the wire has not yet drawn any attention .
recent studies@xcite for a ring attached to the wire showed the existence of bound states , which are bounded in the ring but with small amplitude at the intersection and almost no amplitude along the wire .
however , this kind of bound state is not the localized state as we mentioned above .
moreover , these bound states have no coupling with the propagating states due to the symmetry.@xcite accordingly , no asymmetric fano line shape appears for this type of bound states . in this work
, we propose a scheme that uses a ring with a tangent connection to the wire as well as a magnetic flux threading the ring for modulation .
we show that there exist localized states trapped at the intersection with energies lying below the each subband .
the localized state embedded in the lower subband has a coupling with the continuum ones which leads to the fano - type interference .
we find that this model can provide a large tunable energy gap in the transmission due to the interplay of the fano - type antiresonance and structure antiresonance .
we further demonstrate that this energy gap is very robust against anderson disorder , making this proposal highly feasible for a real application .
this paper is organized as follows . in sec .
[ model ] , we set up the model and lay out the formalism . in sec .
[ results_a ] , we show that for the scheme in one - dimension , there exists only one localized state trapped at the intersection due to the geometric confinement
. however , no fano asymmetry appears in the transmission since the energy of the localized state lies below the energy band . in sec .
[ results_b ] , we find that for the quasi - one - dimensional model in practice , there exist localized states trapped at the intersection with energies lying below the each subband .
we systematically investigate these localized states and the resulting fano - type interference in the transmission .
we summarize in sec .
[ summary ] .
represents the magnetic flux threading the ring .
@xmath0 denotes the azimuthal angle on the ring.,width=325 ] a schematic view of the ring structure in our study is shown in fig .
[ figyw1 ] , wherein a quantum ring with width @xmath1 and inner radius @xmath2 , threaded by a magnetic flux @xmath3 , has a tangent connection to a current wire with the same width @xmath1 and length @xmath4 .
the inner and the outer rings both lie tangent to the wire edges . @xmath0 denotes the azimuthal angle on the ring .
the wire is connected to the left / right leads with the same width @xmath1 through perfect ideal ohmic contacts .
we describe the model by the tight - binding hamiltonian with the nearest - neighbor approximation , @xmath5 where @xmath6 is the hamiltonian of the entire ring , @xmath7 are the hamiltonian of the left , right leads and the wire except the part overlapping with the ring , respectively .
@xmath8 stands for the couplings between the wire and the leads and between the wire and the ring , respectively .
these terms are written as @xmath9 the index @xmath10 is the site coordinate in the structure and the leads and @xmath11 denotes that the sum is restricted to the nearest neighbors .
@xmath12 describes the hopping in the structure and the leads and between the structure and the leads . here , @xmath13 and @xmath14 stand for the effective mass and lattice constant , respectively .
@xmath15 represents the on - site energy .
@xmath16 is the flux quantum .
the localized state is studied by diagonalizing the hamiltonian @xmath17 numerically . within the framework of the landauer - bttiker
approach,@xcite the transmission amplitude is given by @xmath18,\ ] ] in which @xmath19 represents the self - energy of the isolated ideal leads and @xmath20 denotes the retarded / advanced green s function for the ring structure.@xcite @xmath21 is the fermi energy in the leads .
in order to show that the existence of the localized state trapped at intersection is indeed due to the geometric confinement effect similar to the cross - shaped structure , we first study an one - dimensional model in sec . [ results_a ] , which shows that there exists only one localized state trapped at the intersection with energy lying below the energy band .
we then lift the energy of this localized state into the energy band by using an on - site gate voltage@xcite and obtain the fano asymmetric line shape in the transmission .
however , for the quasi - one - dimensional model in practice ( sec .
[ results_b ] ) , due to the subband effect , there exist localized states trapped at the intersection with energies lying below the each subband .
the transmission therefore is strongly modulated by the fano - type interference especially at the fano antiresonance due to the coupling between the localized state and the continuum ones .
for the one - dimensional model ( @xmath22 ) , we take the ring size @xmath23 , i.e. , the ring contains 11 lattice points .
the diagonalization of the one - dimensional structure hamiltonian @xmath24 shows that there exists only one localized state trapped at the intersection ( shown in fig .
[ figyw2 ] ) with energy lying below the energy band . moreover , the energy of this localized state @xmath21 is around @xmath25 , which is close to the energy of the localized state in one - dimensional cross - shaped structure with the same on - site and hopping energy parameters.@xcite the transmission therefore only shows the oscillatory behavior due to the structure resonance and antiresonance as shown in fig .
[ figyw3](a ) , where the transmission amplitude is plotted against the fermi energy @xmath21 in the absence of magnetic flux .
moreover , the corresponding resonant energies in transmission , except the two outermost resonances which are close to the energy band edges , coincide with the eigenvalues of the isolated mesoscopic ring with the same size.@xcite the energies of the two outermost resonances are also very close to the eigenvalues of the isolated mesoscopic ring . in fig .
[ figyw3](b ) , we plot the transmission amplitudes of one resonance against the fermi energy @xmath21 with different magnetic fluxes @xmath26 .
one observes that the resonance peak doubles when @xmath27 since the magnetic flux breaks the symmetry of the clockwise and counter - clockwise propagation on the ring , and the two resonance peaks shifted in the opposite directions with increasing the magnetic flux .
particularly , when @xmath28 , the resonance and the antiresonance interchange the position in contrast to @xmath29 by comparing figs .
[ figyw3](a ) and ( c ) .
when @xmath30 , the transimission spectrum returns to the case with @xmath29 due to the periodicity of the magnetic flux .
it is noted that we neglect the effect of the magnetic flux on the left and right leads in the computation .
this is because that the ab phase coherence is due to the phase accumulated in a closed path rather than the phase accumulated on the leads .
our calculations also confirm that there is no difference for the results with / without the magnetic flux on the leads .
_ vs _ fermi energy of the leads for one - dimensional scheme , ( a ) with the magnetic flux @xmath29 , ( b ) with different magnetic fluxes , ( c ) with the magnetic flux @xmath28 , ( d ) with an on - site gate voltage @xmath31 at the intersection and @xmath29 . , width=419 ] following the idea of introducing the localized state into the energy band , we lift the energy of the localized state by using an on - site gate voltage . we apply a positive voltage@xcite @xmath31 in the region where the probability density of the localized state is larger than @xmath32 ( as show in fig .
[ figyw2 ] , including the intersection point , two points on the line and two points on the ring ) .
we find that the energy of the localized state is lifted to @xmath33 .
the transmission amplitudes against the fermi energy are plotted in fig .
[ figyw3](d ) , where one observes that the asymmetric fano line shape appears at @xmath34 . for the quasi - one - dimensional model in practice
, the diagonalization of the structure hamiltonian @xmath24 shows that there exist localized states trapped at the intersection with energies lying below the each subband . the states which we are interested in are the ones whose energies lying inside the lowest subband .
we find that when the inner radius of the ring @xmath2 is one order of magnitude smaller than the width @xmath1 of the wire , there exists only one localized state trapped at the intersection with energy lying inside the first subband .
however , with increasing the inner radius of the ring @xmath2 under the same width @xmath1 , two localized states appear in the first subband and tend to be degenerate .
geometrically , the increase of the inner radius separates the two positions where the ring intersects into the wire , the intersection then gradually evolves into two small equivalent intersections of the t - shaped structure .
the two localized states embedded in the first subband concentrate at these two intersections . since we are interested in the localized states due to the geometric confinement , we therefore concentrate on the case with small inner radius of the ring ( one order of magnitude smaller than the width @xmath1 of the wire ) . in fig .
[ figyw4 ] , we plot the probability density of the localized state for the system with @xmath35 and @xmath36 , and with the eigen - energy @xmath37 lying inside the first subband .
one finds from the figure that for this localized state @xmath38 the electron indeed concentrates at the intersection , and decays exponentially away from the intersection .
moreover , the wave - function amplitude in the wire is small but finite , which implies that this localized state embedded in the first subband has a coupling @xmath39 with the continuum ones @xmath26 and therefore manifests a quasi - localized behavior . for the system with different inner radii @xmath2 under the same width @xmath1
, the coupling @xmath39 mainly comes from the region where the localized state concentrates and the localized state trapped at a larger region has a stronger coupling ( larger wave - function amplitude in the wire ) .
furthermore , we also find the adjustment of the magnetic flux has little influence on the position of the localized state trapped at the intersection in the energy spectrum . ,
@xmath35 and @xmath28 with eigen - energy @xmath37.,width=309 ] _ vs _ fermi energy of the leads with different magnetic fluxes when @xmath36 and @xmath35 .
, width=362 ] _ vs _ fermi energy of the leads with different disorder strengths @xmath40 when @xmath36 , @xmath35 and @xmath29 , ( a ) with both the fano antiresonance and the structure antiresonance .
( b ) with an ordinary structure antiresonance .
, width=362 ] in fig .
[ figyw5 ] , we plot the transmission amplitudes against the fermi energy for the system with @xmath35 and @xmath36 , but with different magnetic fluxes .
the energy window we choose in the figure lies inside the first subband .
one finds that the asymmetric fano line shape appears at the vicinity of @xmath37 .
the position of the fano line shape coincides with the corresponding localized state obtained from the diagonalization .
one observes that the magnetic flux has little effect on the position of the fano line shape but has a marked influence on its shape .
this interesting modulation is caused by the shift of the structure resonance which has been demonstrated in the case of one - dimensional model in the previous subsection .
when the fano dip is close to an antiresonance dip ( @xmath29 ) , a wide energy gap @xmath41 $ ] appears in the transmission ( see the red solid curve ) .
moreover , this gap can be turned off via tuning the magnetic flux . specifically , when the fano dip is close to a resonance peak ( @xmath28 ) , a sharp dip with a sharp peak nearby appears in the transmission and a large transmission amplitude @xmath42 is obtained for the gap window @xmath41 $ ] ( see the blue dotted curve ) .
this feature demonstrates that the energy gap can be turned off or on by tuning the magnetic flux .
therefore , this model can work as a transistor with on and off features by tuning the magnetic flux .
we now show the feasibility of the above proposed model for a real application by analyzing the robustness of the energy gap against the anderson disorder . in our simulation ,
the anderson disorder is introduced by generating random on - site energies at the structure sites : @xmath43 in eq .
( 3 ) . here
@xmath40 is the anderson disorder strength and @xmath44 is a random number with a uniform probability distribution in the range @xmath45 .
the converged transmission amplitudes , which are averaged over 3000 random configurations for @xmath35 and @xmath36 , are plotted in fig .
[ figyw6](a ) against the fermi energy in the vicinity of the energy gap with different anderson disorder strengths @xmath40 .
one finds that the vanishingly small transmission amplitudes in the gap window @xmath41 $ ] become larger as the disorder strength @xmath40 increases .
but even for the very large disorder strength @xmath46 , the corresponding transmission amplitudes for the gap window are still smaller than @xmath47 , which is one order of magnitude smaller than the on " transmission amplitudes @xmath48 as mentioned above . for comparison , we also show the robustness of an ordinary antiresonance ( with the energy window @xmath49 $ ] and @xmath29 ) without a fano antiresonance nearby in fig .
[ figyw6](b ) .
the transmission amplitudes in this energy window rapidly increase with the strength of the disorder .
specifically , they already reach over @xmath50 at @xmath51 .
therefore , the transistors only based on the structure antiresonance are very weak against the disorder . _ vs _ fermi energy of the leads with different inner radii @xmath2 of the ring when @xmath36 and @xmath52 . , width=309 ] in fig .
[ figyw7 ] , we plot the transmission amplitudes against the fermi energy for the system with @xmath36 and @xmath52 but with different inner radii @xmath2 , one observes that the asymmetric fano antiresonance dip which appears in transmission when @xmath53 ( see the red solid curve ) gradually fades away with increasing the inner radius of the ring under the same width @xmath36 .
especially , when @xmath54 under the same width @xmath36 , this dip in transmission is barely visible and two sharp resonance peaks appear ( see the pink chain curve ) . as mentioned above , under the same width @xmath1 , the increase of the inner radius separates the intersection into two small ones and there appear two localized states each trapped at a small intersection . since the strength of the coupling @xmath39 between the localized state @xmath55 and the continuum ones @xmath3 depends on the area of the localized region , these two localized states trapped at the small intersection therefore have weak couplings with the continuum ones , leading to two sharp resonance peaks in the transmission
in summary , we have extended a special kind of localized state trapped at the intersection due to the geometric confinement , first proposed in a three - terminal - opening t - shaped structure,@xcite into a ring geometry with a tangent connection to the wire .
this kind of the localized states has long been overlooked in the ring geometry attached to the wire . in this ring geometry , where we use a magnetic flux to thread the ring for modulation
, we find that when the inner radius of the ring is one order of magnitude smaller than the width of the attached wire , there exists one localized state trapped at the intersection with energy lying inside the lowest subband .
the fano - type interference due to the coupling between this localized state and the continuum ones strongly modulates the transmission , leading to a fano line shape .
however , the increase of the inner radius of the ring weakens this coupling and the asymmetric fano dip fades away . by tuning the magnetic flux for the structure with small inner radius of the ring , we find that a wide energy gap appears in the transmission when the fano antiresonance and the structure antiresonance are close to each other .
we propose that our structure can be used as a transistor since large transmission amplitudes can be obtained in the same energy gap region when the two types of the antiresonance are tuned away from each other by changing the magnetic flux .
we also demonstrate the strong robustness of this energy gap against the anderson disorder ( in contrast to an ordinary structure antiresonance ) .
such features suggest that the proposed structure has great potential to work as transistors for a real application .
this work was supported by the national natural science foundation of china under grant no .
11334014 , the national basic research program of china under grant no .
2012cb922002 , and the strategic priority research program of the chinese academy of sciences under grant no .
. one of the authors ( f.y . ) would like to thank l. wang and t. yu for valuable discussions .
lin , y .- n .
chen , and d .- s .
chuu , phys .
b * 64 * , 193316 ( 2001 ) .
l. a. openov , euro .
. lett . * 55 * , 539 ( 2001 ) .
f. sols , m. macucci , u. ravaioli , and k. hess , appl .
. lett . * 54 * , 350 ( 1989 ) . , edited by s. p. beaumont and c. m. s. torres ( springer , berlin , 1990 ) , p. 107 . x. y. feng , j. h. jiang , and m. q. weng , appl .
lett . * 90 * , 142503 ( 2007 ) .
k. shen and m. w. wu , phys .
b * 77 * , 193305 ( 2008 ) .
h. tong and m. w. wu , phys .
b * 85 * , 205433 ( 2012 ) .
j. g. xu , l. wang , and m. q. weng , j. appl .
phys . * 114 * , 153701 ( 2013 ) .
u. fano , phys . rev . * 124 * , 1866 ( 1961 ) .
a. e. miroshnichenko , s. flach , and y. s. kivshar , rev .
82 * , 2257 ( 2010 ) . y. aharonov and d. bohm , phys .
* 115 * , 485 ( 1959 ) .
y. gefen , y. imry , and m. ya .
azbel , phys .
. lett . * 52 * , 129 ( 1984 ) .
y. aharonov and a. casher , phys .
* 53 * , 319 ( 1984 ) . m. v. berry , proc .
london * 392 * , 45 ( 1984 ) .
r. a. webb , s. washburn , c. p. umbach , and r. b. laibowitz , phys .
* 54 * , 2696 ( 1985 ) .
l. p. lvy , g. dolan , j. dunsmuir , and h. bouchiat , phys .
rev . lett . * 64 * , 2074 ( 1990 ) .
v. chandrasekhar , r. a. webb , m. j. brady , m. b. ketchen , w. j. gallagher , and a. kleinsasser , phys .
lett . * 67 * , 3578 ( 1991 ) .
d. mailly , c. chapelier , and a. benoit , phys .
* 70 * , 2020 ( 1993 ) .
j. nitta , f. e. meijer , and h. takayanagi , appl .
75 * , 695 ( 1999 ) .
a. g. malshukov , v. v. shlyapin , and k. a. chao , phys .
b * 60 * , r2161 ( 1999 ) .
a. fuhrer , s. lscher , t. ihn , t. heinzel , k. ensslin , w. wegscheider , and m. bichler , nature ( london ) * 413 * , 822 ( 2001 ) .
d. frustaglia , m. hentschel , and k. richter , phys .
87 * , 256602 ( 2001 ) .
j. nitta , t. koga , and h. takayanagi , physica e * 12 * , 753 ( 2002 ) .
a. g. malshukov , v. v. shlyapin , and k. a. chao , phys .
b * 66 * , 081311(r ) ( 2002 ) .
m. popp , d. frustaglia , and k. richter , nanotechnology * 14 * , 347 ( 2003 ) .
r. ionicioiu and i. damico , phys .
b * 67 * , 041307(r ) ( 2003 ) .
d. frustaglia and k. richter , phys .
b * 69 * , 235310 ( 2004 ) .
m. hentschel , h. schomerus , d. frustaglia , and k. richter , phys .
b * 69 * , 155326 ( 2004 ) .
d. frustaglia , m. hentschel , and k. richter , phys .
b * 69 * , 155327 ( 2004 ) .
r. capozza , d. giuliano , p. lucignano , and a. tagliacozzo , phys .
lett . * 95 * , 226803 ( 2005 ) .
m. knig , a. tschetschetkin , e. m. hankiewicz , j. sinova , v. hock , v. daumer , m. schfer , c. r. becker , h. buhmann , and l. w. molenkamp , phys .
lett . * 96 * , 076804 ( 2006 ) .
t. bergsten , t. kobayashi , y. sekine , and j. nitta , phys .
97 * , 196803 ( 2006 ) .
f. qu , f. yang , j. chen , j. shen , y. ding , j. lu , y. song , h. yang , g. liu , j. fan , y. li , z. ji , c. yang , and l. lu , phys .
lett . * 107 * , 016802 ( 2011 ) .
f. nagasawa , j. takagi , y. kunihashi , m. kohda , and j. nitta , phys .
lett . * 108 * , 086801 ( 2012 ) .
f. nagasawa , d. frustaglia , h. saarikoski , k. richter , and j. nitta , nat . commun . * 4 * , 2526 ( 2013 ) . f. k. joibari , ya .
m. blanter , and g. e. w. bauer , phys .
b * 88 * , 115410 ( 2013 ) .
a. chaves , g. a. farias , f. m. peeters , and b. szafran , phys .
b * 80 * , 125331 ( 2009 ) . m. r. poniedzialek and b. szafran , j. phys . : condens .
matter * 22 * , 465801 ( 2010 ) .
m. p. nowak , b. szafran , and f. m. peeters , phys .
b * 84 * , 235319 ( 2011 ) .
m. bttiker , phys .
rev . lett . * 57 * , 1761 ( 1986 ) .
s. datta , _ electronic transport in mesoscopic systems _ ( cambridge university press , new york , 1995 ) .
p. g. silvestrov and k. b. efetov , phys .
lett . * 98 * , 016802 ( 2007 ) .
p. recher , j. nilsson , g. burkard , and b. trauzettel , phys .
b * 79 * , 085407 ( 2009 ) .
s. k. maiti , physica e * 36 * , 199204 ( 2007 ) .
l. liao , j. w. bai , r. cheng , y. c. lin , s. jiang , y. huang , and x. f. duan , nano lett .
* 10 * , 1917 ( 2010 ) .
|
we extend a special kind of localized state trapped at the intersection due to the geometric confinement , first proposed in a three - terminal - opening t - shaped structure [ euro .
. lett . * 55 * , 539 ( 2001 ) ] , into a ring geometry with a tangent connection to the wire . in this ring geometry
, there exists one localized state trapped at the intersection with energy lying inside the lowest subband .
we systematically study this localized state and the resulting fano - type interference due to the coupling between this localized state and the continuum ones .
it is found that the increase of inner radius of the ring weakens the coupling to the continuum ones and the asymmetric fano dip fades away .
a wide energy gap in transmission appears due to the interplay of two types of antiresonances : the fano - type antiresonance and the structure antiresonance .
the size of this antiresonance gap can be modulated by adjusting the magnetic flux . moreover , a large transmission amplitude can be obtained in the same gap area .
the strong robustness of the antiresonance gap is demonstrated and shows the feasibility of the proposed geometry for a real application .
|
recently , a new subdiscipline of condensed matter physics , spintronics , is emerging rapidly and generating great interests@xmath2}$ ] .
much attention has been focused , especially , on the spin - dependent transport dominated by the spin - orbit interaction ( soi)@xmath3}$ ] in the low dimensional semiconductor nanostructures , such as wires , rings , spirals , and other structures .
for example , the influence of spin on electrons moving in a mesoscopic ring have been studied in several contexts@xmath4}$ ] since the original proposal of the spin field effect transistor by datta and das@xmath5}$ ] . on the other hand
, many low - dimensional systems are curvilinear and more complicated in geometry , such as the v - shaped quantum wire@xmath6}$ ] , the spiral inductors@xmath7}$ ] , nanotubes@xmath8}$ ] , and so on . due to the potential of providing new physical features and new functionalities for electronic devices@xmath9}$ ] , the effects of the curvilinear geometry on electronic states and electron transports are the subjects of many recent works@xmath10}$ ] . especially , the influence of geometry on the soi is considered by some groups@xmath11}$ ] .
the systems concerned are quasi - one - dimensional circles or two - dimensional ( 2d ) nanostructures .
however , for one dimensional ( 1d ) system , the electron behavior in curved quantum wires with arbitrary shape is still a challenging question .
the purpose of this paper is to study the curvature effects on the electron transport through a 1dcurved nanostructure taking into account the soi .
this paper is organized as follows . in sec .
ii , we investigate a 1d planar curvilinear quantum wire with arbitrary shape . the effective hamiltonian is derived using perturbation theory . in sec .
iii we study the one - electron energy spectrum and eigenstates of curvilinear quantum wire .
then we consider the corresponding transmission of electrons .
a detailed numerical discussion is given .
conclusions and remarks follow in sec .
we consider the adiabatic motion of electrons bounded on a planar quantum wire as shown in fig .
1(a ) . under the effective - mass approximation ,
the schrdinger equation for an electron in the presence of rashba soi@xmath12 } $ ] is given by @xmath13\psi & = e\psi { \bf , } \\ \hat{h}_{so}=\alpha { \bf \hat{\sigma}}\cdot { \bf n}\times { \bf p } , & \nonumber\end{aligned}\ ] ] where @xmath14 is the effective electron mass , @xmath15 is the pauli matrices and @xmath16 is the surface normal .
here @xmath17 represents the potential which confines electrons to the wire , while the rashba term results from the asymmetric confinement along the direction perpendicular to the plane .
the parameter @xmath0 is the rashba parameter which represents the average electric field along the @xmath16 direction . in order to obtain the effective hamiltonian for electrons on a 1d wire , we take the same approach as da costas@xmath18}$ ] .
let @xmath19 parametrically specify the 1d planar curve where @xmath20 is the arc length along the curve .
any point in the surface surrounding the curve can be expressed as follows : @xmath21 where @xmath22 is the normal of @xmath23 and @xmath24 is the curvature of the curve , the curve width @xmath25 is small and is assumed to be less than the curvature radii .
the twosome @xmath26 forms a curvilinear coordinate system in the 2d plane .
the concerned electron moving along the wire is confined near the curve @xmath23 in the normal direction by a potential @xmath27 . in this curvilinear
coordinate system , the corresponding metric tensor is @xmath28 so the wave function should be normalized according to the condition @xmath29.\ ] ] it is convenient to introduce a new function @xmath30 to eliminate @xmath31 from the area element in the normalization condition .
after some substitution , the schrdinger equation is rewritten and the new function @xmath32 is separable in @xmath20 and @xmath33 , @xmath34 , since @xmath27 does not depend on @xmath20 .
the eigenfunction @xmath35 depicts the transversal motion in the confining potential @xmath27 . for small @xmath25 or deep confining potential ,
the confining energy in the transversal direction is much larger than the so energy and the kinetic energy in the longitudinal direction .
this allows us to divide the electron transversal motion from the longitudinal motion and the soi . in the limit of 1d wires , @xmath36
, the electron will be in the lowest transversal mode @xmath37 .
then we have an infinitely degenerate set of states @xmath38 where the @xmath39 denote a complete set of spinors in the @xmath20 direction .
accordingly the corresponding effective hamiltonian can be found in a way similar to ref .
24 . in order to obtain the exact expression of the effective hamiltonian
, we have to calculate the lowest transversal mode for a given confining potential . for generality and simplicity
, we assume the confining potentials to be symmetric : for example , the harmonic potential [ @xmath40 , the hard - wall potential or the @xmath41 potential .
we find that all these different potentials give the same results as a result of the symmetry of the potential .
the 1d effective hamiltonian in the presence of soi for a planar quantum wire is given by@xmath42,\]]where @xmath43 @xmath44 and @xmath45 @xmath46 , expressed by the usual pauli spin matrices @xmath47 , are the spin matrices on the normal and tangent direction in the curvilinear coordinate respectively .
here we have redefined @xmath0 ( @xmath48 ) .
obviously , the equation gives an effective geometric potential term @xmath49 in agreement with the term given by da costa@xmath18}$ ] .
specially the last term in eq .
( 5 ) @xmath50,\]]stands for the rashba soi in a planar curved quantum wire .
it can be easily seen that the geometry of the wire , via the curvature @xmath51 , has influence on the soi .
note that this 1d soi hamiltonian is universal for planar wire with arbitrary shape . as a consequence
, we can get the corresponding hamiltonian for a ring which is the same as meijer _
et al_ s@xmath52}$ ] .
the structure of the device we consider is shown schematically in fig .
1(b ) , consisting of two straight lines conjugated with an arc of a circumference@xmath53}$ ] . due to the presence of geometric potential ,
the curved section can be represented as a rectangular potential well ( see fig .
1(d ) ) with a width @xmath54 and a depth @xmath55 , where @xmath56 is the radius of the circumference and @xmath1 is the angle between two rectilinear parts of the wire .
there exits one and only one bound state for @xmath57 in such a system@xmath9}$ ] . in order to study the problem of electron transportation ,
we need to investigate the eigenstates of this system first . for convenience ,
the device is divided into three parts as shown in the fig .
1(c ) , and the two straight lines are assumed to be semi - infinite . to take into account the geometrical effects ,
the soi is considered for all the three sections . using the above 1deffective hamiltonian , we can easily write the hamiltonians for each part . in the selective coordinate of panel ( c ) in fig
. 1 , they read@xmath58,\text { } \theta \in \lbrack 0,\text { } \theta _ { w } ] , \nonumber \\ \hat{h}_{3 } & = -\frac{\hbar ^{2}}{2m}\frac{\partial ^{2}}{\partial s^{2}}-i\alpha \sigma _ { b_{3}}\frac{\partial } { \partial s},\text { } s\in ( \theta _ { w}\rho , \text { } \infty ) , \nonumber\end{aligned}\]]where
we have introduced the polar angle @xmath59 for the second part and substituted @xmath59 for @xmath20 .
the spin operators are in the forms@xmath60 we label the eigenstates of the system @xmath61 where @xmath62 denote the corresponding region of the device respectively , @xmath63 indicates the direction of motion along the wire and @xmath64 is the spin states . for the two parts of straight line , the energy spectrum @xmath65 and unnormalized eigenstates @xmath66 and @xmath67 are found to be@xmath68where @xmath69 takes @xmath70 at @xmath71 , or else @xmath72 .
here the spinors are given by @xmath73here we take the wave vector @xmath74 .
for the second part , i.e. the curved section , the corresponding eigenfunctions and eigenvalues are obtained similarly as @xmath75 where the orbit quantum number @xmath76 and @xmath69 is the same as eq .
( 10 ) . here
the spinors @xmath77 take the forms of @xmath78 where the angle @xmath79 , satisfying @xmath80 in the adiabatic limit @xmath81 , is given by @xmath82 with @xmath83 and @xmath84 .
the associated eigenenergies read@xmath85.\]]we will use these results in the following section to study thetransport properties .
we consider an electron which is transported from the 1st section to the 3rd section through the arc to study the transport properties of the system . in this case
it is appropriate to apply a spin - dependent version of the griffith s boundary conditions@xmath86}$ ] at the intersections .
this reduces the electron transport through the device to an exactly solvable 1d scattering problem .
accordingly there are two boundary conditions : ( i ) the wave functions must be continuous , and ( ii ) the spin probability current density must be conserved . due to the conservation of energy ,
the total wave functions corresponding to the same energy in the three regions can be written as linear combination of the corresponding eigenstates . in the presence of rashba coupling ,
the energy splitting is such that electrons with fermi energy @xmath87 have different wave numbers depending on spin ( @xmath88 ) and the direction of motion ( @xmath89 ) .
the quantities @xmath90 and @xmath91 are obtained by solving @xmath92 @xmath87 in eq .
( 9 ) and eq .
( 14 ) respectively .
let the incoming electron wave function @xmath93 be @xmath94 , where the parameter @xmath95 determines the probability amplitude of each eigenstate , the wave functions of the device read@xmath96respectively .
here @xmath97 stands for the reflection coefficients while @xmath98 are transmission coefficients for spin polarization @xmath99 respectively , and @xmath100 , @xmath101 , @xmath102 , @xmath103 are the corresponding coefficients of the wave function .
using the boundary conditions , we can determine the reflection and transmission probabilities which are expressed as the following forms : @xmath104^{2}}{(a+b)^{4}+(a - b)^{4}-2(a^{2}-b^{2})^{2}\cos ( 2\beta \theta _ { w } ) } , \\
r_{- } & = & \left| r_{-}\right| ^{2}=\frac{4[(a^{2}-b^{2})\sin ( \beta \theta _ { w})\sin \phi ] ^{2}}{(a+b)^{4}+(a - b)^{4}-2(a^{2}-b^{2})^{2}\cos ( 2\beta \theta _ { w } ) } , \\
t_{+ } & = & \left| t_{+}\right| ^{2}=\frac{8a^{2}b^{2}(\cos ^{2}\phi + \cos \varphi \cos ( \varphi -2\phi ) \cos [ ( 1 + 2\gamma ) \theta _ { w}]+\sin
^{2}(\varphi -\phi ) ) } { ( a+b)^{4}+(a - b)^{4}-2(a^{2}-b^{2})^{2}\cos ( 2\beta \theta _ { w } ) } , \\
t_{- } & = & \left| t_{-}\right| ^{2}=\frac{8a^{2}b^{2}(\cos ^{2}(\varphi -\phi ) -\cos \varphi \cos ( \varphi -2\phi ) \cos [ ( 1 + 2\gamma ) \theta _ { w}]+\sin ^{2}\phi ) } { ( a+b)^{4}+(a - b)^{4}-2(a^{2}-b^{2})^{2}\cos ( 2\beta \theta
_ { w})},\end{aligned}\ ] ] satisfying the normalization condition , @xmath105 .
here we have introduced @xmath106 , @xmath107 and @xmath108 .
the parameter @xmath109 and @xmath110 are defined by @xmath111 and @xmath112 , respectively .
finally , the conductance in mesoscopic structures can be expressed by means of the landauer conductance formula which in our case reads @xmath113where the conductance quantum @xmath114 . for numerical calculation
, we consider the curved 1d quantum channels based on inas .
the electrons at the fermi level in inas have the effective mass @xmath115 and the fermi energy @xmath116 mev . from eqs .
( 15 - 18 ) and ( 19 ) , we know that the device geometry , i.e. the radius @xmath56 of the arc and the angle @xmath1 , will affect the electron transmission . by using specific fabrication method , the scale of @xmath56 is chosen over a wide range : @xmath117 nm . while @xmath118 to avoid intersection .
in addition , the so strength @xmath0 is assumed to be a tunable quantity too .
for example , the soi strength @xmath0 can be controlled by a gate voltage with typical values in the range @xmath119 evm within an ingaas - based 2d electron gas@xmath120}$ ] .
first of all , we investigate the transmission of single incidentelectron with the polarized spin .
the incoming electron state is chosen to be @xmath121 , i.e. @xmath122 . so one can find that the reflection probability @xmath123 .
this means that there are no spin - flip process when the electron is reflected at the first junction .
another choice of the reversed electron spin incoming states will give the analogous results .
we investigate the relationship of reflection and transmission probabilities on the radius @xmath56 shown in fig .
2 . panel ( a - c ) are for different values of soi strength @xmath0 and fixed @xmath124 @xmath125 , while panel ( d - f ) are for different @xmath1 at a fixed @xmath0 . from these figures ,
one can see that the electron is reflected completely at @xmath126 .
this is induced by our model that the geometric potential becomes infinite , i.e. @xmath127 , when @xmath128 and @xmath126 .
in fact , the quantum wire has the width , so the radius @xmath56 can not achieve to zero for the condition of a smooth transition .
thereby this result may not be measurable .
we only deal with the conditions of @xmath129 nm . generally speaking , the reflection probabilities are very small . as @xmath56 increases ,
we find that the reflection probabilities @xmath130 decrease to zero quickly and the transmission probabilities @xmath131 and @xmath132 . for @xmath130 ,
the profiles of curves are slightly affected by the so coupling and mainly determined by the angle @xmath1 in panels ( a ) and ( d ) . from eqs .
( 15 - 18 ) , one can see that the so strength @xmath0 shows its effect through the two parameters @xmath133 and @xmath134 .
considering the ranges of all of the physical quantities , the term consisted @xmath135 is found to be smaller than other terms by three orders of magnitude at least . while for @xmath136 and @xmath137 , another parameter @xmath110 need to be considered .
so both the angle @xmath1 and the soi can affect the profiles of the curves obviously , as seen in panels ( b - c ) and ( e - f ) .
figure 3 shows the reflection and transmission probabilities as function of the angle @xmath1 for several @xmath56 and @xmath135 .
it is obvious that the profiles of the curves are symmetric on both sides of @xmath138 . for @xmath139 , we get @xmath140 and @xmath141 as expected . in our model ,
the 2nd section of the device vanishes and the device becomes a whole straight line when @xmath139 .
then the hamiltonians @xmath142 and @xmath143 in eq . ( 7 ) take the same form , the electron state should have no change through the process of transmission . in fig .
3 , the reflection and transmission probabilities are shown to be periodic functions . from equations ( 15 - 18 ) ,
the periods of the reflection and transmission probabilities are derived as@xmath144the amplitudes can also be calculated .
for example , the amplitude of @xmath130 is ( @xmath145 .
due to the same reason as fig .
2 , the profiles of @xmath130 vs. @xmath1 , including the periods and the amplitudes , are mainly determined by @xmath56 .
we plot the dependences of the probabilities on the soi @xmath0 for different @xmath1 and @xmath56 in figure 4 .
those oscillation vs. @xmath0 , influenced by the device geometry , remain for all of the curves . as increasing the so strength ,
the soi is enhanced and can weaken the geometric potential .
so one can find that the reflection probabilities @xmath146 tend to zero and the transmission probabilities @xmath131 and @xmath132 with the increscent @xmath0 .
when the soi is absent , the spin state of the outgoing electron is mainly determined by the angle @xmath1 .
now we turn to the electron conductance . figs .
5(a)-(b ) show the conductance vs. @xmath56 for different set of @xmath1 and @xmath0 . as increasing @xmath56 , the conductance increases rapidly and goes to @xmath147 oscillatorily .
the changes are mainly effected by the angle @xmath148 . while the relationships of the conductance to the angle @xmath1 , plotted in figs .
5(c ) and 5(d ) .
obviously , one can see that the conductance is periodical and symmetrical .
finally the dependences of the conductance on @xmath0 are shown in figs .
5(e ) and 5(f ) .
the conductance shows oscillation and goes to @xmath147 along with the increasing soi strength . as a whole
, we get @xmath149 @xmath150 using the normalization condition .
so the conductance has the similar behavior with the reflection probability @xmath130 . for further quantitative analysis
, we can see that the effects of the device geometry and the soi are very small and the conductance shows a periodic variation in the @xmath151 .
so these effects on the conductance may be difficult to be observed .
furthermore , we consider the electron energy in a wide range , not only localized on the fermi level .
6 shows the relationship of the conductance versus the electron energy @xmath152 in units of @xmath87 for different values of the soi strength @xmath0 , the angle @xmath1 and the radius @xmath56 .
one can see that the conductance also has the behavior of oscillation and goes to @xmath147 with the increase of @xmath152 . especially , there are resonant transmissions at certain energy , similar to the scattering problem of typical square well potential . finally , we turn to the incoming electron with the mixed state . from eq .
( 19 ) , one can easily see that the conductance is independent of the parameter @xmath95 . in order to study the influence on the transmission for different incoming electron state
, we calculate the polarization ratio @xmath153 for the outgoing electron .
considering eqs .
( 17 ) and ( 18 ) , the polarization ratio @xmath154 takes the form@xmath155+\sin ( \varphi ) \sin ( \varphi -2\phi ) .\]]fig .
7 shows the dependences of @xmath154 on the parameter @xmath95 for different values of @xmath56 , @xmath0 and @xmath1 .
it is obvious that the profiles of all curves are periodical on @xmath95 .
the period of the polarization ratio @xmath154 can be found to be @xmath156 through eq .
the outgoing electron state can be spin polarized when we choose the appropriate parameters . especially for the spin polarized incident electron , such as @xmath122
, we investigate the relationship of the polarization ratio @xmath154 on the device geometry and the soi shown in fig . 8 .
we find that the spin state of the outgoing electron can also be polarized and just rotated around the @xmath16 direction with the angle @xmath1 as the condition of @xmath157=1 $ ] is satisfied .
in conclusion , we have derived the 1d effective hamiltonian for a planar curvilinear quantum wire in the presence of rashba soi .
the 1d form of the rashba soi is obtained analytically .
then we consider the electron transport in a quantum wire consisting of two straight lines conjugated with an arc of a circumference .
for the electron transportation of spin polarized incidence , we calculate the reflection and transmission probabilities and find that the electron transmission is influenced by three parameters : the device geometry ( @xmath1 and @xmath56 ) and the soi strength @xmath0 .
the relationships between them are discussed in detail through numerical analysis .
the device geometry can influence the electron transportation obviously , so does the so strength @xmath0 except for the electron reflection .
there is no spin - flip reflection for the device .
the general behavior of oscillating versus @xmath56 , @xmath1 or @xmath0 are found , especially for the angle @xmath1 the electron transportation shows the periodic character . using the landauer formalism
, we obtain that the conductance varies near @xmath147 and goes to @xmath147 oscillatorily .
the relationship of the conductance vs. the electron energy is also investigated .
the electron resonant transmissions occur at some certain energy . finally , we study the influence of the incoming electron state on the electron transmission , especially the spin polarization ratio for the outgoing electron .
we can achieve the rotated spin polarized electron output under certain conditions .
there is one point that should be noted here . in our system ,
the soi is considered for the whole device for the sake of the geometrical effects .
this also makes the soi controllable through the electron field easily . on the side ,
the system without the soi on the two straight lines is also investigated and will be published later .
even so , the study does give us one way to modulate the electron transmission through the correlative parameters , especially the device geometry .
as an example , the device is placed on an elastomeric substrate , then the quantum wire can deform in company with the substrate .
thereby the wire geometry changes when the substratum deforms ; accordingly the electron transmission changes .
we may determine the deformation if we can detect the changes of the electron transmission .
furthermore , this study also suggests that we may construct the custom - built device through the geometry method .
we thank s. zhao , l. zhang , r. liang , y. liu , and z. yao for helpful discussions .
this work is supported by nsf of china under grant no .
e. zhang is also supported by the doctoral foundation grant of xian jiaotong university ( xjtu ) no .
dfxjtu2004 - 10 and by the nsf of xjtu ( grant no . 0900 - 573042 ) .
|
the one - dimensional effective hamiltonian for a planar curvilinear quantum wire with arbitrary shape is proposed in the presence of the rashba spin - orbit interaction .
single electron propagation through a device of two straight lines conjugated with an arc has been investigated and the analytic expressions of the reflection and transmission probabilities have been derived .
the effects of the device geometry and the spin - orbit coupling strength @xmath0 on the reflection and transmission probabilities and the conductance are investigated in the case of spin polarized electron incidence .
we find that no spin - flip exists in the reflection of the first junction .
the reflection probabilities are mainly influenced by the arc angle and the radius , while the transmission probabilities are affected by both spin - orbit coupling and the device geometry .
the probabilities and the conductance take the general behavior of oscillation versus the device geometry parameters and @xmath0 .
especially the electron transportation varies periodically versus the arc angle @xmath1 .
we also investigate the relationship between the conductance and the electron energy , and find that electron resonant transmission occurs for certain energy .
finally , the electron transmission for the incoming electron with arbitrary state is considered . for the outgoing electron ,
the polarization ratio is obtained and the effects of the incoming electron state are discussed .
we find that the outgoing electron state can be spin polarization and reveal the polarized conditions .
|
since their beginnings about half a century ago @xcite , bayesian optimization algorithms have been increasingly used for derivative - free global minimization of expensive to evaluate functions . typically assuming a continuous objective function @xmath0 , single - objective bayesian optimization algorithms consist in sequentially evaluating @xmath1 at promising points under the assumption that @xmath1 is a sample realization ( _ path _ or _ trajectory _ ) of a random field @xmath2 .
such algorithms are especially popular in the case where evaluating @xmath3 requires heavy high - fidelity numerical simulations ( or _ computer experiments _ , see notably @xcite ) , where @xmath4 stands for some design parameters to be optimized over .
such expensive simulations are classically encountered in the resolution of partial differential equations from physical sciences , engineering and beyond @xcite . in recent years
, bayesian optimization also has attracted a lot of interest from the machine learning community @xcite , be it to optimize simulation - based objective functions @xcite or even to estimate tuning parameters of machine learning algorithms themselves @xcite . in both communities , a gaussian random field ( or _ gaussian process _ , gp ) model is often used for @xmath5 , so that prior information on @xmath1 is taken into account through a trend function @xmath6 and a covariance kernel @xmath7 .
once @xmath8 and @xmath9 are specified , possibly up to some parameters to be inferred based on data , the considered gp model can be used as an instrument to locate the next evaluation point(s ) via so - called infill sampling criteria , also referred to as _ acquisition functions _ or simply as _
criteria_. while a number of bayesian optimization criteria have been proposed in the literature ( see , e.g. , @xcite and references therein ) , we concentrate here essentially on the _ expected improvement _ ( ei ) criterion @xcite and on variations thereof , with a focus on its use in synchronous batch - sequential optimization .
denoting by @xmath10 points were @xmath1 is assumed to have already been evaluated and by @xmath11 a batch of candidate points where to evaluate @xmath1 next , the multipoint ei is defined as @xmath12 where @xmath13 refers to the conditional expectation knowing the event @xmath14 .
one way of calculating such criterion is to rely on monte carlo simulations .
figure [ fig : explanation ] illustrates both what the criterion means and how to approach it by simulations , relying on three samples from the multivariate gaussian distribution underlying equation .
our main focus here , in contrast , is on deriving equation in closed form , studying the criterion s differentiability , and ultimately calculating and efficiently approximating its gradient in order to perform efficient batch optimization using derivative - based deterministic search . for @xmath15 , @xmath16 , @xmath17 .
left : gaussian process prediction of a function @xmath1 from observations @xmath18 ( depicted by black crosses ) .
the green horizontal line stands for @xmath19 , the smallest response value from @xmath18 .
three conditional simulation draws are plotted in orange and various point symbols represent their respective values at two unobserved locations @xmath20 and @xmath21 .
right : distribution of the random vector @xmath22 knowing @xmath18 ( black contours ) . for each point
symbol , the length of the purple segment represents the improvement realized by the corresponding sample path .
the multipoint ei is the expectation of this length , or in other words , it is the integral of the improvement ( grey - scale function ) with respect to the conditional distribution of @xmath22 knowing @xmath18 . ]
now , for @xmath23 , it is well known that ei can be expressed in closed form as a function of @xmath24 and @xmath25 as follows @xmath26 where @xmath27 ( defined for @xmath28 ) and @xmath29 are the cumulative distribution function and probability density function of the standard gaussian distribution , respectively .
when deriving equation , equation happens ( hence for @xmath23 ) to involve a first order moment of the truncated univariate gaussian distribution .
as shown in @xcite and developed further here , it turns out that equation can be expanded in a similar way in the multipoint case ( @xmath30 ) relying on moments of truncated gaussian vectors .
this is essential for the open challenges tackled here of efficiently calculating and optimizing the multipoint criterion of equation .
the applied motivation for having batch - sequential ei algorithms is strong , as distributing evaluations of bayesian optimization algorithms over several computing units allows significantly reducing wall - clock time and with the fast popularization of clouds , clusters and gpus in recent years it is becoming always more commonplace to launch several calculations in parallel . even at a slightly inflated price and
scripting effort , reducing the total time off is often a primary goal in order to deliver conclusions involving heavy experiments , be they numerical or laboratory experiments , in studies subject to hard time limitations .
obviously , given its practical importance , the question of parallelizing ei algorithms and alike by selecting @xmath31 points per iteration has been already tackled in a number of works from various disciplinary horizons ( including notably @xcite ) .
here we essentially focus on approaches relying on the maximization of equation and related multipoint criteria .
the multipoint ei of equation has been defined in @xcite and first calculated in closed form for the case @xmath17 in @xcite . for the case
@xmath32 , a monte carlo scheme and some sub - optimal batch selection strategies were proposed .
further work on monte carlo simulations for multipoint ei estimation can be found in @xcite ; besides this , stochastic simulation ideas have been explored in @xcite for maximizing this multipoint ei criterion via a stochastic gradient algorithm , an approach recently investigated in @xcite . meanwhile , a closed - form formula for the multipoint ei relying on combinations of @xmath33-and @xmath34-dimensional gaussian cumulative distribution functions was obtained in @xcite , a formula which applicability in reasonable time is however restricted to moderate @xmath34 ( say @xmath35 ) in the current situation . building upon @xcite ,
@xcite recently calculated the gradient of the multipoint ei criterion in closed form and obtained some first experimental results on ( non - stochastic ) gradient - based multipoint ei maximization .
our aim in the present paper is to present a set of novel analytical and numerical results pertaining to the calculation , the computation , and the maximization of the multipoint ei criterion . as most of these novel results apply to a broader class of criteria , we first present in section [ sec : framework ] a generalization of the multipoint ei that allows accounting for noise in conditioning observations and also exponentiating the improvement .
this generalized criterion is calculated using moments of truncated gaussian vectors in the flavour of @xcite .
the obtained formula is then revisited in the standard case ( noise - free with an exponent set to @xmath36 ) , leading to a numerical approximation of the multipoint ei with arbitrary precision and very significantly reduced computation time .
next , the @xmath37-dimensional maximization of the multipoint ei criterion is discussed in section [ sec : discussion ] , where the differentiability of the generalized criterion is studied , its analytical gradient is calculated , and further numerical approaches for fast gradient approximations with controllable accuracy are presented . finally , section [ sec : application ] is dedicated to numerical experiments where , in particular , a multistart derivative - based multipoint ei maximization algorithm highlighting the benefits of the considered methodological principles and the proposed fast approximations is tested and compared to baseline strategies .
throughout this section the objective function @xmath1 may be observed noise - free or in noise , meaning that at some arbitrary iteration @xmath38 the observed value may be @xmath39 or @xmath40 where @xmath41 is a realization of a zero mean gaussian random variable with known ( or estimated and plugged - in ) variance .
@xmath1 is assumed to be one realization of a random field @xmath5 , where @xmath5 has a gaussian random field ( grf ) distribution conditionally to events of the form @xmath42 ( with conditioning on @xmath43 in the noisy case , see for instance @xcite ) .
this setup naturally includes the case where @xmath5 is a grf , but also the so - called universal kriging settings where @xmath5 is the sum of a trend with an improper prior and a grf @xcite . note that in noisy cases the @xmath41 s are generally assumed to be independent ( although the case of @xmath41 s forming a gaussian vector is tractable ) , but more essentially they are assumed independent of @xmath5 .
in batch - sequential bayesian optimization we are interested in computing sampling criteria @xmath44 depending on @xmath45 new points @xmath46 . at any step of corresponding synchronous parallel algorithms , the next batch of @xmath34 points @xmath47
is then defined by globally maximizing @xmath44 over all possible batches : @xmath48 values of such criteria typically depend on @xmath49 through the conditional distribution @xmath50 , simplifying to @xmath51 in the noiseless context .
conditional mean and covariance functions are analytically formulated via the so - called _ kriging equations _ , see e.g. @xcite . working out these criteria
thus generally boils down to gaussian vector calculus , which may become intricate and quite cumbersome to implement as @xmath34 ( or @xmath52 , in noisy settings ) increases .
our considered generalized version of the multipoint ei criterion , that allows accounting for a gaussian noise in the conditioning observations and also for an exponentiation in the definition of the improvement , is defined as : @xmath53 where @xmath54 , @xmath55 and @xmath56 .
this form gathers several sampling criteria notably including @xmath57 , both in noiseless and noisy settings , and also a multipoint version of the generalized ei of @xcite .
in addition , the obtained results apply to batch - sequential versions of the expected quantile improvement @xcite ( eqi ) and variations thereof , by a simply change of process from @xmath5 to the quantile process
. we will show in proposition [ proposition : analyticqei ] that such generalized multipoint ei criteria can be formulated as a sum of moments of truncated gaussian vectors . in the next subsection , in order to get a closed form for the generalized ei we first define these moments and derive some first analytical formulas , that might also be of relevance in further contexts .
we fix @xmath58 and @xmath59 in noisy settings or @xmath60 in noiseless settings .
+ let @xmath61 be a gaussian vector with mean @xmath62 and covariance matrix @xmath63 , where @xmath64 is the cone of positive definite matrices of @xmath65 .
for all positive integer @xmath66 , we define the function @xmath67 on @xmath68 by @xmath69 where the inequality @xmath70 is to be interpreted component - wise .
if @xmath61 is composed of values of a grf at a batch of @xmath34 locations @xmath49 , we use the notation @xmath71 .
we obtain the moments @xmath72 of a truncated gaussian distribution by an extension of tallis technique @xcite to any order , presented in the following proposition : [ proposition : analyticalmoment ] the function @xmath73 defined by @xmath74 where @xmath75 is the cumulative distribution function of the centered @xmath76-variate normal distribution , is infinitely differentiable , and the moments @xmath67 are given by : @xmath77 the proof of this proposition is given in appendix [ appendix : analyticalmoment ] and relies on calculating the moment generating function @xmath78 . even if an analytical formula can be obtained at any order of differentiation @xmath79 , the complexity of derivatives in equation increases rapidly .
we give below the results for @xmath79 equals 1 and 2 .
differentiating @xmath81 with respect to @xmath82 yields : @xmath83 where @xmath84 is the gradient of @xmath85 ( see appendix [ appendix : cdfgradient ] for an analytical derivation ) .
taking @xmath86 in the previous equation gives @xmath87 where @xmath88 is the @xmath89 column of @xmath90 .
it is shown in appendix [ appendix : cdfgradient ] that computing each of the @xmath76 components of @xmath84 requires to compute a multivariate cdf of the normal distribution in dimension @xmath91 .
the number of calls to this function for computing the first moment of the truncated gaussian distribution is thus of @xmath92 .
similarly , differentiating @xmath81 twice with respect to @xmath82 yields @xmath94 for readability , the detailed formula of @xmath95 , the hessian matrix of @xmath85 , is sent to appendix [ appendix : cdfhessian ] .
the number of calls to the multivariate normal cdf is of @xmath96 .
the previous results obtained for the moments of the truncated normal distribution turn out to be of interest for computing the generalized @xmath57 introduced in equation , as shown by the following proposition . [ proposition : analyticqei ] for @xmath97 , the criterion @xmath98 defined by exists for all @xmath79 and can be written as a sum of moments of truncated normal distributions @xmath99 with @xmath100 a vector of size @xmath101 defined , noting @xmath102 , by + @xmath103 [ cols= " < , < " , ] indeed , the first @xmath104 components of @xmath105 reflect @xmath106 , and the last components reflect @xmath107 and @xmath108 .
let @xmath110 be an event , @xmath5 be a mean - squared differentiable gaussian process and @xmath111 .
then we have : @xmath112 by mean - squared differentiability of @xmath113 .
the term @xmath114 comes from a symmetry occurring when summing terms with different index but actually equal . at fixed summation index @xmath115 in ,
we denote @xmath116 the @xmath117 term in the scalar product in for each @xmath118 required for @xmath57 : @xmath119_i.\ ] ] then the following symmetry between indices @xmath38 and @xmath9 occurs : @xmath120 indeed , using the formula of the derivative of cdf , ( appendix [ appendix : cdfgradient ] ) , leads to : @xmath121 , _ a tutorial on bayesian optimization of expensive cost functions , with application to active user modeling and hierarchical reinforcement learning .
report , dept . of computer science , university of british columbia , 2009 . , _ _ learning and intelligent optimization - 7th international conference , lion 7 , catania , italy , january 7 - 11 , 2013 , revised selected papers _ , chapter fast computation of the multipoint expected improvement with applications in batch selection , pages 59 - 69 _ , springer , ( 2014 . ) .
, _ expected improvements for the asynchronous parallel global optimization of expensive functions : potentials and challenges _ , in lion 6 conference ( learning and intelligent optimization ) , paris : france , 2012 . , _ machine learning , optimization , and big data : first international workshop , mod 2015 , taormina , sicily , italy , july 21 - 23 , 2015 , revised selected papers _ , springer international publishing , cham , 2015 , ch . differentiating the multipoint expected improvement for optimal batch design , pp .
|
we deal with the efficient parallelization of bayesian global optimization algorithms , and more specifically of those based on the expected improvement criterion and its variants .
a closed form formula relying on multivariate gaussian cumulative distribution functions is established for a generalized version of the multipoint expected improvement criterion . in turn , the latter relies on intermediate results that could be of independent interest concerning moments of truncated gaussian vectors .
the obtained expansion of the criterion enables studying its differentiability with respect to point batches and calculating the corresponding gradient in closed form .
furthermore , we derive fast numerical approximations of this gradient and propose efficient batch optimization strategies .
numerical experiments illustrate that the proposed approaches enable computational savings of between one and two order of magnitudes , hence enabling derivative - based batch - sequential acquisition function maximization to become a practically implementable and efficient standard .
* keywords : * kriging , expected improvement , parallel optimization .
|
since the discovery of a bow shock in the periphery of the bullet cluster 1e 0657 - 558 @xcite , it has been well established that shock waves exist in and around galaxy clusters .
using cosmological hydrodynamic simulations for the large - scale structure ( lss ) formation of the universe , the origin and nature of shock waves in the intracluster medium ( icm ) as well as in the intergalactic medium ( igm ) have been extensively studied @xcite .
these studies demonstrated that abundant shocks are produced by supersonic flow motions during the process of hierarchical clustering of nonlinear structures , and that they could be classified into two categories , according to their locations relative to the host structures . _
external shocks _ are formed at the outermost surfaces surrounding clusters , filaments , and sheets of galaxies by the accretion of cool ( @xmath1 k ) , tenuous gas in voids onto those nonlinear structures . since the accretion velocity around clusters can be as high as @xmath2 a few @xmath3 and the sound speed of accreting gas
is @xmath4 , external shocks are strong with mach number as large as @xmath5 . on the other hand , _ internal shocks _ are produced inside the nonlinear structures , where the gas has been already heated to high temperature by previous episodes of shock passage , so their mach number is typically low with @xmath6 . while most of internal shocks have @xmath7 , those with @xmath8 play the most important role in dissipating the shock kinetic energy into heat in the icm ( e.g. , * ? ? ?
* ; * ? ? ?
internal shocks could be further classified by their origins into a number of types , which in fact may not be mutually exclusive .
_ turbulent shocks _ are induced by turbulent flow motions in the icm and they are expected to have very small mach numbers , @xmath0 @xcite .
turbulent motions in the icm could be generated by several different processes : major or minor mergers , agn jets , galactic winds , galaxy wakes , and etc ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
bow shocks found in the outskirts of merging clusters are commonly referred as _ merger shocks _ @xcite . in most cases merger shocks are weak with @xmath7 ( e.g. , * ? ? ?
_ infall shocks _ form by infall of the warm - hot intergalactic medium ( whim ; @xmath9 k ) into the hot icm along adjacent filaments @xcite .
they could have relatively high mach numbers , reaching up to @xmath10 , and thus the ensuing cosmic - ray ( cr ) acceleration can be more efficient than in other types of shocks . including the shock in the bullet cluster , a number of shocks have been found in x - ray observations ( e.g. , * ? ? ?
? * ; * ? ? ?
* ) . in these observations ,
shocks are detected as sharp discontinuities in the temperature and surface brightness distributions , and their physical properties including the sonic mach number , @xmath11 , are estimated from the ` deprojected ' temperature and density jumps @xcite .
most of the shocks detected in x - ray observations are weak with @xmath12 .
in addition , shocks in the icm have been detected by radio observations , especially as the so - called radio relics ( see , e.g. , * ? ? ? * ; * ? ? ? * for reviews ) .
radio relics are usually found in the cluster outskirts around the virial radius , @xmath13 .
the radio emission is understood as synchrotron radiation from cr electrons with the lorentz factor of @xmath14 that are believed to be accelerated at the shocks associated with them ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
high energy nonthermal particles can be produced via diffusive shock acceleration ( dsa ) at astrophysical shocks , such as interplanetary and supernova remnant shocks as well as cluster shocks in collisionless tenuous plasma @xcite .
moreover , it has been shown that turbulence flow motions in the icm can produce magnetic fields of up to @xmath15 level ( e.g. , * ? ? ? * ) . since the radio - emitting electrons with the lorentz factor of @xmath14 would not advect nor diffuse in the icm more than @xmath16 kpc away from the shock surface before they lose the energy due to radiative cooling via synchrotron emission and inverse compton ( ic ) scattering ( e.g. , * ? ? ?
* ) , it is commonly thought that their acceleration sites are likely to be close to where the synchrotron emission is seen .
so the physical properties of shocks in radio relics are inferred from observed quantities such as the injection spectral index at the shock edge , @xmath17 , and the spatial profile of surface brightness @xcite . note that the spectral index for the synchrotron spectrum _ integrated _ over the downstream behind a steady planar shock is expected to be @xmath18 ( e.g. * ? ? ?
* ; * ? ? ?
* ) . in most cases ,
the mach number of shocks in radio relics was found to be in the range of @xmath19 ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
pre - acceleration of thermal electrons to suprathermal energies and subsequent injection into the fermi first - order process at shock have been one of the outstanding problems in the dsa theory .
it is thought that the particle injection and dsa acceleration might be very inefficient at weak shocks ( @xmath20 ) because of small density compression across shocks ( e.g. , * ? ? ?
especially , in the so - called `` thermal leakage '' injection model , proton injection is expected to be strongly suppressed at weak shocks ; the relative difference between postshock , proton thermal and flow speeds is greater and so it is less likely for postshock protons to recross the shock front at weaker shocks ( e.g. , * ? ? ?
although postshock electrons move faster than protons , they are tied to magnetic field fluctuations more tightly because of smaller rigidities ( i.e. @xmath21 ) .
so it was speculated that the dsa of electrons at weak shocks would not be efficient either , for instance , not enough to explain the observed radio flux of spectacular giant radio relics such as the sausage relic .
a pre - exiting population of electrons with the lorentz factor of @xmath22 was suggested as a possible solution to the low electron injection problem at weak cluster shocks @xcite .
moreover , @xcite has suggested that a @xmath23-like distribution of suprathermal electrons may exist in high beta ( @xmath24 ) icm plasmas just like in solar winds , and facilitate the electron injection at weak shocks .
recently , using particle - in - cell ( pic ) simulations of weak shocks in high beta plasmas , @xcite and @xcite have shown that some of the incoming electrons gain energy via shock drift acceleration ( sda ) and are reflected specularly toward the upstream region .
those reflected particles can be scattered back to the shock surface by plasma waves excited in the upstream region , and then undergo multiple cycles of sda , resulting in a power - law type suprathermal population up to @xmath25 .
these studies suggest a possibility that `` self pre - acceleration '' of thermal electrons to suprathermal energies via kinetic plasma processes at the shock itself might provide seed electrons enough to explain the observed flux level of bright radio relics .
the morphology of radio relics is , in some cases , observed to be elongated and arc - like with a sharp edge on one side . and
some of radio relics are found as a pair in the opposite side of clusters .
so they are often interpreted as products of binary mergers ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
several numerical studies have suggested that merger shocks with sufficient amount of shock kinetic energy flux could produce radio relics @xcite . in these studies ,
synthetic radio maps were constructed by identifying shocks in simulated clusters and modeling cr electron injection / acceleration and magnetic field strength .
however , there remain a few issues to be resolved , before the simple picture of merger shocks being the origin of radio relics is accepted .
the first issue concerns the frequency of observed radio relics .
although structure formation simulations have demonstrated that shocks are produced frequently during mergers and they should last for the cluster dynamical time of @xmath26 gyr , only about @xmath27 of x - ray luminous clusters host some radio relics , putative merger shocks , and the fraction of merging clusters with giant radio relics is even much lower @xcite .
recently , it has been suggested that icm shocks may light up as radio relics only when they encounter fossil relativistic electrons that are left over from either a previous episode of shock / turbulence acceleration or a radio jet from agn @xcite . in such a scenario , only a small faction of icm shocks become radio - emitting structures for a fraction of the dynamical time ( @xmath28 myr ) . so the rareness of radio relics could be explained .
of course , this model can be justified only if the injection and acceleration of electrons at weak shocks in the icm is very inefficient , so that the re - acceleration of pre - existing cr electrons must be required for the birth of radio relics .
the second issue involves the discrepancies in the shock properties inferred from radio and x - ray observations of a few radio relics . among several dozens of observed radio relics , only a fraction of shocks associated with them
also have been detected in x - ray observations ( * ? ? ?
* and references therein ) . in the case of the toothbrush relic in 1rxs j0603.3 , for exmaple , the radio index was measured to be @xmath29 , indicating a ` radio mach number ' , @xmath30 @xcite .
but the temperature and density discontinuities in x - ray observations suggest an ` x - ray mach number ' , @xmath31 , and the position of shock identified in x - ray observations is shifted from that in radio observations by @xmath32 kpc @xcite .
in addition , @xcite suggested that for the radio relic in a2256 , if the observed index @xmath33 measured between 63 and 1360 mhz is interpreted as the injection index , the radio mach number can be estimated to be @xmath34 , while the temperature jump measured in x - ray observations implies the x - ray mach number , @xmath35 . in the case of the sausage relic in ciza j2242.8 , @xcite used the observed radio spectral index near the edge ( shock surface ) , @xmath36 , to obtain a radio - inferred shock mach number , @xmath37 . but x - ray - inferred mach number reported later by suzaku and chandra observations turned out to be lower with @xmath38 @xcite . recently , however , @xcite estimated a steeper value , @xmath39 , by performing a spatially - resolved spectral fitting , implying a weaker shock with @xmath40 in good agreement with x - ray observations . in ( * ? ? ? * hereafter , paper i ) , we studied properties of shock waves in the outskirts of simulated galaxy clusters using sets of lss formation simulations .
we find that in addition to merger shocks , infall shocks are produced in icms by continuous infall of density clumps ( i.e. minor mergers ) along filaments of galaxies to hot icms .
unlike weak bow shocks ( @xmath0 ) driven by major mergers , infall shocks do not show pairing structures and have higher mach numbers ( @xmath41 ) . in a few cases ( e.g. , coma 1253 + 275 and ngc 1265 ) , observed radio relics
are thought to be associated with infall shocks @xcite . in this paper
, we consider a scenario in which suprathermal electrons are injected directly at the shocks without a help of fossil cr electrons , and accelerated to radio - emitting energies .
then the frequency of radio relics depends on the shock statistics , while the radio luminosity of radio relics is determined by the shock kinetic energy flux and the assumed dsa efficiency model ( e.g. * ? ? ? * ) . in this scenario
the rareness of radio relics , compared to the frequency of the icm shocks , can be controlled by adjusting the models for the dsa acceleration efficiency and the magnetic field amplification .
for instance , @xcite showed , using structure formation simulations and generating mock radio maps of simulated cluster samples , that radio emission tends to increase toward the cluster periphery and peak around @xmath42 , mainly because the kinetic energy dissipated at shocks peaks around @xmath43 and the mach number of shocks tend to increase toward the cluster outskirts .
such findings can explain why radio relics are rarely found in the cluster central regions .
we study the properties of radio and x - ray shocks in the simulated cluster sample of paper i , based on a dsa model in which suprathermal electrons are injected and accelerated preferentially at shocks with higher mach numbers .
we first calculate synchrotron emission at shocks by modeling primary cr electron population based on recent dsa simulations @xcite and magnetic field distribution based on a turbulent dynamo model @xcite .
we produce mock radio and x - ray maps by projecting the synchrotron and bremsstrahlung emission , respectively .
we then identify `` synthetic radio relics '' in the radio map and extract their properties , such as radio and x - ray shock mach numbers , @xmath44 and @xmath45 , and their locations in the map .
we examine the radio and x - ray properties of relic shocks , and attempt to understand discrepancies derived from radio and x - ray observations in some radio relics . in section
[ sec : numerics ] , we present numerical details such as the calculation of synchrotron emission at shock , the construction of mock radio and x - ray maps , and the extraction of radio and x - ray shock mach numbers . in section [ sec : projection ] , we discuss how the two - dimensional ( 2d ) projection of three - dimensional ( 3d ) shock distributions affects the radio and x - ray observations of icm shocks . in section [ sec :
stats ] , we describe the properties of shocks in 2d projection as well as the properties for synthetic radio relics .
summary follows in section [ sec : summary ] .
the construction of galaxy cluster sample and the identification of shocks are described in detail in paper i , so here they are summarized only briefly .
the lss formation simulations adopted a standard @xmath46cdm model with the following parameters : baryon density fraction @xmath47 , dark matter density fraction @xmath48 , cosmological constant fraction @xmath49 , hubble parameter @xmath50 , rms density fluctuation @xmath51 , and the primordial spectral index @xmath52 .
an updated version of a particle - mesh / eulerian hydrodynamic code described in @xcite was used for the simulations .
three sets of simulations were performed : ( 1 ) 16 different realizations of lss formation in a comoving cubic box of @xmath53 with @xmath54 uniform grid zones , ( 2 ) 16 different realizations in a comoving cubic box of @xmath55 with @xmath54 zones , and ( 3 ) one realizations in a comoving cubic box of @xmath53 with @xmath56 zones .
the sets of ( 1 ) and ( 2 ) were performed in an adiabatic ( non - radiative ) way , while ( 3 ) includes a mild feedback from star formation and cooling / heating processes . in paper
i , we find that , in all three types of simulations , the statistics of the physical properties of clusters and shocks are similar . in the simulation data ,
clusters are identified as regions around local peaks in the spatial distribution of x - ray emissivity . around each peak , the x - ray emission - weighted mean temperature , @xmath57 , is obtained for the spherical volume within @xmath58 . here , @xmath59 is the radius within which the gas overdensity is 200 times the mean gas density of the universe . those with @xmath60 kev for @xmath61 box simulations and @xmath62 kev for @xmath63 box simulations are selected as synthetic clusters , resulting in a total of 228 clusters from the three sets of simulations .
here , @xmath64 is the boltzmann constant shocks ( actually , shock zones ) are identified by a set of criteria given in paper i , and their sonic mach numbers , @xmath11 , are estimated from the temperature jump condition , @xmath65 @xmath66 , where @xmath67 is the gas temperature .
hereafter , the subscripts `` 1 '' and `` 2 '' indicate the preshock and postshock quantities , respectively . only shocks with @xmath68
are considered .
note that a shock surface normally consists of many of these shock zones .
the shock speed and shock kinetic energy flux at shock zone are , then , calculated as @xmath69 and @xmath70 , respectively , where @xmath71 is the preshock thermal pressure and @xmath72 is the gas adiabatic index .
the energy flux of cr protons can be estimated as @xmath73 where @xmath74 is the cr acceleration efficiency via the dsa process .
for the efficiency , the values presented in @xcite ( also shown in figure 2 of paper i ) are adopted . in paper
i , we find that the morphology of shock surfaces is quite complex due to the dynamic history of clusters , and in general a connected shock surface can consist of a number of grid zones of different types of shocks including merger shocks ( see figure 7 in paper i ) . to calculate radio synchrotron emission at cluster shocks , we need to model the strength of magnetic fields as well as the energy spectrum of cr electrons .
the icm is observed to be permeated with magnetic fields of a few @xmath75 g in the cluster core region and a few @xmath76 g in the cluster periphery ( e.g. * ? ? ?
* ; * ? ? ?
. suggested ideas for the generation and amplification of magnetic fields in the icm include processes during primordial phase transitions , biermann battery mechanism , plasma processes at collisionless shocks , different types of turbulence dynamo , and the ejection of galactic winds and agn jets ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . here
we adopt the turbulent dynamo model of @xcite , which assumes that turbulent flow motions are induced via stretching and compression of the vorticity generated behind curved surfaces of shocks in clusters , and the magnetic fields are amplified by the turbulent flows .
then the strength of the resulting magnetic fields can be modeled in terms of the number of local eddy turn - over by the following fitting formula : @xmath77 & \textrm { for } t / t_{\rm eddy } < 4 \\ ( 0.36/41 ) \cdot ( t / t_{\rm eddy}-4 ) + 0.04 & \textrm { for } t / t_{\rm eddy } > 4 \end{array } \right . , \label{bdynamo}\ ] ] where @xmath78 is the turbulent energy density and @xmath79 is the reciprocal of the vorticity calculated from the local flow speed .
the fitting function @xmath80 represents the fraction of the turbulent energy transferred to the magnetic energy via turbulent dynamo , and it is derived from a magneto - hydrodynamic simulation @xcite .
this model predicts that the magnetic field strength reaches to a few @xmath81 in the cluster center and decreases to @xmath82 g toward the cluster outskirts .
this is in a good agreement with the observed magnetic field strength in actual clusters . for the energy spectrum of cr electrons
, we first assume that the cr acceleration at shock is described by the test - particle model @xcite , since most of the shocks found in clusters are weak ( see introduction ) .
then , the momentum distribution of cr protons at the shock position is described by the power - law form for @xmath83 , @xmath84 where @xmath85 is the dimensionless proton momentum , @xmath86 is the proton mass , and @xmath87 is the spectral index .
the shock compression ratio is determined by the sonic mach number as @xmath88}$ ] .
the normalization factor @xmath89 at each shock zone is determined by setting the cr energy flux through the shock zone as @xmath90 where @xmath91 is the postshock flow speed and @xmath92 is given in equation ( [ fcrp ] ) .
we do _ not _ consider the re - acceleration of pre - existing crs in this work .
here , @xmath93 is the injection momentum above which particles can participate in the dsa process .
according to recent pic and hybrid simulations , both protons and electrons initially can gain energies via sda while confined between the shock front and the preshock region by scatterings due to self - generated upstream waves ( * ? ? ?
* ; * ? ? ?
* see introduction ) .
particles should have the rigidity ( @xmath94 ) large enough to cross the shock front , i.e. , @xmath95 , in order to take part in the full first - order fermi process .
here , @xmath96 is the most probable momentum of thermal protons of postshock gas with temperature @xmath97 , while the most probable momentum of thermal electrons is @xmath21 .
in fact , @xmath98 is expected to depend on the shock mach number , obliquity angle , and shock speed . here , for the sake of simplicity , we set @xmath99 in all shock zones regardless of the shock speed and mach number and the icm temperature .
since the dsa process operates on cr protons and electrons with the same rigidity in the same manner , the momentum distribution of primary cr electrons at shocks should follow that of cr protons , except less efficient injection and radiative cooling ( e.g. , * ? ? ?
* ; * ? ? ?
the injection rate of cr electrons is expected be much lower than that of cr protons , because postshock thermal electrons need to be pre - accelerated from the thermal momentum to @xmath98 in order to get injected into the dsa process . since the electron pre - acceleration is not yet fully constrained by plasma physics ( despite of recent pic simulations , see introduction ) , it is often parameterized by the cr electron - to - proton ratio , @xmath100 .
different types of observations have indicated a wide range of @xmath101 ( e.g. , * ? ? ?
* ; * ? ? ?
so we adopt the cr electron momentum distribution of @xmath102 for @xmath83 with @xmath103 . adopting a higher value of @xmath100 , as suggested in some recent pic simulations @xcite ,
will increase the amplitude of @xmath104 and therefore the synchrotron radiation flux uniformly for all shocks . at the same time of gaining energy via the dsa process at shock ,
cr electrons lose energy via synchrotron emission and ic scattering . as a results ,
an equilibrium momentum exists where the momentum gain is balanced by the radiative losses
@xcite : @xmath105^{1/2},\ ] ] where all the quantities except @xmath106 are expressed in cgs units and the electron equilibrium momentum is normalized as @xmath107 .
here , @xmath108 takes account for the loss due to ic scattering of cosmic background photons ( @xmath109 ) as well as the synchrotron loss .
then , the cr electron spectrum _ at the shock location _ has the form of @xmath110 $ ] .
as the primary cr electrons are advected downstream in the postshock region , they also lose energy with the cooling time @xmath111 .
so the cutoff momentum due to radiative losses decreases with the postshock distance ( @xmath112 ) away from the shock front as @xmath113 . as a result ,
the downstream , integrated electron spectrum steepens by one power of @xmath114 as @xmath115 for @xmath116 , where the ` break ' momentum , @xmath117 is the lowest momentum up to which the postshock cr electrons have cooled down . in our simulation sample
, a shock is defined as discontinuity within a grid zone of thickness @xmath118 @xmath119 .
the advection time for cr electrons to pass such thickness is @xmath120 for @xmath121 , which is of the order of the dynamical time of typical clusters or the merger time .
for the cr electrons emitting at ghz in @xmath75 g magnetic field ( @xmath122 ) , the cooling time becomes shorter than the advection time ( @xmath123 ) , so they have cooled down before existing the shock zone .
hence , we assume that the _ volume - averaged _ spectrum of cr electrons within each shock zone has the following form : @xmath124 & \textrm { for } { \mathinner{\tilde{p } } } > { \mathinner{\tilde{p}}}_\mathrm{br } \\
\end{array } \right .
\label{fep}\ ] ] note that we made several assumptions and simplifications such as the dsa efficiency in equation ( [ fcrp ] ) , the magnetic field strength , @xmath125 , in equation ( [ bdynamo ] ) , the electron to proton ratio , @xmath100 , and the volume - averaged electron spectrum in equation ( [ fep ] ) .
modifications to models for those will lead to rescaling of synchrotron radiation flux .
so these may affect quantitatively some of the shock properties in the flux - limited , mock sample of radio relics , because a radio relic will consist of multiple shocks in the projected maps ( see section 2.4 ) .
but their impacts on the estimates of derived radio mach numbers would be marginal , so the main results of this work should remain mostly unaffected ( see section 2.5 ) .
we calculate the radio synchrotron emission at shock zones , using the magnetic field strength given in equation ( [ bdynamo ] ) and the cr electron spectrum given in equation ( [ fep ] ) . for a single electron with @xmath126 , the synchrotron power at frequency @xmath127
is given by @xmath128 , where @xmath129 is the angle between the electron momentum and magnetic field directions , @xmath130 ( eb \sin \theta ) / ( 4\pi m_e c)$ ] is the characteristic frequency , and @xmath131 ( @xmath132 is modified bessel function ) .
then , the synchrotron emissivity ( in cgs units of erg @xmath133 s@xmath134 hz@xmath134
str@xmath134 ) in shock zone can be estimated as the sum of @xmath135 for all cr electrons of momentum spectrum @xmath104 in the zone , assuming random angular distribution ( i.e. , @xmath136 ) , @xmath137 note that @xmath138 is in fact the volume - averaged emissivity , because @xmath104 in equation ( [ fep ] ) is averaged over the volume of the grid zone taking account for radiative cooling in the postshock region .
the calculation of the bolometric x - ray emissivity due to thermal bremsstrahlung in the icm is straightforward and can be done with @xmath139 where the he abundance is assumed to be 7.9 % by number ( i.e. , the mass fractions of @xmath140 and @xmath141 ) and the gaunt factor @xmath142 for @xmath143k is used . note that @xmath144 is calculated at all grid zones , while @xmath138 only at shock zones .
mock radio and x - ray maps of each synthetic cluster are constructed by projecting the synchrotron emissivity , @xmath145 , and the bolometric x - ray emissivity , @xmath146 , along a depth of @xmath147 .
the choice of the depth should not affect our results , since both the synchrotron and x - ray emissions beyond @xmath148 from the cluster center are in general negligible .
all line - of - sights ( loss ) are assumed to be parallel to each other , since the angular sizes of observed clusters and radio relics are usually much smaller than one radian .
assuming that the system is optically thin , the synchrotron / x - ray intensities , or surface brightnesses , are calculated by integrating the synchrotron / x - ray emissivities along each los , i.e. , @xmath149 and @xmath150 .
since the spatial distribution of shock zones in clusters is quite complicated , they can be translated into radio structures of different morphologies , depending on the projection direction ( e.g. , * ? ? ?
* ; * ? ? ?
so we choose 24 equally - spaced viewing angles for projection , resulting in 24 realizations of radio / x - ray maps for each of 228 synthetic clusters . as a result ,
a total of 5472 radio / x - ray maps are generated and used for the statistics below .
radio telescopes have finite resolutions , so in practice they measure the surface brightness convolved with telescope beams .
if a beam has effective angular area @xmath151 , then the measured synchrotron flux within the beam is approximately @xmath152 , where @xmath153 is the redshifted observation frequency . in this study , we adopt @xmath154 and a beam of @xmath155 for synthetic radio observation .
the beam size is chosen to comparable to those of future radio surveys such as those of the square kilometre array ( ska ) .
hereafter , we set @xmath156 ghz as the representative radio frequency , and present @xmath157 and @xmath158 at the frequency as radio quantities for synthetic observation . for the calculation of radio spectral index ,
two frequencies of 142 mhz and 1.4 ghz are used ( see the next subsection ) . with shocks of different mach numbers and kinetic energy fluxes existing in clusters as mentioned in introduction , the radio / x - ray structures projected to the sky may consist of shock surfaces of different characteristics .
so it may not be straightforward to define the properties for observed structures such as radio relics . in our 3d simulation data cube ,
each ` shock zone ' is specified by the sonic mach number , cr spectrum , and radio emissivity , while each zone is assigned with the gas density , velocity , temperature , magnetic field strength and x - ray emissivity , as described in previous subsections . in observations ( real or synthetic ) , on the other hand ,
the physical properties of shocks must be extracted from 2d projections of radio / x - ray emissivities . here
, we describe the quantification of sonic mach number @xmath11 of shocks associated with 2d projected radio / x - ray structures .
we specify the _ derived _ mach numbers in two different ways .
first , the _ weighted _ mach numbers , @xmath159 and @xmath160 , are defined as the averages along los , weighted by their synchrotron and x - ray emissivities , i.e. , @xmath161 and @xmath162 , respectively .
we consider that these weighted mach numbers represent the true properties of shocks associated with 2d projected structures .
second , the _ observed _ mach numbers , @xmath163 and @xmath164 , on the other hand , are designed to mimic the mach numbers inferred from radio / x - ray observations . in 2d projected map ,
a ` radio pixel ' is the pixel that has at least one shock zone along its los and so has @xmath165 . for each radio pixel
, @xmath166 is calculated from the integrated spectral index of synchrotron flux , i.e. , @xmath167 , obtained between @xmath168 mhz and @xmath169 ghz , following the same way that radio observers usually interpret their observed radio spectra ( see table 1 ) .
note that the value of @xmath166 is not sensitive to the choice of low frequency with our model of cr electron spectrum .
however , this practice is justified only when the break momentum of the volume - integrated electron spectrum is @xmath170 , that is , when the break frequency in the integrated radio spectrum is @xmath171 mhz . for the calculation of @xmath172 ,
the x - ray emission - weighted temperature is calculated along each los as @xmath173 in mock maps of clusters .
pixels are tagged as ` x - ray shocks ' , if @xmath174 , as in the shock identification scheme for 3d volume ( see paper i ) . for x - ray shocks , @xmath172 is estimated from the temperature jump , i.e. @xmath175 . the larger of the values of @xmath172 estimated along the two primary ( @xmath176 and @xmath177 ) directions on projected maps is assigned as the observed mach number , i.e. , @xmath178 .
again , only x - ray shocks with @xmath179 are considered .
note that a pixel identified as x - ray shock may not be a ` radio pixel ' and vice versa .
in fact , only a small fraction ( @xmath180 ) of radio pixels are identified as x - ray shocks .
although the fraction may differ in real observations , this suggests that x - ray observations could miss a substantial fraction of radio shocks .
since it is caused by the smoothing in projection , this problem may be difficult to be overcome , even if the angular resolution of x - ray observation is improved .
in this section , we examine how the projection affects the radio / x - ray observations of shocks such as the morphologies and derived shock parameters . as a representative example , we consider a cluster from @xmath61 box simulation with @xmath56 grid zones .
this cluster is identical to the one appeared in the left panels of figures 1 and also in figure 7 of paper i. it has the x - ray emission - weighted temperature @xmath181 kev and @xmath182 .
the shock that produces the largest amount of cr protons in this cluster is an infall shock with @xmath183 and @xmath184 ( see paper i ) . figure [ fig : map ] shows projected maps of the cluster , viewed from four different angles , zoomed around the area of @xmath185 ( dashed circle ) , where @xmath186 is the distance from the cluster center in the projected plane . here , the synchrotron flux , @xmath158 , is shown as contours of black solid lines , which is superposed with the bolometric x - ray surface brightness , @xmath187 , in gray - scale .
the lowest contour level for the synchrotron flux is @xmath188 , which could be detected by future radio observatories such as the ska .
the projection direction of each map is given in terms of polar and azimuthal angles , @xmath129 and @xmath80 , respectively .
the four maps in figure [ fig : map ] exhibit morphologies of radio structures , distinct from each other .
for instance , the radio map in the upper - left panel seems to show a paired structures on the opposite side of the cluster , which could be interpreted as radio relics due to a pair of merger shocks . here
, we tag the radio pixels associated with the left structure as `` l - relic '' ( red color ) , while those associated with the right structure as `` r - relic '' ( green color ) .
in other three panels , if the shock zone with the largest synchrotron contribution along a given los belong to either l - relic or r - relic , we color the corresponding radio pixel as red or green , respectively . in the upper - right and lower - left panels ,
elongated radio structures with length @xmath189 are composed of both red ( l - relic ) and green ( r - relic ) pixels .
this illustrates that even a single connected radio structure in the sky may consist of a number of disconnected shock surfaces in the real 3d volume .
so it could be misleading , if we try to extract the nature of underlying shocks only from the morphology or shock parameters inferred from radio observations .
the radio map in the lower - right panel , on the other hand , has a number of small structures , and the structure located near the center could be interpreted as so - called `` radio mini - halo '' ( see , e.g. , * ? ? ?
* ; * ? ? ?
* for reviews ) . here
, we do not intend to address the natures of observed paired structures or halo - like structures in detail , but we just point that the morphology of observed radio structures critically depends on the projection of underlying 3d structures . in figure
[ fig : slice ] , we compare the spatial distributions of @xmath67 , @xmath190 , and @xmath11 in a slice ( left panels ) with those of @xmath191 , @xmath166 , and @xmath172 in 2d projection ( right panels ) .
the slice passes through @xmath192 away from the cluster center , and contains the most radio - luminous shock zones for the l and r - relics , which are projected at @xmath193 @xmath194 and @xmath195 @xmath196 in the 2d maps , respectively .
they are identified as infall shocks with @xmath197 for l - relic and @xmath198 for r - relic .
the right panels are from the same map as that for the upper - left panel of figure [ fig : map ] , and the contours are drawn with the level of @xmath199 . for @xmath166 , only pixels with @xmath200 are plotted , while for @xmath191 and @xmath172 , only pixels with @xmath201 are plotted .
note that the bolometric x - ray surface brightness in outskirts near @xmath202 ranges typically @xmath203 in observed clusters ( e.g. , * ? ? ?
comparison of @xmath11 in the slice map ( left - bottom ) and @xmath166 in the 2d map ( right - middle ) indicates that the mach number and location of shocks with @xmath204 agree well in the two distributions , especially for the shocks associated with l - relic .
this is because the radio spectrum in projected maps is governed by one or a few radio - luminous shock zones with high mach numbers in a given los .
so @xmath166 derived from radio observations could be a good proxy for the mach number of radio - luminous shocks , implying that the properties of such shocks could be extracted reasonably well from radio observations . on the other hand ,
there are noticeable differences between the distributions of @xmath67/@xmath11 in the slice maps ( left - top / bottom ) and those of @xmath191/@xmath172 in the 2d maps ( right - top / bottom ) . as seen in top panels , the distribution of @xmath191 is smoother than that of @xmath67 , partly because the contribution from the x - ray bright icm dominates over the emissions from the whim in surrounding filaments , and also because the projection ( integration along loss ) inevitably irons out any sharp features .
the temperature jump across a shock in the @xmath191 map looks reduced .
we find that @xmath172 derived from x - ray observations tends to underestimate the actual shock mach number ( see the next chapter ) .
we further examine in detail the region near l - relic , where the slice maps indicate a shock of @xmath205 formed by the infall of the whim along a filament .
there is @xmath206 of positional shift between the locations of shocks in the @xmath166 ( right - middle ) and @xmath172 ( right - bottom ) maps .
this is because @xmath166 picks up the infall shock , while due to the dominant contribution of x - ray from the hot and dense icm along the los , @xmath172 picks up a foreground shock with smaller mach number formed in the icm ( sse below ) .
this may explain spatial offsets between the shock surfaces inferred from radio and x - ray observations in some radio relics ( e.g. , * ? ? ?
* ; * ? ? ?
figure [ fig:3d ] shows the distributions of radio and x - ray emissivities , @xmath157 and @xmath144 , as functions of @xmath11 for shocks within @xmath185 in our simulated clusters .
both distributions have rather large spreads , but the following points are clear , firstly , @xmath144 decreases with increasing @xmath11 , while @xmath157 increases with @xmath11 , peaks at @xmath207 and decreases slightly for higher @xmath11 .
these behaviors can be understood as follows .
weaker shocks which tend to form in hot and dense regions near the cluster core produce more @xmath144 , while shocks formed in cluster outskirts with @xmath208 several accelerate crs and produce @xmath157 most efficiently .
secondly , on average , @xmath144 varies over two orders of magnitude , while @xmath157 varies over 10 orders of magnitude .
although both radio / x - ray surface brightnesses on projected maps are dominated by contributions from a small number of bright shock zones along each los , the tendency is much stronger in the case of radio .
in this section , we first examine the surface brightnesses and derived mach numbers of pixels with shocks , i.e @xmath209 , @xmath210 , @xmath211 and @xmath212 , obtained in projected radio / x - ray maps .
as noted in section 2.5 , radio pixels which contain shocks in radio maps may not be identified as shocks in x - ray maps , and vice versa .
so when we study a correlation between any two quantities , for instance , @xmath166 versus @xmath172 , we use a subset of pixels in which both quantities are specified .
figure [ fig : xlum_sync_mach_all ] displays the relations between @xmath209 vs @xmath211 , @xmath166 vs @xmath209 , and @xmath172 vs @xmath211 for pixels within @xmath185 from the cluster center .
the bottom panel shows that bright x - ray shocks with large @xmath211 tend to be weak with small @xmath172 ; the brightest x - ray shocks have @xmath213 .
the middle panel , on the other hand , shows that the synchrotron flux @xmath209 of radio pixels increases with @xmath166 up to @xmath214 and then decreases for larger @xmath166 .
these behaviors should be guessed from those of @xmath157 and @xmath144 shown in figure [ fig:3d ] . as a result ,
the correlation between @xmath209 and @xmath187 turns out to be rather poor with wide variations in the top panel .
what is clear is that the shocks which are brightest in radio are not necessarily brightest in x - ray . as a matter of fact ,
brightest x - ray shocks have modest radio brightnesses , and vice verse .
figure [ fig : machs_all ] displays the relations among mock - observed mach numbers , @xmath166 and @xmath172 , and weighted mach numbers , @xmath215 and @xmath216 , for pixels within @xmath185 . as noted in the previous chapter ,
the x - ray mach numbers tend to be smaller than the radio mach numbers ( upper panels ) .
again , this is mainly because x - ray observations incline to pick up weaker shocks than radio observations along given loss .
the correlation between @xmath210 and @xmath212 is rather poor ; the pearson s linear correlation coefficient of @xmath166 and @xmath172 is @xmath217 ( upper - left panel ) and that of @xmath215 and @xmath216 is @xmath218 ( upper - right panel ) .
so it could be misleading if @xmath172 is guessed from @xmath166 , and vice versa .
it is expected that @xmath219 and @xmath220 for the same band are correlated better .
we find that @xmath166 is often smaller than @xmath215 as shown in the lower - left panel , but the correlation between them is quite good with @xmath221 .
the correlation between @xmath172 and @xmath216 in the lower - right panel , on the other hand , is still poor with @xmath222 .
this demonstrates that the smoothing in projection would make the estimation of shock properties harder in x - ray observations than in radio observations . in figure
[ fig : rad_all ] , we examine the radial distributions of observable properties of shock pixels .
both @xmath166 and @xmath172 tend to increase with radius , and @xmath166 shows a larger variance than @xmath172 .
shocks with x - ray mach number @xmath223 a few would be rare within @xmath185 , while shocks with radio mach number @xmath166 up to several could be found .
radio bright shocks , for instance , those with @xmath224 , are found mostly in @xmath225 .
this suggests that future radio observations could detect many more shock structures in cluster outskirts . on the other hand , x - ray bright shocks
are preferentially located and so found close the center .
the bolometric x - ray surface brightness of x - ray shocks follows @xmath226 with @xmath227 for @xmath228 , which is consistent with the observed x - ray profiles in cluster outskirts ( e.g. , * ? ? ?
we build up a catalog of `` synthetic radio relics '' by finding connected structures in our radio maps , which meet the following conditions .
( 1 ) all pixels within structures have @xmath229 .
( 2 ) a structure has at least five pixels .
note that the area of a single pixel is @xmath230 ( @xmath231 or @xmath232 ) .
so five pixels corresponds to , for instance , @xmath233 in simulations of @xmath234 box with @xmath54 grid zones . using these conditions , we have 9583 radio relics samples : 4626 from simulations of @xmath234 with @xmath54 grid zones , 4850 from simulations of @xmath235 with @xmath54 grid zones , and 107 from simulation of @xmath234 with @xmath56 grid zones .
we then assign the following quantities to synthetic radio relics .
the average distance from the cluster center , @xmath236 , is defined by the synchrotron - weighted average of the distance of the pixels that belong to a radio relic , @xmath237 , where @xmath186 is the distance of pixels from the cluster center .
the derived mach numbers of radio relics are defined by the synchrotron / x - ray weighted averages of the corresponding derived mach numbers of pixels , @xmath166 , @xmath215 , @xmath172 , and @xmath216 .
for example , @xmath238 and @xmath239 . the number of pixels with @xmath172 could be smaller than that with @xmath166 , since @xmath172 is specified only at the pixels identified as x - ray shocks .
in fact , about 40% of sample radio relics contain no pixel with @xmath172 , so the corresponding x - ray mach number is not assigned to them .
the synchrotron power at 1.4 ghz and bolometric x - ray luminosity of a radio relic are calculated with @xmath240 and @xmath241 , respectively .
figure [ fig : machs_relic ] displays the relations among the derived mach numbers for sample radio relics .
the relations follow and so look similar to those for pixels in projected maps shown in figure [ fig : machs_all ] , but the correlations are tighter with @xmath242 @xmath243 @xmath244 , @xmath245 @xmath246 @xmath247 , @xmath242 @xmath248 @xmath249 , and @xmath250 @xmath251)@xmath252 . especially , the correlation between @xmath253 and @xmath254 is very good , indicating that the mach number estimated from radio spectral index would be a fair representation of suitably averaged mach number of shocks associated with radio relics .
most of our sample radio relics have @xmath255 .
this is partly because with the model cr acceleration efficiency we employ @xcite , the amount of cr electrons emitting synchrotron is very small for weak shocks with @xmath256 ( see figure 3 of paper i ) .
if the cr electron acceleration at weak shocks would be more efficient than in our model , radio relics with smaller @xmath253 could be more common .
as noted in introduction , merger shocks are expected to have mostly @xmath7 .
in fact , we find that our synthetic radio relics normally involve projections of multiple shocks along loss , resulting in different morphologies for different viewing angles ( see figure [ fig : map ] ) . in @xmath257 of sample radio relics , the brightest pixels with largest @xmath209 include infall shocks along loss , suggesting that infall shocks may account for some radio relics with flat spectra .
this seems reasonable in the sense that infall shocks could be the major sources of crs in clusters , as discussed in paper i. figure [ fig : rad_relic ] displays the radial distributions of @xmath253 , @xmath258 , @xmath259 , and @xmath260 .
upper panels show that most of our synthetic radio relics are found at @xmath261 .
although there are wide variations , @xmath253 and @xmath258 tend to increase toward outskirts , and on average the radio mach number is larger than the x - ray mach number .
the lower - left panel shows that the radio power at a given radial distance can vary widely , yet most powerful radio relics are found at @xmath262 .
this results mainly from the combined effects of the radial distribution of shock kinetic energy flux and @xmath11 , the strong dependence of cr acceleration efficiency on @xmath11 , and the geometrical increase of relic surface area toward outskirts ( see also * ? ? ? * ) .
the lower - right panel shows a strong radial decline of the bolometric x - ray luminosity of sample radio relics ( not including the contributions from the background icm ) toward outskirts .
this reflects mostly the radial decreases of gas density and temperature ; the effect of the increase of the mach number and relic surface area is not large here .
we compare our synthetic radio relics with observed radio relics .
table 1 lists the properties of the observed radio relics we use ; here , @xmath263 is the redshift of associated clusters , @xmath264 is the largest linear scale on the sky , and @xmath265 is the integrated radio spectral index . they are chosen from @xcite and @xcite with the following criteria ; ( 1 ) @xmath266 ( our radio relic sample is constructed at @xmath154 ) , ( 2 ) @xmath267 with @xmath268 , and ( 3 ) both @xmath259 and @xmath265 available .
note that @xmath265 in table 1 was calculated between @xmath269 mhz and @xmath270 ghz .
although some observations suggested that the radio spectra of radio relics could have curvatures and @xmath265 may increase at high frequencies @xmath271 ghz ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , here we ignore the effect .
figure [ fig : relic ] shows @xmath259 vs @xmath236 and @xmath265 for synthetic ( colors ) and observed ( filled squares ) radio relics .
the distribution of @xmath259 of observed radio relics seems to fall within the range predicted by our synthetic observation .
on the other hand , there are noticeable differences in the distributions of @xmath236 and @xmath265 between synthetic and observed radio relics .
first , four observed radio relics are found with @xmath272 mpc among twelve , while the fraction of synthetic radio relics with @xmath272 mpc is very small , less than @xmath273 .
second , while the integrated spectral index of synthetic radio relics lies @xmath274 ( corresponding to @xmath275 ) , some of observed radio relics have steeper spectral indices . for example , the radio relics found at abell 548b b and abell s753 have @xmath276 ( @xmath277 ) , and the radio relic at abell 2345w has @xmath278 ( @xmath279 ) . the discrepancies could be partly because weak shocks close to cluster core might be under - represented in our structure formation simulations due to the limited spatial resolution and omission of some non - gravitational processes .
but it could be also possible that weak shocks accelerate cr electrons more efficiently than we assume here , as hinted in recent pic simulations ( see , e.g. * ? ? ?
* ; * ? ? ?
* ) . on the other hand , the re - acceleration of `` pre - existing '' cr electrons , which is not considered in this study , could play an important role in the production of cr electrons at weak shocks with @xmath0 ( e.g. * ? ? ? * ; * ? ? ?
especially , if fossil relativistic electrons are present in the form of isolated clouds , instead of being spread throughout the general icm space , the relative rareness of radio relics can be understood , as discussed in the introduction @xcite .
in such a scenario , ` lighting - up ' of radio relics is governed mainly by the occasional encounters of icm shocks with fossil electron clouds , instead of the mach - number - dependent dsa efficiency and the shock kinetic energy flux . as a result
, the frequency and radio luminosity of radio relics may depend less sensitively on the shock mach number , and much weaker shocks could be turned on as radio relics with steeper spectral index .
these features are more consistent with the observational data in figure 9 .
moreover , only a small fraction of icm shocks become radio sources for the period much shorter than the cluster dynamical time sale @xcite .
so the model naturally explains why radio relics are rare , while structure formation simulations predict shocks should be ubiquitous in the icm ( e.g. * ? ? ?
* ; * ? ? ?
the exploration of this scenario requires extensions of the models we employ here , so beyond the scope of this paper .
the existence of shocks in clusters of galaxies has been established through observations of radio relics due to synchrotron emission from shock - accelerated cr electrons ( e.g. , * ? ? ?
* ) as well as through x - ray observations of shock discontinuities ( e.g. , * ? ? ?
* ) . however , it is not clear whether the properties of shocks inferred from radio and x - ray observations represent their true nature , because we can only measure the quantities projected onto the sky , integrated along loss .
in fact , for a few relics such as the tooth brush relic and the radio relic in a2256 , radio and x - ray observations have reported inconsistencies in quantities such as shock mach number and position @xcite . in this paper
, we explored a scenario in which electrons freshly injected and accelerated to high energies via dsa process at the structure formation shocks produce radio relics .
we constructed synthetic maps of galaxy clusters in radio and x - ray by employing the cosmological hydrodynamic simulation data reported earlier in paper i. first , the volume - averaged synchrotron emissivity at shock zones was calculated by adopting the cr electron acceleration efficiency based on a dsa model @xcite and magnetic fields based on a turbulent dynamo model @xcite .
the bolometric x - ray emissivity at grid zones was also calculated using the gas density and temperature from simulations .
then , mock maps of the synchrotron and x - ray surface brightnesses in 2d projection were produced by integrating volume emissivities along loss for simulated clusters . in the synthetic maps ,
the properties of shocks and radio relics , such as the shock mach number and location , were examined in detail .
\1 ) in most cases , radio and x - ray shocks in 2d maps are the outcomes of projection of multiple shock surfaces along loss , because shocks are abundant in the icm with a mean separation of shock surfaces , @xmath280 . as a result ,
the morphology of shock distributions in 2d maps depends on the projection direction ; for the same cluster , very different morphologies may turn out for different viewing angles .
\2 ) synchrotron emissivity depends sensitively on the shock mach number , especially with our dsa model for the cr electron acceleration , while bremsstrahlung emissivity depends on the gas density and temperature .
hence , radio observations tend to pick up shocks of @xmath208 a few to several along a los , while x - ray observations preferentially select weaker shocks ( @xmath0 ) with high density and temperature .
consequently , the properties of shocks in 2d projected maps could be different in radio and x - ray observations .
the shock mach number estimated with x - ray observations tends to be smaller than that from radio observations , if a radio relic consists of multiple shocks along loss .
in addition , the location of shocks from x - ray observations could be shifted with respect to that from radio observations .
\3 ) for radio relics , the mach number estimated from the radio spectral index seems to be a fair representation of suitably averaged mach number of shock surfaces associated with them . on the other hand , the discontinuities in x - ray temperature tend to be smeared due to projection effects including possible multiple shocks and multiple icm components , resulting in the x - ray mach number lower than the real mach number of the associated shock .
\4 ) when the properties of our synthetic radio relics are compared with those of observed radio relics , there are clear differences in the statistics of radial location and spectral index ; more radio relics have been observed closer to the cluster core and with steeper spectral indices than our synthetic observation predicts .
this discrepancy may imply that weak shocks in high beta plasmas in fact accelerate cr electrons more efficiently than we model here ( see , e.g. * ? ? ?
* ; * ? ? ?
alternatively , as shown in the previous studies of @xcite and @xcite , the re - acceleration of pre - existing electrons in the icm may enhance the production of cr electrons at weak shocks with @xmath0 .
finally , we could also conjecture that radio relics might be activated only when shocks encounter clouds , which contain fossil electrons either accelerated earlier by shocks / turbulence or left over from old radio jets ( see , e.g. * ? ? ?
* ; * ? ? ?
this model may explain how very weak shocks can turn on the synchrotron emission and why radio relics are rare relative to putative shocks in galaxy clusters .
the authors thank the anonymous referee for his / her thorough review and constructive suggestions that lead to an improvement of the paper .
seh was supported by the national research foundation of korea through grant 2007 - 0093860 .
hk was supported by basic science research program through the national research foundation of korea(nrf ) funded by the ministry of education ( 2014r1a1a2057940 ) .
the work of dr was supported by the year of 2014 research fund of unist ( 1.140035.01 ) .
lcccccl abell 2256 & 0.058 & 3.95 & 300 & 1100 & 1.2 & @xcite + abell
1240 n & 0.159 & 0.427 & 700 & 650 & 1.2 & @xcite + abell 1240 s & 0.159 & 0.730 & 1100 & 1250 & 1.3 & @xcite + abell
2345 e & 0.177 & 2.62 & 890 & 1500 & 1.3 & @xcite + abell 2345 w & 0.177 & 2.83 & 1000 & 1150 & 1.5 & @xcite + abell 3667 & 0.056 & 17.4 & 1950 & 1920 & 1.1 & @xcite + abell
548b b & 0.042 & 0.250 & 430 & 370 & @xmath281 & @xcite + coma 1253 + 275 & 0.023 & 0.284 & 1940 & 850 & 1.18 & @xcite + abell 2163 & 0.203 & 2.23 & 1550 & 450 & 1.02 & @xcite + abell s753 & 0.014 & 0.205 & 410 & 350 & 2.0 & @xcite + abell 115 & 0.197 & 16.7 & 1510 & 1960 & 1.1 & @xcite + abell 610 & 0.095 & 0.444 & 310 & 330 & 1.4 & @xcite + ( top ) , gas density @xmath190 ( middle ) , and shock mach number @xmath11 ( bottom ) in a slice which passes through @xmath282 away from the center of the same cluster as in figure [ fig : map ] .
right panels : projected maps of x - ray emission - weighted temperature @xmath191 ( top ) , mach number estimated from the radio spectral index , @xmath166 ( middle ) , and mach number of x - ray shocks detected in the @xmath191 map , @xmath172 ( bottom ) , with the same viewing angle as the upper - left panel of figure [ fig : map ] .
the yellow ( top panels ) and black ( middle and bottom panels ) contours correspond to @xmath283 . for @xmath191 and @xmath172 ,
the pixels with @xmath284 are shown , while for @xmath166 , the pixels with @xmath285 are shown .
dashed circles show @xmath148 . ]
vs @xmath211 , @xmath166 vs @xmath209 , and @xmath172 vs @xmath211 for shock pixels within @xmath185 from the cluster center in projected maps . here
, @xmath209 is the synchrotron flux at 1.4 ghz , @xmath211 is the bolometric x - ray surface brightness , @xmath166 is the mach number estimated from the radio spectral index , and @xmath172 is the mach number of x - ray shocks detected in the @xmath191 map .
colors code the relative frequency , ranging from @xmath286 ( blue ) to 1 ( red ) in the top and middle panels and from @xmath287 ( blue ) to 1 ( red ) in the bottom panel . ]
|
radio relics detected in the outskirts of galaxy clusters are thought to trace radio - emitting relativistic electrons accelerated at cosmological shocks . in this study , using the cosmological hydrodynamic simulation data for the large - scale structure formation and adopting a diffusive shock acceleration ( dsa ) model for the production of cosmic - ray ( cr ) electrons , we construct mock radio and x - ray maps of simulated galaxy clusters that are projected in the sky plane .
various properties of shocks and radio relics , including the shock mach number , radio spectral index and luminosity are extracted from the synthetic maps and compared with observations .
a substantial fraction of radio and x - ray shocks identified in these maps involve multiple shock surfaces along line of sights ( loss ) , and the morphology of shock distributions in the maps depends on the projection direction . among multiple shocks in a given los ,
radio observations tend to pick up stronger shocks with flatter radio spectra , while x - ray observations preferentially select weaker shocks with larger kinetic energy flux . as a result
, the shock mach numbers and locations derived from radio and x - ray observations could differ from each other in some cases .
we also find that the distributions of the spectral index and radio power of the synthetic radio relics are somewhat inconsistent with those of observed real relics ; a bit more radio relics have been observed closer to the cluster core and with steeper spectral indices .
we suggest the inconsistency could be explained , if very weak shocks with @xmath0 accelerate cr electrons more efficiently , compared to the dsa model adopted here .
|
the starting up of the large hadron collider ( lhc ) operating in an energy range far above the electroweak scale , offers a possibility to reveal new phenomena and to explore the phenomenology extracted from the expected huge amount of experimental data . in this way
, it is expected new physics manifestations throughout the appearance of new degrees of freedom such as new charged and neutral fermions , superpartners , new gauge bosons , and higgs scalar(s ) . to take into account these new possible degrees of freedom
we must go beyond the sm .
there are many ways to construct extensions of the electroweak sm .
however , if we are concerned with new gauge bosons we must consider a larger gauge group .
whatever the extension of the sm is , it must present the same results of the sm at the electroweak energy scale in order to be consistent with present data .
although hadron colliders present a high background ( pile - up ) compared to electron - positron ( beamstrahlung ) colliders , they seem to be more suitable for discovering new degrees of freedom . since high centre - of - mass energy and high precision
can not be achieved simultaneously in the same experiment , both types of colliders are needed in order to have a detailed description of a particular phenomena .
this is easily realized if we look to a recent past , when the synergy between hadron and lepton colliders , brought up complementary information for each other , generating benefits to high energy physics . as examples we have the discovery of the @xmath9 boson , the gluon , and the quark top .
the discovery of the @xmath9 boson occurred in a @xmath10 collider ( sps ) but its properties ( decay widths , couplings to fermions , asymmetries , mass , etc ... ) were measured with high precision in lepton colliders ( slc , lep ) . in this way , the recent studies started at lhc can direct the search for new physics in future linear colliders ( ilc / clic ) . among the main goals of any accelerator proposal we find the physics of an extra neutral gauge boson @xmath11 .
this new boson is predicted by many extensions of the standard model like : * the models from @xmath12 group , also know as rank 5 models ( er5 m ) @xcite ; * the left - right which are based in the group @xmath13 @xcite ; * the models with @xmath11 in little higgs scenario @xcite ; * the sequential standard models - ssm @xcite ; * the kaluza klein ( kk ) models , predicted by extra dimensions @xcite ; * the models with @xmath14 gauge symmetry , known as 3 - 3 - 1 models @xcite ; * models where a strong dynamics is involved in the electroweak spontaneous symmetry breaking .
we can cite , for instance , topcolor and bess ( breaking electroweak symmetry strongly ) models @xcite .
some models which have an extra @xmath15 gauge group @xcite also deserve a detailed study in order to predict new physics behavior .
the cumulative study of different models , candidates to describe the electroweak interaction , can guide the lhc data analysis if in fact a @xmath11 is discovered .
the extra @xmath15 group gives rise to the @xmath11 vector boson .
the new constraints on these models generate few free parameters , if we compare them to other models which have origin in larger gauge groups . following the arguments here exposed ,
first we have considered the @xmath11 boson observables for electron - positron colliders @xcite , and now we extend the study for hadron colliders . in this paper , we study the phenomenology of the @xmath16 boson originated from two different extensions of the electroweak sm with an extra @xmath15 gauge factor .
we predict the possible signals of new physics in the process @xmath17 , at the tree level .
we compare several quantities as total cross sections , number of events , and decay widths assuming that both @xmath15 factors correspond to the local @xmath2 symmetry , and that the breaking of this symmetry occurs at an energy scale above the tev scale . in order to have a more realistic scenario
, we choose the parameters and constraints of the two different models to compare them at the same @xmath0 boson mass .
we use @xmath18 gev , but we also show the results for @xmath19 gev , for two centre - of - mass energies : @xmath20 and @xmath21 tev . this article is organized in the following way . in section
ii we introduce the models . in section
iii we present the results and in section iv we present our conclusions .
in this paper we consider two extensions of the electroweak sm which have an extra @xmath15 local factor , resulting in the total gauge symmetry @xmath22 .
more specifically , we are concerned with models which are based on the gauge symmetries : * @xmath23 , called as secluded model @xcite ; * @xmath24 called as flipped model @xcite . in the secluded model
, @xmath25 is a new @xmath15 charge , and @xmath26 is the electroweak sm hypercharge . in the flipped one
, @xmath27 is a new extra @xmath15 charge , and the @xmath2 charge assignments are those of the sm . in this last model the sm hypercharge is recovered after the first spontaneous symmetry breaking . in the secluded model , the electric charge has no component in the @xmath28 and in particular , depending on the value of this charge , several versions of this model can be studied .
our analysis will be focused on comparing both models only when the @xmath25 charge of the secluded model coincides with @xmath2 . besides ,
in this minimal version , considered here , both models have the same scalar sector , composed by one doublet , @xmath29 , with hypercharge @xmath30 , and a complex singlet with @xmath31 .
the number of right - handed neutrinos is three , in order to be consistent with the anomaly cancellations .
we had set them to be heavy enough in order to have @xmath11 decay into heavy right - handed neutrinos kinematically suppressed .
besides that , the minimal version considered here is also absent of exotic decay channels like heavy quarks or heavy charged leptons .
so , our results of decay widths , branchings fractions for these two particular @xmath11 models are related only to standard model fermions .
we will parameterize the neutral currents in terms of the mass eigenstate fields as follows : @xmath32\psi_i .
\label{nc}\ ] ] in this paper we briefly present both models .
the complete details about both of them were subject of study in a recent work @xcite . moreover , the scalar sector of this sort of models , with the respective feynman rules , is considered in reference @xcite .
the charge operators for the secluded and the flipped models , are respectively given by : @xmath33 @xmath34 .
\label{gn1}\end{aligned}\ ] ] when the @xmath25 charge of the secluded model is equal to @xmath2 it implies that the parameter @xmath35 ( the @xmath25-charge of the sm higgs doublet ) , discussed in refs .
@xcite , is equal to zero . in this case
, there is no tree level mixing in the mass matrix between the neutral gauge bosons @xmath36 , @xmath37 and @xmath38 , and we have @xmath39 and @xmath40 , i.e. the mass eigenstates and the symmetry eigenstates coincide .
this is not the case for the flipped model . here
, we will assume that the mixing between the two neutral vector bosons is small enough so that @xmath41 and @xmath42 .
both models are well motivated since , in general , they can be thought as an intermediate symmetry , remnant of a larger ( unknown ) gauge symmetry , that was broken at a very large energy scale .
moreover , they are simpler compared to other class of models with an extra @xmath0 neutral vector boson : simpler gauge groups and fewer free parameters .
we have first chosen @xmath18 gev in order to compare the results coming from the two different models . for the secluded model ,
this choice respects the bounds already established on the @xmath11 boson mass and its coupling constants by the direct search at the tevatron @xcite , by the electroweak precision tests ( ewpt ) at lep ii , and by the low - energy neutral current experiments @xcite , which are well studied in the literature . on the other hand ,
the flipped model has not received the same level of attention .
we find that it also deserves a detailed study to establish constraints on its parameters and numerical couplings . for the flipped model we use the following inputs : @xmath43 , @xmath44 , @xmath45 gev , @xmath46 gev , @xmath47 , and @xmath48 @xcite . with these parameters
we obtain @xmath49 gev , @xmath50 gev , and the values of the neutral current coupling constants shown in table 1 .
the @xmath9 couplings with sm fermions remain the same so we emphasize only the new contribution from the @xmath11 boson .
[ acoplflipp ] @xmath51 for the secluded model we use as inputs : @xmath52 , @xmath53 , @xmath54 , @xmath55 gev , and @xmath56 gev . with these parameters
we obtain @xmath57 gev , @xmath58 gev , and the values of the neutral current coupling constants shown in table 2 . due to the choice @xmath52 in the secluded model
, the @xmath0 boson has only vector couplings to fermions , and presents a _ leptofilic _ behavior around the peak region .
the _ leptofilic _
character means that the @xmath0 boson couples preferentially to leptons , and that the cross section for lepton production presents a higher peak compared to quark cross sections .
this behavior can be seen with more evidence in @xmath59 process , where @xmath60 is a lepton .
[ acoplsecl ] @xmath61 we can realize through the partial decay widths that , as we said before , for the secluded model , @xmath11 couples preferentially to leptons , as shown in table 3 . due to this character , and depending on the choice of the @xmath0 parameters , the secluded model can give an explanation for the positron excess in cosmic rays , as presented by the pamela experiment @xcite .
[ branching ] .the @xmath11 partial decay widths for fermions and charged vector boson for both @xmath2 models . [ cols="^,^,^",options="header " , ] besides the total cross section and decay width for leptonic final state , we present : invariant mass , rapidity , forward - backward asymmetry , transverse momentum and angular distribution . in figures 2a and 2b we present the dilepton invariant mass distribution .
the flipped model presents a distinct behavior with a higher peak far above the standard model and has very good changes to be separated from the background . on the other , the secluded model distribution in the region off
peak in figure 2 ( down panel ) has a behavior similar to that of the sm , while the on peak region is not so powerful to distinguish it from the sm .
this difference has origin mainly in the choice of @xmath62 for the secluded model , which leads to pure vector couplings of the @xmath63 boson to fermions .
a different choice for @xmath64 , say @xmath65 , will restore the axial couplings but , in this case , the @xmath25-charge will not be equal to @xmath2 .
the case of @xmath65 is out of the scope of the present work . the dilepton invariant mass distribution for @xmath66 process for the flipped , secluded and standard models considering @xmath18 gev at @xmath67 tev .
first cut set ( up ) and second cut set ( down).,title="fig : " ] the dilepton invariant mass distribution for @xmath66 process for the flipped , secluded and standard models considering @xmath18 gev at @xmath67 tev .
first cut set ( up ) and second cut set ( down).,title="fig : " ] the rapidity distributions can also be used to disentangle the @xmath2 models .
this observable is powerful even when we consider both light and heavy quarks or only the light ones , this result is showed in figures 3a , 3b and 3c .
the signal has been improved for both models but once again the flipped model produces a signal above the ones of the other two models .
one intrinsic characteristic of the flipped model is the increasing signal until it reaches the region where the rapidity of each muon is @xmath68 .
both , the secluded model and the sm do not have this behavior .
when we apply more severe cuts on rapidity , as @xmath69 , the resulting data reveal that the three models have less possibilities to be disentangled .
so , for this observable , the first cut set is more efficient than the more severe ones . considering @xmath18 gev for the flipped , secluded and standard models at @xmath70 tev .
the valence and sea quark contributions ( up ) and only the valence quarks ( middle ) , using the first cut set .
the same rapidity distribution using the second cut set ( down).,title="fig : " ] -0.8 cm considering @xmath18 gev for the flipped , secluded and standard models at @xmath70 tev .
the valence and sea quark contributions ( up ) and only the valence quarks ( middle ) , using the first cut set . the same rapidity distribution using the second cut set ( down).,title="fig : " ] -0.8 cm considering @xmath18 gev for the flipped , secluded and standard models at @xmath70 tev . the valence and sea quark contributions ( up ) and only the valence quarks ( middle ) , using the first cut set . the same rapidity distribution using the second cut set ( down).,title="fig : " ] -1.5 cm another important observable , very sensitive to the new physics contributions , is the forward - backward asymmetry ( @xmath71 ) . in @xmath72 colliders ,
@xmath71 is measured with high precision , due to the well known initial beam direction .
however , the situation is quite different for hadron colliders where the original quark direction is completely unknown . to solve this problem ,
the quark direction can be approximated by the boost direction , which connects the dimuon reference system with the original beam direction . in order to achieve this goal we have applied the @xmath73 cut for the muon pair rapidity , following the references @xcite .
in contrast with the flipped model , the secluded one has no @xmath74 axial couplings to fermions and , as a consequence of it , the forward - backward asymmetry , even being a powerful tool to get information on the @xmath11 boson parameters , does not receive contributions from the axial couplings on - peak . in the figure 4 we show this observable versus @xmath75 , and we see that the two @xmath2 models can be distinguished . as a function of the dilepton invariant mass for both @xmath2 models , considering @xmath18 gev , @xmath70 tev , and applying the second cut set . ]
if we consider all the studied observables , after applying the second cut set , we will realize that one particular numerical cut can put in evidence one observable and others not .
the cut @xmath76 is not so helpful for rapidity distributions , but it is necessary for forward - backward asymmetries in order to guarantee that the quark direction has more probability to be the correct one .
it means that the probability of obtaining more significant measurements of forward - backward asymmetries increases for higher rapidities .
we present in the figure 5 our results for the muon transverse momentum distribution .
one of the characteristics of this distribution is that the peak is located right at the position @xmath77 , for both @xmath2 models .
we have adopted more restricted cuts in the final dimuon mass , @xmath78 gev @xcite in order to emphasize this behavior .
the flipped has once again , much better chances to be separated from the background , in opposition , the secluded one can be mistaken as standard model background . for
the flipped , secluded and standard models , considering @xmath18 gev , @xmath70 tev , and applying the second cut set . ]
the muon angular distribution is less sensitive to @xmath63 contributions , but can be used as a previous result before the calculation of @xmath71 . from figure 6
we can realize that the signals related to the sm are below the signals of the two @xmath2 models .
we can see a small asymmetry when the curves cross each other in the region @xmath79 , where @xmath80 is the angle between the beam and the muon directions . for
the flipped , secluded and standard models , considering @xmath19 gev and @xmath70 tev . ]
we now consider a @xmath63 boson which mass is 1000 gev and @xmath20 and @xmath21 tev in order to extract more conclusions about the same observables presented above .
the related graphs are presented in the figures 7 , 8 , 9 and 10 .
the increase of the centre of mass energy from 7 to 14 tev has as direct consequences the increase of the cross sections and , hence , the number of events .
the invariant mass distribution shows also a small peak for the secluded model if compared to the flipped one , as we can see in figures 7a and 7b .
the rapidity distributions presented in figures 8a and 8b show that for @xmath19 gev both models can not be clearly distinguished from each other in this case .
if we consider @xmath81 tev the rapidity distributions graphs are separated , mainly in the region @xmath82 , but this separation is not so noticeable as the one for @xmath49 gev , already presented in figures 3a and 3b .
the transverse momentum distributions , presented in the figures 9a and 9b , do not show a clear peak for @xmath83 for the secluded model , but the flipped one is put in evidence once again with higher values for this distribution on peak .
the muon angular distributions present a better behavior for disentangling both models , mainly if we consider @xmath81 tev , as can be seen from figures 10a and 10b .
process in flipped and secluded models considering @xmath84 gev at @xmath85 tev ( up ) and @xmath70 tev ( down ) with the first cuts set.,title="fig : " ] process in flipped and secluded models considering @xmath84 gev at @xmath85 tev ( up ) and @xmath70 tev ( down ) with the first cuts set.,title="fig : " ] for both @xmath2 models , considering @xmath19 gev , @xmath85 tev ( up ) and @xmath70 tev ( down ) with the first cut set .
, title="fig : " ] for both @xmath2 models , considering @xmath19 gev , @xmath85 tev ( up ) and @xmath70 tev ( down ) with the first cut set .
, title="fig : " ] for both @xmath2 models , considering @xmath19 , @xmath85 tev ( up ) and @xmath70 tev ( down ) , applying the first cut set .
, title="fig : " ] for both @xmath2 models , considering @xmath19 , @xmath85 tev ( up ) and @xmath70 tev ( down ) , applying the first cut set .
, title="fig : " ] for both @xmath2 models , considering @xmath19 , @xmath85 tev ( up ) and @xmath70 tev ( down ) , applying the first cut set .
, title="fig : " ] for both @xmath2 models , considering @xmath19 , @xmath85 tev ( up ) and @xmath70 tev ( down ) , applying the first cut set .
, title="fig : " ]
in order to explore the properties of a new neutral vector boson belonging to models with @xmath2 local symmetry , we study the drell - yan channel with a muon pair production , for two lhc proposed scenarios : @xmath20 tev ( @xmath86 fb@xmath87 ) and @xmath81 tev ( @xmath88 fb@xmath87 ) . our study is focused on two values of the @xmath11 boson mass , @xmath89 gev and @xmath90 gev .
+ first we can say that if @xmath8 is smaller than the centre of mass energy of the colliding protons , it can be discovered at the lhc .
for instance , for @xmath91 gev and @xmath81 tev we obtain @xmath92 events for both @xmath2 models .
the signals of @xmath11 bosons come from their direct production . in order to perform @xmath0 studies in @xmath93 collisions ,
many observables can be used .
we can cite the total cross - section and also several asymmetries very well studied by @xmath94 colliders .
the measurement of some of these observables is very difficult due to the signal , which needs to be detected above the background of the hadronic experiment .
thus , it is necessary to apply cuts taking into account the detector acceptance for reducing the background contribution . the invariant mass distribution of the muon pair final state is centered around the @xmath0 mass .
this behavior allows us to separate the signal from the background and to measure the @xmath0 boson mass .
several cuts can be applied to guarantee more precision on measurements .
the signal from the flipped model is above the one from the secluded model , and the corresponding difference can be enhanced if more severe cuts are considered . on the other hand
, the sm background remains below the signals of both @xmath2 models , as shown in figure 2 .
the muon transverse momentum distribution centered around the half of @xmath0 boson mass is a helpful observable for distinguishing the models . in this case
we have observed a dominance of the flipped model over the secluded one , which has a behavior similar to the sm .
see figure 5 .
the forward - backward charge asymmetry is another powerful tool to provide some information about the @xmath0 couplings to quarks in @xmath93 collisions , when leptons are in the final state , even being measured indirectly due to the unknown of original quark direction . to guarantee the correct direction of the initial parton we applied a cut on final muon pair @xmath95 .
the resulting @xmath96 asymmetries , for both b - l models , present clear different behaviors , mainly for the muon pair invariant mass in the range @xmath97 gev , as shown in figure 4 .
in addition to @xmath96 , the rapidity distributions are very sensitive to the @xmath0 couplings to quarks and in their computing we can consider light and heavy quarks .
they are very useful to disentangle the different models .
different cuts on it should be analyzed to exploit different responses from the same experiment .
the combination of the forward - backward asymmetry and rapidity distribution generates another interesting observable , @xmath98 .
if the lhc discovers a @xmath0 boson , the @xmath98 can be used as a refinement to select to which model the @xmath0 boson belongs to .
this combined observable in not under consideration in this work . besides the study for lhc operating in the high energy regime
, we have performed an analysis at @xmath20 tev for @xmath19 gev .
the obtained results present a clear signature for the @xmath0 boson production .
the dimuon invariant mass distribution shows a sharp peak around @xmath0 boson mass , above the sm background , as shown in figure 7 .
we note that the transverse momentum distribution once again reveals a peak on @xmath99 .
for these two distributions , the ones of the flipped model are far above those of the secluded model .
in general , the presented models can be well exploited by using the cited observables above at the lhc first energy stage .
some of them are more suitable for distinguishing the models .
the choice of the applied cuts is also important .
for instance , the same cuts that we had used to the forward - backward asymmetries are not appropriate to the rapidity distributions .
therefore , a combined analysis and different cuts should be applied to disentangle the models and guarantee more realist theoretical prediction on the @xmath11 couplings to fermions .
the lhc running at @xmath85 tev already is a powerful tool for unraveling physics beyond the sm .
1.cm acknowledgements e. c. f. s. fortes was supported by fapesp under the process 07/59398 - 2 ; j. c. montero was partially supported by cnpq under the process 307807/2006 - 1 ; y. a. coutinho thanks faperj for financial suport .
99 d. p. barber _ et al .
lett . * 43 * , 830 ( 1979 ) .
g. weiglein _ et al . _
( lhc / lc study group ) , phys . rept . * 426 * , 47 ( 2006 ) . r. w. robinett and j. l. rosner , phys . rev .
d * 25 * , ( 1985 ) 3036 ; p. langacker and m. luo , _
ibid_. d * 45 * , ( 1992 ) 278 ; j. erler and p. langacker , phys .
lett . * 84 * , ( 2000 ) 212 .
d. london and j. l. rosner , phys .
d * 34 * , ( 1986 ) 1530 ; j. kang and p. langacker , phys .
d * 71 * ( 2005 ) 035014 ; p. langacker , r. w. robinett and j. l. rosner , phys .
d * 30 * , 1470 ( 1984 ) ; j. l. hewett , t. g. rizzo , phys .
rep . * 183 * , 193 ( 1989 ) .
a. davidson , phys .
d * 20 * , ( 1979 ) 776 ; v. barger , e. ma and k. whisnant , phys .
d * 26 * , ( 1982 ) 2378 ; m. k. parida and a. raychaudhuri , phys .
d * 26 * , ( 1982 ) 2364 .
n. arkani - hamed , a. g. cohen and h. georgi , phys .
b * 513 * , 232 ( 2001 ) ; n. arkani - hamed , a. g. cohen , e. katz and a. e. nelson , jhep * 0207 * , 034 ( 2002 ) .
vernon d. barger , wai - yee keung , ernest ma .
b * 94*,(1980 ) 377 ; vernon d. barger , wai - yee keung , ernest ma .
d * 22 * ( 1980 ) 727 ; vernon d. barger , wai - yee keung , ernest ma .
* 44 * ( 1980 ) 1169 ; h. georgi and s. weinberg , phys .
d * 17 * ( 1978 ) 275 ; e. h. de groot , d. schildknecht and g. j. gounaris , university of bielefeld report no .
bi - tp 79/37 ( unpublished ) .
pedagogical introduction to extra dimensions " , thomas g. rizzo , slac - pub-10753 , ssi-2004-l013 , sep 2004 , arxiv : hep - ph/0409309v2 . f. pisano and v. pleitez , phys .
d * 46 * , 410 ( 1992 ) .
p. h. frampton , phys .
lett . * 69 * , 2889 ( 1992 ) .
j. e. cieza montalvo , m. d. tonasse , phys .
d * 67 * , 075022 ( 2003 ) .
de curtis , stefania , the bess model revisited as a higgsless linearmoose @xmath100 the lhc , arxiv:1002.2361 .
t. appelquist , b. a. dobrescu , and a. r. hopper , phys .
d * 68 * , 035012 ( 2003 ) .
j. c. montero and v. pleitez , phys .
b * 765 * , 64 ( 2009 ) .
e. c. f. s. fortes , j. c. montero , v. pleitez , phys .
d * 82 * , 114007 ( 2010 ) .
w. emam , and s. khalil , eur .
j. c52 , 625 ( 2007 ) .
a. abulencia _
et al . _ , ( cdf collaboration ) , phys .
96 , 211801 ( 2006 ) ; t. aaltoni _ et al .
_ , ( cdf collaboration ) , phys . rev
102 * , 091805 ( 2009 ) .
j. erler , p. langacker , s. munir , and e. rojas , jhep * 08 * , 017 ( 2009 ) .
g. cacciapaglia , c. cski , g. marandella , and a. strumia , phys .
d * 74 * , 033011 ( 2006 ) .
e. salvioni , g. villadoro , and f. zwirner , jhep * 11 * , 068 ( 2009).x k. nakamura _ et al .
( particle data group ) , j. phys .
g * 37 * , 075021 ( 2010 ) .
m. boezio _ et al .
_ , new j. phys . * 11 * , 105023 ( 2009 ) .
e. boos _ et al .
_ , [ comphep collaboration ] , comphep 4.4 : automatic computations from lagrangians to events , nucl .
instrum . meth .
a * 534 * , 250 ( 2004 ) .
michael dittmar , phys .
d * 55 * , 161 ( 1997 ) .
michael dittmar , anne - sylvie nicollerat , abdelhak djouadi , phys .
b * 583 * , 111 ( 2004 ) .
e. ramirez barreto , y. a. coutinho , j. s borges , phys .
b. * 689 * , 36 ( 2010 ) .
|
we study the @xmath0 phenomenology for two extensions of the electroweak standard model ( sm ) which have an extra @xmath1 gauge factor .
we show the capabilities of the lhc in distinguishing the signals coming from these two extensions and both of them from the standard model background . in order to compare the behavior of these @xmath2 models we consider the reaction @xmath3 and
compute some observables as the total cross sections , number of events , forward - backward asymmetry , final particle distributions like rapidity , transverse momentum , and dimuon invariant mass , for two lhc regimes : @xmath4 tev ( @xmath5 ) and @xmath6 tev ( @xmath7 ) for @xmath8 = 1000 gev and 1500 gev .
we show that by using appropriate kinematic cuts some of the observables considered here are able to extract different properties of the @xmath0 boson , and hence providing information about to which @xmath2 model it belongs to .
pacs:14.70.pw , 12.60.cn , 12.15.ji [email protected] ; [email protected];[email protected]
|
the birman - murakami - wenzl algebras are deformations of the brauer centralizer algebras @xcite .
they are quotients of the artin braid groups algebras , and have appeared in connection with the kauffman link invariant and the quantum groups of types b , c and d. the birman - murakami - wenzl algebras are generically semi - simple , and their structure was given by wenzl @xcite .
they play a key role in the construction of quantum invariants , modular categories and topological quantum field theories , as was shown by turaev and wenzl @xcite .
our purpose here is to study the structure of these algebras without using their representation theory . in a separate article
, we will pursue turaev and wenzl s program and construct four series of modular categories . together with the present paper
this construction will be reasonably self - contained .
our main results are the following .
* we give explicit formulas for minimal idempotents in the birman - murakami - wenzl algebras @xmath0 .
these are then used to obtain the semi - simple decomposition of @xmath0 , together with a basis of _ matrix units_. similar results were obtained by ram and wenzl @xcite using jones _ basic construction_. * we give a skein theoretic proof of the wenzl formula for the quantum dimensions of these minimal idempotents .
the key point is here the proof of the recursive formula ( [ rec ] ) .
this formula is further used to derive versions of the wenzl formula corresponding to the quantum group specializations and to discuss the existence of idempotents in the non generic case
. * conventions . * throughout this paper , the manifolds are compact , smooth and oriented . by a _ link _ we mean an isotopy class of an unoriented framed link . here
, a framing is a non - singular normal vector field up to homotopy . by a _ tangle _ in a @xmath1-manifold @xmath2 we mean an isotopy class of a framed tangle relative to the boundary . here
the boundary of the tangle is a finite set of points in @xmath3 , together with a non zero vector tangent to @xmath3 at each point .
note that a framing together with an orientation is equivalent to a trivialization of the normal bundle up to homotopy . by an _
oriented link _ we mean an isotopy class of a link together with a trivialization of the normal bundle up to homotopy . by an _ oriented tangle _ we mean an isotopy class of a tangle together with a trivialization of the normal bundle , up to homotopy relative to the boundary . here
the boundary of the tangle is a finite set of points in @xmath3 , together with a trivialization of the tangent space to @xmath3 at each point . in the figures ,
a preferred convention using the plane gives the framing ( blackboard framing ) .
let @xmath2 be a 3-manifold ( possibly with a given finite set @xmath4 of framed points on the boundary ) .
we denote by @xmath5 ( resp .
@xmath6 ) the @xmath7-module freely generated by links in @xmath2 ( and tangles in @xmath2 that meet @xmath3 in @xmath4 ) modulo ( the relative isotopy and ) the kauffman skein relations in figure [ kaufrel ] .
@xmath8 @xmath9 @xmath10 we suppose that @xmath7 is an integral domain containing the invertible elements @xmath11 , @xmath12 and that @xmath13 lives in @xmath7 .
we call @xmath5 the skein module of @xmath2 . for example , @xmath14 .
the isomorphism sends any link @xmath15 in @xmath16 to its kauffman polynomial @xmath17 , normalized by @xmath18 .
the birman - murakami - wenzl category @xmath19 is defined as follows . an object of @xmath19 is a disc @xmath20 equipped with a finite set of points and a non zero tangent vector at each point . unless otherwise specified , we will use the second vector of the standard basis ( the vector @xmath21 in complex notation ) . if @xmath22 and @xmath23 are two objects , the module @xmath24 is @xmath25,l_0\times 0\amalg l_1\times 1)$ ] .
the notation @xmath26 and @xmath27 will be used for @xmath24 and @xmath28 , respectively . for composition
, we use the _ covariant _ notation : @xmath29 in our figures the time parameter goes upwards , so that the morphism @xmath30 is depicted with @xmath31 lying above @xmath32 , and the normal vector field is orthogonal to the plane and points ` inside the blackboard ' .
the birman - murakami - wenzl category is a ribbon ab - category ( see ( * ? ? ?
* ch ii ) ) .
ribbon categories admit a theory of traces of morphisms and dimensions of objects for which we will use the terminology _ quantum trace _ and _ quantum dimension_. in the case of the category of finite dimensional vector spaces , equipped with trivial braiding and twist , these traces and dimensions coincide with the usual ones ( * ? ? ?
* section i.1.7 and lemma ii.4.3.1 ) .
we will use the notation @xmath33 for the quantum trace of @xmath34 .
this quantum trace is equal to the value of the closure of @xmath32 in @xmath35 obtained by gluing of @xmath36 and @xmath37 in @xmath27 along the identity map .
we denote by @xmath38 the object formed by the @xmath38 points @xmath39 ( @xmath40 is the trivial object ) .
if we consider only these standard objects @xmath38 , @xmath41 , we obtain a full subcategory equivalent to @xmath19 , which was named the kauffman category by turaev who first introduced it in ( * ? ? ?
* section 7.7 ) .
the algebra @xmath42 is isomorphic to the birman - murakami - wenzl algebra which is the quotient of the braid group algebra @xmath43 $ ] by the kauffman skein relations @xcite . for a proof of the above isomorphism ,
see @xcite or @xcite .
this algebra is a deformation of the brauer algebra ( i.e. the centralizer algebra of the semi - simple lie algebras of types b , c and d ) .
if @xmath7 is a field , then the algebra @xmath0 is known to be semi - simple @xcite , except possibly if @xmath12 is a root of unity , or @xmath44 for some @xmath45 .
its simple components correspond to the partitions @xmath46 with @xmath47 , @xmath48 $ ] .
the birman - murakami - wenzl algebra @xmath0 is generated by the identity @xmath49 , positive transpositions @xmath50 and hooks @xmath51 drawn in figure [ generators ] .
if the @xmath52 are supposed to be invertible , then a complete system of relations @xcite is given by @xmath53 the quotient of @xmath0 by the ideal @xmath54 generated by @xmath55 is isomorphic to hecke algebra @xmath56 .
we will use the knowledge of this hecke algebra @xmath57 to study @xmath0 . note that @xmath58 .
the hecke category @xmath59 is defined similarly as above , using the homfly skein theory .
an object in this category is a disc @xmath60 equipped with a set of points with a trivialization of the tangent space at each point .
if @xmath61 and @xmath62 are two objects , the module @xmath63 is the homfly skein module @xmath64,l_0\times 0\amalg l_1\times 1)$ ] . here
the homfly skein module in @xmath2 is freely generated by oriented framed tangles in @xmath2 modulo ( the relative isotopy and ) the homfly relations given in figure [ homflyrel ] .
note that there we have specialized the three variable homfly skein theory for framed links .
( 0,0)(1,1 ) ( .1,0)(.9,1 ) ( .9,0)(.6,.4 ) ( .4,.6)(.1,1 ) @xmath65 ( 0,0)(1,1 ) ( .9,0)(.1,1 ) ( .1,0)(.4,.4 ) ( .6,.6)(.9,1 ) @xmath66 ( 0,0)(1,1 ) ( .1,0)(.4,.5)(.1,1 ) ( .9,0)(.6,.5)(.9,1 ) + ( 0,0)(1,1 ) ( .1,0)(.15,.45)(.45,.75)(.7,.5)(.4,.2)(.20,.38 ) ( .14,.62)(.1,.9)(.1,1 ) @xmath67 ( 0,0)(1,1 ) ( .5,0)(.5,1 ) ( 0,0)(1,1 ) ( .1,1)(.15,.55)(.45,.25)(.7,.5)(.4,.8)(.20,.62 ) ( .14,.38)(.1,.1)(.1,0 ) @xmath68 ( 0,0)(1,1 ) ( .5,0)(.5,1 ) + @xmath69 ( 0,0)(1,1 ) ( .5,.5).4 @xmath70 we also simply denote by @xmath38 the object formed by the @xmath38 points @xmath39 , equipped with the standard trivialization .
the positive permutation braids @xmath71 represent a basis of the module @xmath57 , indexed by permutations @xmath72 .
the symmetrizers and antisymmetrizers in @xmath57 are represented respectively by the following elements @xmath73 and @xmath74 of the braid group algebra . @xmath75!}s^{-\frac{n(n-1)}{2 } } \sum_{\pi\in \mathcal{s}_n } s^{l(\pi)}w_\pi\ , \ ] ] @xmath76!}s^{\frac{n(n-1)}{2 } } \sum_{\pi\in \mathcal{s}_n } ( -s)^{-l(\pi)}w_\pi\ .\ ] ] here @xmath77 is the length of the permutation @xmath72 .
we work in the generic case .
this means that , in the domain @xmath7 , the quantum integers @xmath78=\frac{s^j - s^{-j}}{s - s^{-1}}$ ] are asked to be invertible for every @xmath79 . for a young diagram @xmath80 of size @xmath38
, we denote by @xmath81 the object of the category @xmath82 formed with one point for each cell @xmath83 of @xmath80 , equipped with the standard trivialization .
if @xmath83 has index @xmath84 ( @xmath85-th row , and @xmath86-th column ) , then the corresponding point in @xmath60 is @xmath87 .
following aiston and morton @xcite , we can define in @xmath88 a minimal idempotent @xmath89 which is a version of the corresponding young idempotent of the symmetric group algebra .
the idea of the construction is to insert symmetrizers along rows and antisymmetrizers along columns , and then to normalize .
a skein computation of the normalizing coefficient ( of slightly different idempotents ) appeared in @xcite .
details about the construction of these idempotents can also be found in @xcite . a standard tableau @xmath90 with shape a young diagram @xmath91 is a labeling of the cells , with the integers @xmath92 to @xmath38 , which is increasing along rows and columns .
we denote by @xmath93 the tableau obtained by removing the cell numbered by @xmath38 .
we define @xmath94 and @xmath95 by @xmath96 @xmath97 @xmath98 here @xmath99 is a standard isomorphism .
the following theorems are shown in @xcite .
[ astruct ] the family @xmath100 for all standard tableaus @xmath101 such that @xmath102 forms a basis for @xmath103 , and ( here @xmath104 is the kronecker delta ) @xmath105 this gives explicitly an algebra isomorphism @xmath106 where @xmath107 is the number of standard tableaus with shape @xmath80 , and @xmath108 is the algebra of @xmath109 matrices with coefficients in @xmath7 .
the diagonal elements @xmath110 are the _ path idempotents _ described in @xcite .
the _ minimal central idempotent _ corresponding to the partition @xmath111 is @xmath112 the minimal idempotents @xmath89 and the path idempotents @xmath113 satisfy the following branching formula .
[ abranch ] @xmath114 we have omitted in these formulas the standard isomorphisms respectively between @xmath115 and @xmath116 , and between @xmath117 and @xmath38 . the result for the quantum dimensions is given in the following theorem @xcite .
here we denote by @xmath118 the quantum trace in the ribbon hecke category .
[ aqdim ] @xmath119 the assertion above can be proven by a skein calculation ( see ( * ? ? ?
* prop.2.4 ) or @xcite ) . here is a sketch of the proof .
we first check the formula for columns @xmath120 , by using the recursive formula for the antisymmetrizers @xmath121 .
we note ( * ? ? ?
* ch.1 ) that the right hand side in theorem [ aqdim ] is the schur polynomial in the @xmath122 .
we then proceed recursively on the number of cells .
if @xmath111 contains two distinct sub - diagrams @xmath123 and @xmath124 with @xmath125 , then we get the result by considering @xmath126 ( considered as a composition of morphisms in @xmath127 ) .
this defines a quasi - idempotent which can be normalized . whence we get a minimal idempotent which belongs to the simple component indexed by @xmath111 , whose quantum trace gives the required formula .
we obtain the remaining cases , namely the rectangular diagrams , @xmath128 by using the branching formula for @xmath129 .
the quotient of @xmath0 by the ideal @xmath54 generated by @xmath55 is isomorphic to the hecke algebra @xmath56 .
we denote by @xmath130 the canonical projection map @xmath131 the main idea of our construction is to define a multiplicative section @xmath132 and to use it for the transport of the idempotents from the hecke algebra to the birman - murakami - wenzl one . in this section , we suppose that the ground ring @xmath7 is the field @xmath133 .
the computation of the quantum dimensions in section [ sectionqdim ] will permit to discuss the non generic case .
[ tr ] there exists a unique multiplicative homomorphism @xmath134 , such that @xmath135 @xmath136 [ cor1 ] @xmath137 .
the theorem above gives minimal central idempotents in @xmath0 , @xmath138 and also minimal path idempotents , @xmath139 the quantum trace of @xmath140 depends only on @xmath141 ; it is denoted by @xmath142 . as we did in section [ hecke ] , for each young diagram @xmath111 , we consider an object @xmath143 whose points correspond to the cells of @xmath111 . from theorem [ tr ] , we get the section @xmath144 , and we obtain a minimal idempotent in @xmath145 , @xmath146 [ orth ] if @xmath111 and @xmath147 are two distinct young diagrams with the same size @xmath148 , then for every @xmath149 one has @xmath150 .
+ we have @xmath151 with @xmath152 , @xmath153 and @xmath154 .
we omit here the isomorphisms between @xmath155 and @xmath156 .
then @xmath157 the result follows from the corresponding property in the category @xmath59 . we will need the following absorbing property which also results from computation in the hecke algebra ( * ? ? ?
* cor.1.10 ) .
if @xmath124 is a young diagram obtained from the young diagram @xmath80 by removing one cell , then one has : @xmath158 8 _ proof of theorem [ tr ] .
_ if the section @xmath159 exists , then it is unique .
this can be seen as follows .
let @xmath160 be the central idempotent corresponding to the factor @xmath54 . if @xmath161 is another section , then we have for every @xmath162 @xmath163 we will construct the section @xmath159 by induction on @xmath38 .
the result is certainly true for @xmath164 , since we have that @xmath165 .
let us assume that we have constructed @xmath166 satisfying the conditions of the theorem , for every @xmath167 , so that we have minimal idempotents , @xmath140 , for every @xmath90 such that @xmath168 .
let @xmath111 be a young diagram whose size is @xmath169 , then we have a minimal idempotent @xmath170 . if @xmath124 is a young diagram included in @xmath111 , such that @xmath171 , then we define @xmath172 , by [ yln]y_(,)= ( y_ ( 0.7,0.65 ) ( 0,0)@xmath92(0.34,0)(0,1)0.65 _ 1)(y_h_1 ) ( y _ ( 0.7,0.65 ) ( 0,0)@xmath92(0.34,0)(0,1)0.65 _ 1 ) .here , the standard isomorphisms between @xmath173 and @xmath174 are omitted .
we need that the quantum dimensions @xmath142 , with @xmath169 are not zero .
this result follows from ( * ? ? ?
* theorem 3.7 ) , and will also be proved , by considering the specializations corresponding to brauer algebras in the next section .
[ ortho ] a ) if @xmath124 and @xmath147 are two distinct young diagrams of size @xmath175 , included in @xmath111 , then @xmath176 + b ) if @xmath124 is a young diagram of size @xmath175 , included in @xmath111 , then @xmath177 is an idempotent . by the induction hypothesis
we can apply lemmas [ orth ] and [ abs ] to young diagrams of size @xmath167 .
the statement @xmath178 follows then from lemma [ orth ] applied to @xmath179 and @xmath123 with @xmath180 .
the square of @xmath181 is equal to @xmath182 times the skein element represented by the following tangle .
@xmath183 let us consider the intermediate morphism @xmath184 ( the subscript in @xmath185 and @xmath186 indicate which isomorphism in @xmath187 is used ) .
the minimality of the idempotent @xmath181 implies that this morphism is equal to @xmath181 , up to a coefficient which is obtained by considering the trace ( we use the absorbing property [ abs ] ) .
[ ptrace](y__/ ) ( y_)(y__/ ) = y _ .
statement @xmath188 follows .
let @xmath90 be a standard tableau whose size is @xmath189 , with shape @xmath190 .
we define @xmath191 and @xmath192 by lifting to the category @xmath19 the elements @xmath193 and @xmath194 defined in section [ hecke ] .
if @xmath193 and @xmath194 are represented by linear combinations of ` braids ' ( elements of the braid groupoid ) , then @xmath195 and @xmath196 are given by the following formulas .
@xmath198 we then have the formula @xmath199 for a young diagram @xmath124 with @xmath175 cells , included in @xmath141 , we define @xmath200 and @xmath201 , by @xmath202 @xmath203 using lemma [ ortho ] , we get the following .
[ ortho2 ] i ) @xmath204 ; @xmath205 ii ) @xmath206 .
we define a linear homomorphism @xmath161 from the braid group algebra @xmath43 $ ] to @xmath0 by [ defsn]xk[b_n ] s_n(x)= _
|(t)|=|()|=n-1 p^+_t x p^+ _ . we will show that this homomorphism induces a well defined linear map @xmath207 which is a section of @xmath130 , and prove multiplicativity
. the proof will be complete with the two following lemmas .
@xmath208 8 [ proofa ] @xmath209 .
we want to show that @xmath210 for any @xmath154 .
we write @xmath211 by the induction hypothesis , we have the result if @xmath212 is in @xmath213 .
so it is enough to consider the case where @xmath214 for @xmath215 an element of the matrix units basis described in section [ hecke ] , i.e. for @xmath216 , where @xmath217 and @xmath218 are standard tableaus with the same shape , whose size is @xmath189 .
if @xmath219 , then we have @xmath220 and the result follows .
it remains to check the case where @xmath221 .
@xmath222 the result @xmath210 follows ; @xmath223 can be obtained similarly . the map @xmath161 induces a well defined multiplicative homomorphism @xmath224 such that @xmath225 .
from lemma [ proofa ] , we can see that the homfly skein relation is respected , whence we have that @xmath159 is well defined .
we have that @xmath226 .
this implies that @xmath225 .
the computation below shows the multiplicativity . @xmath227
brauer centralizer algebras were introduced in @xcite in relation with the representation theory of the orthogonal and symplectic groups ( see also @xcite ) .
their structure was obtained by wenzl in @xcite .
we emphasize also nazarov s computations of the action of generators on their irreducible representations in @xcite .
his work includes the dimension formulas and inspired our computation of the quantum dimensions in section [ sectionqdim ] .
we have defined the birman - murakami - wenzl algebras @xmath0 by using kauffman skein theory .
brauer algebras can be defined in a similar way by using the classical version of kauffman relations given in figure [ clkaufrel ] .
@xmath228 @xmath229 @xmath230 here any coefficient ring is allowed , and @xmath231 could be an indeterminate .
if @xmath231 is a natural number , then we obtain brauer algebras with complex coefficients as a specialization of birman - murakami - wenzl algebras , with coefficient ring @xmath232 $ ] and @xmath233 , by setting @xmath234 .
it is a classical fact ( * ? ? ?
* ch.5 ) , that there exists an algebra homomorphism @xmath235 from this brauer algebra , denoted by @xmath236 , to the centralizer algebra @xmath237 , where @xmath238 is the fundamental representation of the orthogonal group @xmath239 .
if we denote by @xmath240 the canonical basis of @xmath241 , then @xmath235 is defined on the generators by @xmath242 @xmath243 in the above , @xmath244 is the kronecker delta . for @xmath245 ,
this homomorphism is injective .
in fact the above assignment extends to a monoidal functor from the specialized bmw category , to the linear category .
this functor is compatible with the ( trivial ) ribbon structures on these categories , and so it respects the ` quantum ' traces . in the case of the linear category
the quantum trace coincides with the usual one .
we get that , with the given specializations , @xmath246 in particular , if @xmath215 is a non trivial idempotent , then its quantum dimension is a non zero natural number .
the quotient of @xmath236 by the ideal @xmath54 generated by @xmath55 is isomorphic to the symmetric group algebra @xmath247 $ ] , the classical counterpart of the hecke algebra .
we denote by @xmath130 the canonical projection map @xmath248\ .\ ] ] [ cltr ] if @xmath231 is an integer greater or equal to @xmath38 , then , there exists a multiplicative homomorphism @xmath249\to d_n(n)$ ] , such that @xmath250}\text { and } \ ] ] @xmath251\;\;\ ; \forall y\in i_n\;\;\ ; s_n(x)y = ys_n(x)=0 \ .\ ] ] [ r1 ] as a corollary we get a non trivial minimal idempotent @xmath252 for every standard tableau of size @xmath38 , whose quantum trace is a non zero natural number . here
@xmath113 is the minimal path idempotent in the symmetric group algebra .
the recursive construction of the preceding section can be done .
the only point to check is that at each step the quantum dimensions @xmath253 are not zero . by the induction hypothesis
, we have non trivial minimal idempotents @xmath252 for every standard tableau of size @xmath189 .
if the shape of @xmath90 is @xmath111 , then we have that @xmath254 is a non zero integer .
the above completes the proof of theorem [ tr ] . at each step of the recursive construction
, we needed that the quantum dimensions @xmath253 with @xmath169 are non zero .
this is the case , because they become non zero integers if we apply the rank @xmath245 brauer specialization .
the same remark shows that theorem [ cltr ] holds if @xmath231 is generic ( e.g. for the brauer algebra with ground field @xmath255 ) .
in this section we will describe a matrix units basis in @xmath0 ( compare with @xcite ) , and show the branching formula . recall that in the categories @xmath59 and @xmath19 , for each young diagram @xmath111 , we have defined an object @xmath143 whose
points correspond to the cells of @xmath111 . in the proof of theorem [ tr ] ,
we have used the section @xmath144 to define a minimal idempotent @xmath256 .
a sequence @xmath257 of young diagrams , in which two consecutive diagrams @xmath258 and @xmath259 differ by exactly one cell will be called an up and down tableau of length @xmath38 , and shape @xmath260 .
those up and down tableaus of length @xmath38 such that @xmath261 ( up tableaus ) correspond bijectively with standard tableaus as described in section [ hecke ] . for an up and down tableau @xmath262 of length @xmath38 ,
we denote by @xmath263 the tableau of length @xmath189 obtained by removing the last young diagram in the sequence @xmath262 .
we define @xmath264 and @xmath265 by @xmath266 if @xmath267 , then @xmath268 @xmath269 if @xmath270 , then @xmath271 @xmath272 here we have omitted the standard isomorphism in @xmath273 and in @xmath274 .
note that for an up tableau , the definition is coherent with the one given for the corresponding standard tableau in the proof of theorem [ tr ] .
[ kbase ] a ) the family @xmath275 for all up and down tableaus @xmath262,@xmath276 of length @xmath38 , such that @xmath277 forms a basis for @xmath0 , and @xmath278 b ) there exists an algebra isomorphism @xmath279 where @xmath280 is the number of the up and down tableaus of length @xmath38 with shape @xmath80 , and @xmath281 is the algebra of @xmath282 matrices with coefficients in @xmath7 .
the diagonal elements @xmath283 are the path idempotents associated with the inclusions @xmath284 the minimal central idempotent corresponding to the partition @xmath111 is @xmath285 if @xmath262 corresponds to a standard tableau @xmath90 ( hecke part ) , then one has @xmath286 , and if @xmath287 then @xmath288 . from lemma [ orth ] if @xmath289 , and from lemma [ ortho2 ] if @xmath290 , we get y_k(_,_)y_=0 .this implies that @xmath291 if @xmath292 .
using the properties [ abs ] and [ ptrace ] , we show that , if @xmath38 is the length of @xmath262 , we have @xmath293 hence we have that @xmath278 the independence follows . to show that the family @xmath275 generate @xmath0 , we proceed recursively on @xmath38 . using the map @xmath159
, we see from the known result in @xmath57 that the @xmath275 , where @xmath260 and @xmath294 is the same young diagram with @xmath38 cells , generate the hecke part @xmath295 of @xmath0 .
it remains to consider @xmath54 . from the induction hypothesis
, we get that @xmath54 is generated by the following elements @xmath296 these are zero if @xmath297 , and else are equal to @xmath298 where @xmath299 .
the path idempotents @xmath300 satisfy the following branching formula .
[ branch ] @xmath301 we recall that in the above formula the shape of @xmath276 is either one cell more , either one cell less than the shape of @xmath262 .
the coordinates of @xmath302 in the standard basis are obtained by computing @xmath303 .
the result is zero unless @xmath276 and @xmath15 have the same shape @xmath147 and @xmath304 ; and in the latter case the result is @xmath305 . as a corollary we also have a branching formula for the minimal idempotents @xmath306 (
the obvious isomorphisms are omitted ) .
[ branch2 ] @xmath307 we can decompose @xmath308 by using the minimal central idempotents in @xmath309 .
we get @xmath310 in the above , only those up and down tableaus @xmath276 with @xmath311 contribute .
\i ) ( twist coefficient ) let @xmath312 be the minimal idempotent , then [ fr]= ^|| s^2_c cn(c ) y _ .here the content of a cell @xmath83 in the @xmath85-th row and the @xmath86-th column of @xmath147 is defined by @xmath313 .
\ii ) [ brad](braiding coefficient ) suppose that @xmath314 and @xmath315 contains only one cell @xmath83 .
then [ oi ] = s^2cn(c ) _ .suppose that @xmath316 and @xmath317 contains only one cell @xmath83 , then [ io ] = ^-2 s^-2cn(c ) _ ( , ) . the statements ( [ fr ] ) and ( [ oi ] ) follow from the corresponding in the hecke algebra ( see ( * ? ? ? * prop.1.11 ) ) . using the definition of the idempotent @xmath318 , we can bring ( [ io ] ) to the form @xmath319 where @xmath320 is depicted below .
@xmath321 by the schur lemma , @xmath322 with @xmath323 .
taking the quantum trace of this morphism , we get @xmath324 the first equality is due to the kauffman skein relations and the statement ( [ oi ] ) above .
the formula for the quantum dimension @xmath325 was obtained by wenzl ( * ? ? ?
* theorem 5.5 ) .
the proof there rests on the representation theory of the quantum group @xmath326 .
we give below an alternative proof for this formula .
our method is inspired by the nazarov computation of the matrix elements of the action of the hook generators on the canonical basis of the irreducible representations of brauer algebras @xcite .
here we work with scalar field @xmath327 , hence we have that all the quantum dimensions @xmath328 are invertible .
suppose that @xmath329 is obtained from @xmath123 by adding one cell in the @xmath85th row .
let @xmath4 be the number of pairwise distinct rows in the diagram @xmath147 .
then one can obtain @xmath330 diagrams by adding a cell to @xmath147 , and @xmath4 diagrams by removing a cell from @xmath147 .
let @xmath331 and @xmath332 be the contents ( defined in prop.[brad ] ) of these cells respectively
. denote by @xmath333 the scalars @xmath334 @xmath335 and by @xmath336 the value among @xmath337 corresponding to the diagram @xmath111 .
[ recdim ] one has [
rec]= b^-1(+1 ) _
b_jb .
let @xmath338 be the element of the algebra @xmath0 defined below .
@xmath339 for @xmath340 the equation in @xmath341 , @xmath342 defines a central element @xmath343 in @xmath344 .
we consider the formal power series in @xmath345 , @xmath346 we have that @xmath347 we denote by @xmath348 the series given by the action of @xmath349 on the simple component of @xmath344 indexed by @xmath147 . a canonical basis of @xmath350 is given in theorem [ kbase ] .
let @xmath262 be an up and down tableau , whose length is @xmath351 , and such that @xmath352 write the products @xmath353 and @xmath354 in the canonical basis .
@xmath355 let @xmath356 be the set of up and down tableaus @xmath357 such that @xmath358 for every @xmath359 .
by considering @xmath360 ( resp .
@xmath361 ) , we get @xmath362 using the three lemmas below the proof can be accomplished as follows .
from lemma [ lemm1 ] , lemma [ lemm2 ] and ( [ serq1 ] ) we have @xmath363 the required formula follows now from ( [ recqb ] ) .
[ lemm1 ] one has @xmath364 let @xmath263 be as usual obtained by removing the last term in the sequence @xmath262 .
we have @xmath365 we will obtain the diagonal term @xmath366 by computing the quantum trace of @xmath353 .
@xmath367 where @xmath368 is represented in the picture below .
+ @xmath369 clearly , @xmath370 and we get the required result for @xmath366 ; @xmath371 is calculated similarly .
let us denote by @xmath372 the residue of @xmath373 at @xmath374 , i.e. the coefficient by @xmath375 in the laurent expansion of this function in the neighborhood of the point @xmath374 .
@xmath376 we have that @xmath377 multiplying on the left by @xmath300 , we can express the above formula in the canonical basis .
we denote by @xmath378 the diagonal term of index @xmath262 . from the left hand side
, we get @xmath379 let us compute the coefficient from the right hand side . @xmath380 in the the above sum , the term of indices @xmath15 and @xmath276 is zero unless @xmath381 , and in this case the action of @xmath338 multiplies by the coefficient among @xmath333 corresponding to the young diagram @xmath294 .
these coefficients are distinct , and we know that @xmath382 is not zero .
this implies that @xmath383 is a rational function in @xmath384 , whose residue at @xmath374 is equal to @xmath366 .
the problem is now to compute the series @xmath385 .
we set [ serq]q_n(u)=z_n(u)+- , [ serq1 ] q(,u)=z(,u)+- .
[ lemq ] [ recqa]q_n+1(u)(-(s - s^-1)^2 ) = ( q_n(u ) ) ( -(s - s^-1)^2 ) [ recqb]q(,u)=(+ ) _
j in the following computations , we will drop some @xmath386 .
this means that by drawing the figures corresponding to these computation , one may have to add a vertical string on the right in order to get coherent equalities .
for example , we write e_n^-1_n+1=_ne_n .from the above , using the skein relation , we obtain @xmath387 this implies the following equalities for the formal series @xmath388@xmath389 we also have the formula symmetric to ( [ deux ] ) @xmath390 we note that @xmath338 and @xmath391 commute , and that @xmath392 multiplying ( [ un ] ) on the left by @xmath393 , we get @xmath394 using ( [ deux ] ) , ( [ trois ] ) and the skein relations , we get @xmath395 we multiply on each side by @xmath396 , and use the relations @xmath397 @xmath398 the recursive formula ( [ recqa ] ) can be deduced .
it can be written @xmath399 hence we have that @xmath400 recall that @xmath401 is the eigenvalue of @xmath338 corresponding to @xmath402 . the formula ( [ recqb ] )
is then established recursively .
note that @xmath403 whence we have the formula for @xmath404 . using theorem [ recdim ] we can deduce wenzl s dimension formula ( * ? ? ?
* theorem 5.5 ) . here
@xmath405 denote the transposed young diagram , so that @xmath406 is the length of the @xmath86th column of @xmath111 .
let @xmath407 and @xmath408 , we set @xmath409=\frac{\a s^d-\a^{-1}s^{-d}}{s - s^{-1 } } , \;\;\;\ ; [ n]=\frac{s^n - s^{-n}}{s - s^{-1}}.\ ] ] [ wen]= _ ( j , j) here , @xmath410 denote the hook - length of the cell @xmath84 , i.e. @xmath411 , and @xmath412 is defined by @xmath413 if we define @xmath414 by @xmath415 then we can write wenzl s formula as follows .
[ wenzltwo]= _ ( i , j) it would be nice to interpret the above formula by decomposing @xmath80 as a tensor product of two objects in some bigger category
. we will prove the formula ( [ wenzltwo ] ) .
we first write the recursive formula ( [ rec ] ) in a more convenient form .
we denote by @xmath416 the unique cell in the skew diagram @xmath417 .
[ recvar ] rcl & = & ( + 1 ) + & & _ j < i + & & _
j<i here the first big product gives the contribution of the coefficients @xmath418 corresponding to cells in the rows @xmath92 to @xmath419 .
note that some factors cancel if two among these rows have equal length .
we can write @xmath420 , where @xmath421 and @xmath422 satisfy the following recursive formulas . @xmath423
@xmath424 by induction , we can obtain the general formulas for @xmath425 and @xmath426 .
@xmath427 @xmath428 whence we get ( [ wenzltwo ] ) .
the following proposition gives the quantum dimension formulas for the specializations corresponding to the quantum groups of b , c , d series ( compare @xcite ) .
we consider here only partitions with at most @xmath38 rows .
[ erl ] a ) for @xmath429 ( @xmath430 specialization ) , one has for a partition @xmath431 ( which may have zero coefficients ) , @xmath432}{[n - j+1/2 ] } \prod_{1\leq i < j\leq n}\frac{[2n+\lambda_i - i+\lambda_j - j+1 ] [ \lambda_i - i-\lambda_j+j ] } { [ 2n - i - j+1 ] [ j - i]}\ ] ] b ) for @xmath433 ( @xmath434 specialization ) , one has for a partition @xmath431 , @xmath435 [ \lambda_i - i-\lambda_j+j ] } { [ 2n - i - j ] [ j - i ] } \text { if $ \lambda_n=0$;}\ ] ] @xmath436 [ \lambda_i - i-\lambda_j+j ] } { [ 2n - i - j ] [ j - i]}\text { if $ \lambda_n\neq0$;}\ ] ] c ) for @xmath437 ( @xmath438 specialization ) , one has for a partition @xmath431 , @xmath439}{[2n+2 - 2j ] } \prod_{1\leq i < j\leq n}\frac{[2n+2+\lambda_i - i+\lambda_j - j ] [ \lambda_i - i-\lambda_j+j ] } { [ 2n+2-i - j ] [ j - i]}\ ] ] observing that @xmath440 , we get formulas for the specializations which are _ symmetric _ to the above ones ( compare with @xcite ) .
for example , for @xmath441 , one has for a partition @xmath80 whose first part @xmath442 is at most @xmath38 , and whose transposed partition is @xmath443 , @xmath444}{[2n+2 - 2j ] } \prod_{1\leq i < j\leq n}\frac{[2n+2+\lambda_i^{\!\vee}-i+\lambda_j^{\!\vee}-j ] [ \lambda_i^{\!\vee}-i-\lambda_j^{\!\vee}+j ] } { [ 2n+2-i - j ] [ j - i]}\ ] ] suppose that @xmath445 , and that @xmath147 is obtained from the young diagram @xmath111 by removing one cell from the @xmath85th row , then we can write the recursive formula ( [ recvar ] ) as follows . if @xmath446 and @xmath447 , then [ recvar1 ] rcl & = & ( + 1 ) + & & _ j < n , else [ recvar2 ] rcl & = & ( + 1 ) + & & _ + & & in the a ) case , @xmath429 , if @xmath446 and @xmath447 , we get @xmath448 [ 1-n-\lambda_j+j ] } { [ n+\lambda_j - j+1 ] [ -n-\lambda_j+j]}\ , \ ] ] else @xmath449 } { [ n+\lambda_i - i-1/2 ] } \prod_{\underset{j\neq i}{1\leq j\leq n } } \frac{[2n+\lambda_i - i+\lambda_j - j+1 ] [ \lambda_i - i-\lambda_j+j ] } { [ 2n+\lambda_i - i+\lambda_j - j ] [ \lambda_i - i-\lambda_j+j-1]}\ .\ ] ] the announced result follows . in the b ) case , @xmath433 , if @xmath446 and @xmath447 , we get @xmath450 [ 1-n-\lambda_j+j ] } { [ n+\lambda_j - j ] [ -n-\lambda_j+j]}\ , \ ] ] else @xmath451 [ \lambda_i - i-\lambda_j+j ] } { [ 2n+\lambda_i - i+\lambda_j - j-1 ] [ \lambda_i - i-\lambda_j+j-1]}\ .\ ] ] the formulas in b ) follow . in the c )
case , @xmath437 , if @xmath446 and @xmath447 , we get @xmath452 [ 1-n-\lambda_j+j ] } { [ n+2+\lambda_j - j ] [ -n-\lambda_j+j]}\ , \ ] ] else @xmath453 } { [ 2n+2\lambda_i-2i ] } \prod_{\underset{j\neq i}{1\leq j\leq n } } \frac{[2n+2+\lambda_i - i+\lambda_j - j ] [ \lambda_i - i-\lambda_j+j ] } { [ 2n+1+\lambda_i - i+\lambda_j - j ] [ \lambda_i - i-\lambda_j+j-1]}\ .\ ] ] the formula follows .
we conclude this paper with a summary of conditions needed to define our minimal idempotents @xmath140 and @xmath306 .
this is of importance in the non generic case where @xmath11 and @xmath12 are roots of unity and some of quantum integers @xmath454 $ ] are non invertible .
let @xmath147 is a young diagram obtained from @xmath111 by removing one cell . from corollary [ branch2 ]
we get a formula for the idempotent @xmath306 . here
we denote by @xmath455 any lifting of @xmath456 .
we have @xmath457 note that the first sum on the right hand side lives in the hecke summand of @xmath145 , hence in each term we can replace @xmath458 by @xmath306 . by using lemmas [ orth ] and [ abs ] we get : y_= y_(y _ * the quantum integer @xmath454 $ ] is non zero for any @xmath459 ; * the idempotent @xmath305 is defined for _ some _ @xmath460 , @xmath461 ; * the coefficient @xmath462 given in theorem [ recdim ] is nonzero for any @xmath463 , @xmath464 .
let @xmath90 be a standard tableau , with shapes @xmath465 , and @xmath466 .
then from the general formula [ defsn ] for the section @xmath159 ( @xmath467 ) , we get p_t= p^+_t p_t p^+_t . here
@xmath468 is any lifting in @xmath0 of the path idempotent @xmath469 .
we obtain a minimal idempotent @xmath140 if the following three conditions are satisfied : 10 a.k .
aiston , h.r .
morton , _ idempotents of hecke algebras of type @xmath470 _ , journal of knot theory and ram .
vol . 7 , no . 4 ( 1998 ) , 463487 .
a. k. aiston , _ a skein theoretic proof of the hook formula for quantum dimension _ , ( preprint 1997 ) .
a. beliakova , c. blanchet , _ modular categories of types b , c and d _ , preprint math.qa/0006227 ( 2000 ) j.s .
birman , h. wenzl , _ braids , link polynomials and a new algebra _ ,
trans . of ams , vol .
313 , no . 1 ( 1989 ) ,
c. blanchet , _ hecke algebras , modular categories and @xmath1-manifolds quantum invariants _ , topology , vol .
39 , no . 1 ( 2000 ) , 193223 .
c. blanchet , n. habegger , g. masbaum and p. vogel , _ topological quantum field theories derived from the kauffman bracket _
, topology .
34 , no . 4 ( 1995 ) , 883927 . j. birman , h. wenzl , _ braids , link polynomials and a new algebra _ , trans .
ams , vol .
313 , no . 1 ( 1989 ) ,
r. brauer , _ on algebras which are connected with the semisimple continuous groups _
, ann . of math .
38 ( 1937 ) , 854872 .
j. erlijman , _ new subfactors from braid group representations _ , trans .
350 , no . 1 ( 1998 ) 185 - 211 .
c. kassel , m. rosso , v. turaev , _ quantum groups and knots invariants _ , panoramas et synthses no .
5 , smf ( 1997 ) .
i. g. macdonald , _ symmetric functions and hall polynomials _ , oxford math .
monographs ( 1979 ) .
h. r. morton and p. traczyk , _ knots and algebras _ , contribuciones matematicas en homaje al professor d. antonio plans sanz de bremond , e. martin - peinador and a. rodez editors , university of saragoza ( 1990 ) , 201220 .
j. murakami , _ the representations of the @xmath471-analogue of brauer s centralizer algebras and the kauffman polynomial of links _
inst . math .
26 , no . 6 ( 1990 ) ,
m. nazarov , _ young orthogonal form for brauer s centralizer algebra _ , j. of algebra .
( 1996 ) , 664693 .
a. ram , h. wenzl , _ matrix units for centralizer algebras _ , j. of algebra .
( 1992 ) , 378395 . v. g. turaev ,
_ quantum invariants of knots and @xmath1-manifolds _ , de gruyter studies in math .
: 18 ( 1994 ) .
v. g. turaev , _ operator invariants of tangles , and @xmath472-matrices _ , math .
35 , no . 2 ( 1990 ) , 411443 . v. turaev , h. wenzl , _ quantum invariants of @xmath1-manifolds associated with classical simple lie algebras _ , int . j. of math .
4 , no . 2 ( 1993 ) ,
v. turaev , h. wenzl , _ semisimple and modular categories from link invariants _ , math .
( 1997 ) , 411461 .
h. wenzl , _ quantum groups and subfactors of type b , c , and d _ , comm . math .
( 1990 ) , 383432 .
h. wenzl , _ hecke algebras of type @xmath473 and subfactors _ , invent .
92 ( 1988 ) , 349383 .
h. wenzl , _ on the structure of brauer s centralizer algebras _ , annals of math .
128 ( 1988 ) , 173193 .
h. weyl , _ the classical groups _ , princeton university press ( 1939 ) .
y. yokota , _ skeins and quantum @xmath474 invariants of @xmath1-manifolds _ , math .
ann . vol . 307
( 1997 ) , 109138 .
|
we give skein theoretic formulas for minimal idempotents in the birman - murakami - wenzl algebras .
these formulas are then applied to derive various known results needed in the construction of quantum invariants and modular categories . in particular , an elementary proof of the wenzl formula for quantum dimensions
is given .
this proof does not use the representation theory of quantum groups and the character formulas .
|
mass predictions for scalar @xmath0 glueballs extracted from qcd sum - rules have been problematic mainly due to discrepancies between analyses which are sensitive to the low - energy theorem for gluonic correlation functions and those which are insensitive to this quantity @xcite .
such a discrepancy would be indicative of two widely - separated states , a result which has already been seen to occur in explicit two - resonance analyses of laplace sum - rules even in the absence of mixing with quark scalar resonances @xcite .
however , there exists substantial evidence that these discrepancies are resolved by the inclusion of instanton @xcite effects in the laplace sum - rules for scalar glueballs @xcite .
recently , techniques for using gaussian sum - rules @xcite to predict hadronic properties have been developed @xcite .
in particular , these methods concentrate on normalized gaussian sum - rules which are independent of the finite - energy sum - rule constraint which is central to the original heat - evolution studies @xcite of the gaussian sum - rules
. advantages of this approach compared with laplace sum - rules include enhanced sensitivity to the hadronic spectral function over a wide range of energy . in this paper ,
these techniques are employed and generalized in an effort to obtain resonance parameter predictions for scalar gluonium .
furthermore , the formulation of these sum - rules is extended to include gaussian kernels weighted by integer powers , @xmath1 } \frac{1}{\pi } \rho(t ) { \ensuremath { \mathrm{d}t } } \quad,\quad k\geq -1\ ] ] where @xmath2 is a hadronic spectral function with physical threshold @xmath3 .
similar to the laplace sum - rules , the low - energy theorem ( let ) @xcite ( see ( [ let ] ) below ) for scalar gluonic currents enters _ only _ the @xmath4 gaussian sum - rule , and instanton contributions to the correlation function serve to mitigate the discrepancy between the @xmath4 and @xmath5 sum - rules . however , theoretical uncertainties associated with the instanton and let parameters are shown to be overwhelming in the @xmath4 sum - rule , rendering it unsuitable for phenomenological analysis .
thus the @xmath6 gaussian sum - rule is the focus of our detailed predictions for scalar glueballs , including an estimate of theoretical uncertainties .
we show that the gaussian sum - rules of scalar gluonic currents contain signatures that the hadronic spectral function is distributed over a broad energy range , ( 0.81.6 ) gev including qcd uncertainties , and that this distribution most likely consists of two separate resonances . since these sum - rules probe the gluonic content of hadronic states , they are sensitive to the glueball component of the observed scalar mesons which in general could be glueball - quark meson mixtures .
thus our results are relevant to the interpretation of the scalar isoscalar resonances in this region : the @xmath7 , @xmath8 , @xmath9 , @xmath10 , and @xmath11 @xcite . in the next section ,
gaussian sum - rules for scalar gluonic currents are developed . in section
[ analysis ] , we develop analysis techniques and employ them to analyze the gaussian sum - rules using a variety of phenomenological models .
the results of this phenomenological analysis , including theoretical uncertainties , are consolidated in section [ discussion ] , and a summary of our results is contained in section [ conclusion ] .
the most important quantity in any sum - rules approach to determining hadron properties is the correlation function for the particular channel under inspection : @xmath12 where @xmath13 is the qcd vacuum state , @xmath14 is the time - ordering operator , and @xmath15 is that current which corresponds to the quantum numbers of interest . in this paper , we wish to focus on scalar glueballs and so we choose the following current : @xmath16 which is renormalization - group invariant in the chiral limit of @xmath17 massless quarks . the gluon field strength tensor @xmath18 is defined by @xmath19 and @xmath20
is the qcd beta function describing the momentum scale dependence of the strong coupling parameter @xmath21
@xmath22 from the asymptotic form and assumed analytic properties of ( [ corr ] ) follows a dispersion relation with three subtraction constants @xmath23 where @xmath2 is the hadronic spectral function is often denoted by @xmath24 . ] with physical threshold @xmath3 .
the spectral function @xmath2 is related to a physical process and is thus determined phenomenologically .
in contrast , @xmath25 is calculated theoretically from qcd , and the constant @xmath26 follows from the low - energy theorem @xcite @xmath27 for these reasons , we shall refer to the left - hand side of ( [ dispersion ] ) as the theoretical side and the right - hand side as the phenomenological side . in this regard , eqn .
( [ dispersion ] ) serves to relate theory to phenomenology , and , in principle , could be used to predict the properties of hadrons from qcd .
however , as it stands , eqn .
( [ dispersion ] ) is not actually that well - suited to this task .
for instance , although the constant @xmath26 is determined by the low - energy theorem ( [ let ] ) , the constants @xmath28 and @xmath29 are not .
further , the theoretical calculation of @xmath25 contains a field theoretical divergence proportional to @xmath30 .
in addition , from a phenomenological perspective , the integral on the right - hand side of ( [ dispersion ] ) is far too sensitive to the high energy behaviour of @xmath2 to effectively probe low - lying resonances . to circumvent these shortcomings
, we consider the one - parameter family of gaussian sum - rules in ( [ srdef ] ) . ]
@xmath31 with the borel transform @xmath32 defined by @xmath33 applying definition ( [ srdef ] ) to both sides of ( [ dispersion ] ) alleviates the difficulties surrounding ( [ dispersion ] ) : the infinite number of derivatives in ( [ borel ] ) annihilate the unwanted low - energy constants and the field theoretical divergence contained in @xmath25 . furthermore ,
as we shall see , a key feature of the resulting sum - rules is the introduction of a gaussian weight factor to the integrand on the phenomenological side of ( [ dispersion ] ) .
this serves to suppress contributions from @xmath2 away from the gaussian peak a desirable situation considering that we wish to extract information concerning low - lying resonances .
let us first consider ( [ srdef ] ) as applied to the theoretical side of ( [ dispersion ] ) .
as noted previously , the low - energy constants @xmath28 and @xmath29 are annihilated by the borel transform ; however , the constant @xmath26 does produce a contribution unique to the case @xmath4 . using the following identity : @xmath34 = \frac{1}{4\tau }
( -a)^n \exp\left ( \frac{-a}{4\tau}\right)\ \mathrm{for}\ n\geq 0 \quad,\ ] ] it is trivial to show that the contribution to the theoretical side of the gaussian sum - rules devolving from the low - energy constant is given by @xmath35 to proceed further , however , we must settle on a specific form for the scalar glueball correlator @xmath25 .
we choose to partition the correlator into the following sum of qualitatively distinct terms : @xmath36 where the superscript _ qcd _ signifies that ( [ gluecorr ] ) is a theoretical approximation to the true correlator .
the first two terms in ( [ gluecorr ] ) devolve from the operator product expansion of the current ( [ current ] ) .
the quantity @xmath37 is the contribution from ordinary perturbation theory whereas @xmath38 is the result of nonzero vacuum expectation values of local gluonic operators ( condensates ) . for three colours and three flavours of massless quarks ( @xmath39 ) ,
@xmath37 is given at three - loop order by @xcite @xmath40 \\
a_0 = -2\left(\frac{\alpha}{{\ensuremath{\mathrm{\pi}}}}\right)^2\left[1+\frac{659}{36}\frac{\alpha}{{\ensuremath{\mathrm{\pi}}}}+ 247.480\left ( \frac{\alpha}{{\ensuremath{\mathrm{\pi}}}}\right)^2\right ] \ , , \quad a_1 = 2\left(\frac{\alpha}{{\ensuremath{\mathrm{\pi}}}}\right)^3\left [ \frac{9}{4 } + 65.781\frac{\alpha}{{\ensuremath{\mathrm{\pi}}}}\right ] \ , , \quad a_2 = -10.1250\left(\frac{\alpha}{{\ensuremath{\mathrm{\pi}}}}\right)^4 \label{pert_coefficients}\end{gathered}\ ] ] where we have omitted the aforementioned field theoretical divergence as this term
is annihilated by the borel operator ( [ borel ] ) . incorporating into @xmath38 next - to - leading order @xcite contributions from @xmath41 , and
the operator basis has been changed from @xmath42 to @xmath43 . ] from the dimension four gluon condensate @xmath43 and leading order @xcite contributions from gluonic condensates of dimension six and eight @xmath44 yields @xmath45 \langle j\rangle + \frac{c_0}{q^2}\left\langle { \cal o}_6\right\rangle + \frac{d_0}{q^4}\left\langle { \cal o}_8\right\rangle \\
b_0 = 4{\ensuremath{\mathrm{\pi}}}\frac{\alpha}{{\ensuremath{\mathrm{\pi}}}}\left [ 1 + \frac{175}{36}\frac{\alpha}{{\ensuremath{\mathrm{\pi}}}}\right ] \ , , \quad b_1 = -9{\ensuremath{\mathrm{\pi}}}\left(\frac{\alpha}{{\ensuremath{\mathrm{\pi}}}}\right)^2 \ , , \quad c_0 = 8{\ensuremath{\mathrm{\pi}}}^2\left(\frac{\alpha}{{\ensuremath{\mathrm{\pi}}}}\right)^2 \ , , \quad d_0 = 8{\ensuremath{\mathrm{\pi}}}^2\frac{\alpha}{{\ensuremath{\mathrm{\pi } } } } \quad .
\label{cond_coefficients}\end{gathered}\ ] ] the final term on the right - hand side of ( [ gluecorr ] ) is a contribution arising from direct instanton effects .
note that in decoupling this term from perturbation theory and the condensate contributions , we have tacitly assumed that interference between classical and quantum fields is small .
further , we also assume that the dominant contribution to @xmath46 comes from bpst single instanton and anti - instanton solutions @xcite and that multi - instanton effects are negligible @xcite . with these provisions , we have @xcite @xmath47 ^ 2 { \ensuremath { \mathrm{d}n } } ( \rho)\quad , \ ] ] where @xmath48 is the modified bessel function of the second kind of order two ( _ c.f . _
@xcite ) , @xmath49 is the instanton radius , and @xmath50 is the instanton density function . before substituting ( [ gluecorr ] ) into ( [ srdef ] ) , it is convenient to first simplify ( [ srdef ] ) by employing a particularly useful identity relating the borel transform ( [ borel ] ) to the inverse laplace transform @xcite @xmath51 = \frac{1}{4\tau}{\ensuremath{\mathcal{l}^{-1}}}[f(\delta^2)]\ ] ] where , in our notation , @xmath52 = \frac{1}{2{\ensuremath{\mathrm{\pi}}}i } \int_{a-{\ensuremath{\mathrm{i}}}\infty}^{a+{\ensuremath{\mathrm{i}}}\infty } f(\delta^2 ) \exp\left ( \frac{\delta^2}{4\tau } \right ) { \ensuremath { \mathrm{d}\delta^2 } } \ ] ] with @xmath53 chosen such that all singularities of @xmath54 lie to the left of @xmath53 in the complex @xmath55-plane .
( note , there is no loss of generality in assuming that @xmath56 . ) rewriting ( [ srdef ] ) using ( [ bor_to_lap ] ) gives @xmath57 if in the first integral above , we make the substitution @xmath58 and in the second , we make the substitution @xmath59 , then ( [ mid ] ) reduces to @xmath60 \pi(w){\ensuremath { \mathrm{d}w } } + 2\int_{\gamma_2 } ( -w)^k \exp\left [ \frac{-(\hat{s}+w)^2}{4\tau}\right ] \pi(w){\ensuremath { \mathrm{d}w } } \right\ } \\
\hspace{2em } & = \frac{1}{\sqrt{4{\ensuremath{\mathrm{\pi}}}\tau}}\frac{1}{2{\ensuremath{\mathrm{\pi}}}i } \int_{\gamma_1 + \gamma_2 } ( -w)^k \exp \left [ \frac{-(\hat{s}+w)^2}{4\tau}\right ] \pi(w ) { \ensuremath { \mathrm{d}w } } \label{finish}\end{aligned}\ ] ] where @xmath61 and @xmath62 are two parabolas ( depicted in figure [ cont_fig ] ) in the complex @xmath63-plane defined by @xmath64 \label{gamma1 } \\
\gamma_1 = -\hat{s}+{\ensuremath{\mathrm{i}}}(a^2+x^2)^{1/4 } \exp\left [ \frac{{\ensuremath{\mathrm{i}}}}{2}{\ensuremath{\mathrm{arctan}}}\left ( \frac{x}{a}\right ) \right ] \label{gamma2}\end{gathered}\ ] ] for all @xmath65 .
now , we must substitute ( [ gluecorr ] ) into ( [ finish ] ) and calculate the resulting complicated integral . towards this end , it is advantageous to consider the closed contour @xmath66 depicted in figure [ deformed ] .
our expression for the correlator ( [ gluecorr ] ) is analytic in the complex @xmath67-plane except for a branch cut along the negative real semi - axis originating from a branch point located at the origin .
consequently , @xmath68 \pi^{{\ensuremath{\mathrm{qcd}}}}(w){\ensuremath { \mathrm{d}w } } \\ & = \frac{1}{2{\ensuremath{\mathrm{\pi}}}i}\frac{1}{\sqrt{4{\ensuremath{\mathrm{\pi}}}\tau } } \left\ { \int_{\tilde{\gamma}_1(r)+\tilde{\gamma}_2(r ) } + \quad\int_{\gamma_{\text{c}}+\gamma_{\epsilon } } + \quad\int_{\gamma_3+\gamma_4+\gamma_5 } \right\ } ( -w)^k \exp \left [ \frac{-(\hat{s}+w)^2}{4\tau } \right ] \pi^{{\ensuremath{\mathrm{qcd}}}}(w){\ensuremath { \mathrm{d}w } } \label{big}\end{aligned}\ ] ] where @xmath69 and @xmath70 are respectively those portions of the contours @xmath61 and @xmath62 ( see ( [ gamma1 ] ) and ( [ gamma2 ] ) ) lying in the interior of a circle of radius @xmath71 centered at @xmath72 . for large @xmath71 ,
the integral over @xmath73 approaches zero and the contours @xmath69 and @xmath70 approach @xmath61 and @xmath62 .
therefore , by rearranging ( [ big ] ) , recalling ( [ finish ] ) , and taking appropriate limits , we get @xmath74 \pi^{{\ensuremath{\mathrm{qcd}}}}(w){\ensuremath { \mathrm{d}w } } \\ & = \frac{1}{2{\ensuremath{\mathrm{\pi}}}{\ensuremath{\mathrm{i}}}}\frac{(-1)^{k+1}}{\sqrt{4{\ensuremath{\mathrm{\pi}}}\tau } } \lim_{\epsilon\rightarrow 0 } \int_{\gamma_{\epsilon } } w^k \exp\left [ \frac{-(\hat{s}+w)^2}{4\tau } \right ] \pi^{{\ensuremath{\mathrm{qcd}}}}(w ) { \ensuremath { \mathrm{d}w } } \nonumber\\ & \hspace{10em}+ \frac{1}{\sqrt{4{\ensuremath{\mathrm{\pi}}}\tau } } \lim_{\epsilon\rightarrow 0 } \int_{\epsilon}^{\infty } t^k \exp \left [ \frac{-(\hat{s}-t)^2}{4\tau } \right ] \frac{1}{{\ensuremath{\mathrm{\pi}}}}{\ensuremath{\mathrm{im}}}\pi^{{\ensuremath{\mathrm{qcd}}}}(t ) { \ensuremath { \mathrm{d}t } } \quad , \label{theorysr}\end{aligned}\ ] ] where @xmath75\quad.\ ] ] eqn .
( [ theorysr ] ) is our final expression for the contribution to the @xmath76-th gaussian sum - rule of scalar gluonium stemming from the correlator ( [ gluecorr ] ) . later in this section , however , we do evaluate ( [ theorysr ] ) for the specific cases @xmath77 ( see ( [ gm1 ] ) and ( [ g0 ] ) ) .
we must now consider the phenomenological contribution to the gaussian sum - rules .
substituting the right - hand side of ( [ dispersion ] ) into ( [ srdef ] ) and again making use of the identity ( [ identity ] ) , it is simple to show that @xmath78 \frac{1}{{\ensuremath{\mathrm{\pi}}}}\rho(t ) { \ensuremath { \mathrm{d}t } } \quad.\ ] ] in sum - rules analyses , it is customary to approximate the spectral function @xmath2 using a `` resonance(s ) + continuum '' ansatz . in this model ,
hadronic physics is ( locally ) dual to qcd above the continuum threshold @xmath79 , and so we write @xmath80 where @xmath81 is the heaviside step function .
( we shall have much more to say concerning @xmath82 in section [ analysis ] . )
substituting ( [ phenom ] ) into ( [ phenom_sr ] ) and comparing the result to the theoretical expression ( [ theorysr ] ) shows us that the continuum contribution @xmath83 \frac{1}{\pi } { \ensuremath{\mathrm{im}}}\pi^{{\ensuremath{\mathrm{qcd}}}}(t ) { \ensuremath { \mathrm{d}t } } \ ] ] is common to both ; therefore , we define @xmath84 and write ( recall the low - energy contribution ( [ gauss_let ] ) unique to the @xmath4 sum - rule ) @xmath85 with @xmath86 \frac{1}{\pi } \rho^{\text{had}}(t ) { \ensuremath { \mathrm{d}t } } \quad,\quad k\geq-1\quad .
\label{g_had_def}\ ] ] we note that ( [ srm1][g_had_def ] ) have meaning for @xmath87 since this simply represents a gaussian kernel whose peak lies outside the @xmath88 physical region , and hence only the gaussian tail extends into the physical region . as will later be shown , the qcd results scale with @xmath89 through the renormalization - group equation , and hence qcd also presents no obstacles to considering @xmath90 . indeed , the seminal work @xcite on gaussian sum - rules considered symmetric and antisymmetric combinations @xmath91 which implicitly employs gaussian sum - rules with negative @xmath92 . in this paper
, we focus exclusively on the @xmath93 gaussian sum - rules . substituting ( [ gluecorr ] ) into ( [ blah ] ) and recalling ( [ theorysr ] )
gives ( for details on simplifying the relevant integrals in ( [ theorysr ] ) , see @xcite ) @xmath94 \left [ ( a_0-\pi^2 a_2 ) + 2a_1\log\left ( \frac{t}{\nu^2 } \right ) + 3a_2 \log^2\left ( \frac{t}{\nu^2 } \right ) \right]{\ensuremath { \mathrm{d}t } } \\ & + \frac{1}{\sqrt{4\pi\tau } } \exp\left ( \frac{-\hat{s}^2}{4\tau}\right ) \left [ -b_0 \langle j\rangle + \frac{c_0 \hat{s}}{2\tau}\left\langle { \cal o}_6\right\rangle -\frac{d_0}{4\tau}\left(\frac{\hat{s}^2}{2\tau } -1 \right ) \left\langle { \cal o}_8\right\rangle \right ] \\ & -\frac{16\pi^3}{\sqrt{4\pi\tau } } \int{\ensuremath { \mathrm{d}n } } ( \rho)\rho^4 \int_0^{s_0 } t \exp\left [ \frac{-(\hat{s}-t)^2}{4\tau}\right ] j_2\left(\rho\sqrt{t } \right ) y_2\left(\rho\sqrt{t } \right ) { \ensuremath { \mathrm{d}t } } \\ & -\frac{128\pi^2}{\sqrt{4\pi\tau } } \exp\left ( \frac{-\hat{s}^2}{4\tau}\right ) \int{\ensuremath { \mathrm{d}n } } ( \rho ) \end{split}\ ] ] @xmath95 \left [ ( a_0-\pi^2 a_2 ) + 2a_1\log\left ( \frac{t}{\nu^2 } \right ) + 3a_2 \log^2\left ( \frac{t}{\nu^2 } \right ) \right ] { \ensuremath { \mathrm{d}t } } \\ & - \frac{1}{\sqrt{4\pi\tau } } b_1\langle j\rangle \int_0^{s_0 } \exp\left [ \frac{-(\hat{s}-t)^2}{4\tau}\right ] { \ensuremath { \mathrm{d}t } } + \frac{1}{\sqrt{4\pi\tau } } \exp\left ( \frac{-\hat{s}^2}{4\tau}\right ) \left [ c_0 \left\langle { \cal o}_6\right\rangle - \frac{d_0 \hat{s}}{2\tau } \left\langle { \cal o}_8\right\rangle \right ] \\ & -\frac{16\pi^3}{\sqrt{4\pi\tau } } \int{\ensuremath { \mathrm{d}n } } ( \rho)\rho^4 \int_0^{s_0 } t^2 \exp\left [ \frac{-(\hat{s}-t)^2}{4\tau}\right ] j_2\left(\rho\sqrt{t } \right ) y_2\left(\rho\sqrt{t } \right ) { \ensuremath { \mathrm{d}t } } \end{split}\ ] ] where @xmath96 and @xmath97 are bessel functions of order two of the first and second kind respectively ( _ c.f . _
renormalization - group improvements @xcite of ( [ gm1 ] ) and ( [ g0 ] ) amount to replacing the strong coupling constant @xmath21 ( contained in the coefficients ( [ pert_coefficients ] ) and ( [ cond_coefficients ] ) ) by the running coupling @xmath98 at the renormalization scale @xmath99 . at three loop order with @xmath39 in the @xmath100 renormalization scheme , we have @xcite @xmath101 \label{run_coupling}\\ l=\log\left(\frac{\nu^2}{\lambda^2}\right)\quad , \quad \bar\beta_i=\frac{\beta_i}{\beta_0 } \quad , \quad \beta_0=\frac{9}{4}\quad , \quad \beta_1=4\quad ,
\quad \beta_2=\frac{3863}{384}\end{gathered}\ ] ] with @xmath102 for three active flavours , consistent with current estimates of @xmath103 @xcite and the subsequent impact on @xmath104 has been investigated through pad approximation techniques @xcite . ] and matching conditions through the charm threshold @xcite .
the normalization of the gaussian sum - rules is related to the finite - energy sum - rules ( fesrs ) @xcite as can be seen by integrating ( [ srm1 ] ) and ( [ sr0 ] ) with respect to @xmath105 to obtain @xmath106 we recognize the quantities on the right - hand sides of ( [ tom_norm_1 ] ) and ( [ tom_norm_2 ] ) from the definition of the ( fesrs ) @xmath107 where @xmath108 represents a qcd prediction .
thus we see that the overall normalization of @xmath109 ( or @xmath110 ) is constrained by the finite - energy sum - rules .
this result is not surprising in light of the seminal work on gaussian sum - rules which established the significance of the fesr constraint by considering the evolution of the gaussian sum - rules through the diffusion equation @xcite .
it was found that this `` heat - evolution '' of the resonance plus continuum model would only reproduce the qcd prediction in the asymptotic regime if the continuum @xmath79 was constrained by the lowest fesr .
hence , the normalization of the gaussian sum - rules , which is constrained by the fesr , should be removed by defining normalized ( unit - area ) gaussian sum - rules @xmath111 where the @xmath112-th moment of @xmath113 is given by @xmath114 note that , for the sake of notational convenience in subsequent sections , we have absorbed the low - energy theorem contribution into the definition of @xmath115 ( see ( [ tom_norm_srm1 ] ) above ) .
this allows us to write our final version of the normalized gaussian sum - rules of scalar gluonium as @xmath116 \frac{1}{\pi}\rho^{\text{had}}(t){\ensuremath { \mathrm{d}t } } } { \int_{t_0}^{\infty } t^k
\frac{1}{\pi}\rho^{\text{had}}(t ) } \quad,\quad k\geq -1\quad.\ ] ] before describing analysis methods for the gaussian sum - rules , we specify the numerical values for the qcd parameters appearing in ( [ gluecorr ] ) that will be employed in our analysis . for the dimension four gluon condensate
, we make the assumption that @xmath117 and then employ the most recently updated value @xcite @xmath118 the dimension six gluon condensate can be related to @xmath119 using instanton techniques ( see @xcite ) @xmath120 further , by invoking vacuum saturation in conjunction with the heavy quark expansion , the authors of @xcite have related the dimension eight gluon condensate to @xmath119 through @xmath121 regarding the instanton contributions , we shall employ shuryak s dilute instanton liquid model @xcite in which @xmath122 with @xmath123
in most sum - rules analyses , it is necessary to make some assumptions concerning the hadronic content of the spectral function ( [ phenom ] ) .
a common choice in the literature is that of a single narrow resonance model .
however , the wealth of scalar states with masses below @xmath124 @xcite , some of which are quite broad , certainly raises a question as to the validity of such an assumption , and suggest that a more general model allowing for a distribution of the resonance strength would be more suitable . in the following subsections , we show that this is indeed the case . using appropriate generalizations of the analysis techniques developed in @xcite , we demonstrate that the single narrow resonance model provides an inadequate description of the hadronic content of the scalar glueball correlator , whereas certain distributed strength models lead to outstanding agreement between the theory and phenomenology parameterized by ( [ ngsr ] ) . in the single narrow resonance model , @xmath82 takes the form @xmath125 where @xmath126 and @xmath54 are respectively the resonance mass and coupling .
with such an ansatz , the normalized gaussian sum - rule ( [ ngsr ] ) becomes @xmath127 \quad.\ ] ] the phenomenological side of ( [ phenom_single ] ) admits an absolute maximum ( peak ) located at @xmath128 , _ independent _ of @xmath89 ; therefore , the theoretical side of ( [ phenom_single ] ) should mimic this behaviour provided that the single narrow resonance model ( [ single ] ) is actually an adequate description of hadronic physics below the continuum threshold ( _ i.e. _ heavier states are weakly coupled enough to be absorbed into the continuum ) .
defining @xmath129 by the condition @xmath130 and denoting by @xmath131 an equally spaced partition of the @xmath89 interval of interest @xmath132 $ ] ( we elaborate on this interval shortly ) , we define the following @xmath133-function @xmath134 ^2\ ] ] as a measure of the difference between the theoretical peak position and the phenomenological peak position .
minimization of ( [ chi_single ] ) with respect to @xmath79 and @xmath135 then provides us with values for these two parameters which correspond to the best possible fit between the theory and phenomenology as represented through ( [ phenom_single ] ) .
lastly , the optimizing condition @xmath136 allows us to write @xmath135 as a function of @xmath79 whereby reducing ( [ chi_single ] ) to a one - dimensional minimization problem : @xmath137 ^2\ ] ] with @xmath138 thus , in a single narrow resonance analysis , we first minimize ( [ final_single_chi ] ) with respect to @xmath79 to determine an optimum choice for the continuum threshold parameter and then substitute this value into ( [ single_mass ] ) to obtain the best fit resonance mass . upon obtaining these optimized parameters
, there exists criteria that can be used to assess the validity of this phenomenological model . for instance , if the single resonance analysis is a reasonable approach , plots of the theoretical gaussian sum - rules ( the left - hand side of ( [ phenom_single ] ) ) and plots of the phenomenological gaussian sum - rules ( the right - hand side of ( [ phenom_single ] ) ) should coincide ( to a large degree ) .
significant deviation of one from the other may be indicative of an inadequate phenomenological model . on a more quantitative note ,
consider the following combination of moments ( [ moments ] ) ( where we suppress the explicit dependence on @xmath89 and @xmath79 ) : @xmath139 in the single narrow resonance model we should find @xmath140 and hence a substantial deviation of @xmath141 from zero signals a failure of the single narrow resonance model . to proceed with the analysis ,
however , we must first choose our region of interest @xmath142 $ ] needed in the definition of the @xmath133 function ( [ chi_single ] ) .
there are a number of factors to be considered in selecting this interval .
the lower bound @xmath143 must be large enough such that the condensate contributions do not dominate perturbation theory and also such that the leading omitted perturbative term in the expansion for the running coupling ( [ run_coupling ] ) is small .
therefore , in accordance with these criteria , we choose a lower bound of @xmath144 to choose an appropriate upper bound on @xmath89 , we first note that the gaussian kernel has a resolution of @xmath145 .
it is important to the analysis that the gaussian sum - rules employed have a resolution less than the non - perturbative ( hadronic physics ) energy scale involved : roughly 23 @xmath146 .
this fact motivates an upper bound of @xmath147 .
therefore , _ in this and in all subsequent analyses _ , we restrict our attention to the range @xmath148 .
an analysis of the @xmath4 normalized gaussian sum - rule through ( [ phenom_single ] ) leads to predictions that are completely unstable under qcd uncertainties . incorporating the error bounds of the dimension four gluonic condensate
( [ dimfour ] ) and allowing for a 15% error in each of @xmath149 and @xmath150 leads to a huge degree of variation in the resulting hadronic parameter estimates , and it is not even possible to ascertain whether or not an extension of the single resonance analysis is warranted .
mass predictions range anywhere from @xmath151 to @xmath152 , an interval far too broad to be of much use .
the only inference we can draw from @xmath4 analysis is that the mass scale obtained is in rough agreement with that which results from an analysis of the next - to - leading order ( @xmath6 ) gaussian sum - rule ( see below ) .
we note further that this consistency in mass scales between the @xmath4 and higher order sum - rules is also observed in @xcite and occurs only when instanton effects are included . due to its extreme sensitivity to small variations in various qcd parameters ,
we are forced to conclude that the @xmath4 gaussian sum - rule is an unreliable probe of the scalar glueball sector , and so , move on to a @xmath6 analysis .
a single narrow resonance analysis of the @xmath6 gaussian sum - rule , however , leads to rather poor agreement between theory and phenomenology .
minimization of ( [ final_single_chi ] ) leads to an optimum threshold at @xmath153 which , when substituted into ( [ single_mass ] ) , yields a resonance mass of @xmath154 . in figure
[ sigma ] , we plot both @xmath155 and @xmath156 versus @xmath89 for @xmath153 .
the graph of @xmath155 appears to have a slope of two , but , if extended to @xmath157 , would not pass through the origin a situation which contradicts the single resonance result ( [ b ] ) . further , in figure
[ sum_rules_single ] , we plot the left- ( theoretical ) and right- ( phenomenological ) hand sides of ( [ phenom_single ] ) versus @xmath105 for @xmath158 using the optimized values @xmath154 and @xmath153 .
the discrepancy between theory and phenomenology is apparent : the theoretical curves consistently underestimate phenomenology near the peak and overestimate phenomenology in the tails .
these observations indicate that a single narrow resonance ansatz is an inadequate description of the scalar gluonium spectral function , hence motivating our subsequent analyses of more general models .
the moment combination @xmath159 measures the width of the qcd distribution , and since @xmath160 , we conclude that the resonance strength is distributed over an energy region broad enough to be sampled by the resolution @xmath145 of the gaussian kernel . in accordance
, we should therefore consider phenomenological models which allow for such a spreading .
we focus on three distinct classes of distributed strength models : 1 .
single non - zero width models , 2 . a double narrow resonance model , and 3 . single narrow resonance + single non - zero width resonance models , ordered according to increasing complexity , _
i.e. _ the ( normalized ) single non - zero width models each contain two free parameters whereas the single narrow resonance + single non - zero width models requires four . in the subsections to follow
, we outline the analysis procedure and the resulting hadronic parameter predications corresponding to each of the models under consideration .
the first non - zero width model we consider is a unit - area square pulse representing a broad , structureless background which has previously been used for the study of the @xmath161 resonance @xcite . to describe such a square pulse centred at @xmath135 and having width @xmath162 , we define @xmath163\quad , \label{square_pulse}\ ] ] which leads to the following normalized gaussian sum - rule ( [ ngsr ] ) for @xmath6 : @xmath164\quad .
\label{gauss_sp}\ ] ] as in the single narrow resonance analysis , the phenomenological side of ( [ gauss_sp ] ) admits a single @xmath89-independent peak , and so we can determine an optimum threshold parameter @xmath79 by following the steps outlined through ( [ chi_single])([single_mass ] ) _ i.e. _ @xmath165 .
we wish to use this optimized value of @xmath79 to generate predictions for the hadronic parameters @xmath166 .
the following combinations of the moments ( [ moments ] ) ( where we suppress the explicit dependence on @xmath89 and @xmath79 ) are useful in this regard : @xmath167 in the spirit of the sum - rules approach , we equate each of these theoretical quantities to its corresponding phenomenological counterpart resulting in the following system of equations : @xmath168 considering that the right - hand ( phenomenological ) sides of ( [ square_params_a])([square_params_c ] ) are all @xmath89-independent , it is an important feature of the analysis that , at the optimized continuum threshold @xmath79 , the left - hand ( theoretical ) sides exhibit negligible dependence on @xmath89 and so are well approximated throughout the interval @xmath132 $ ] by averaged values .
such @xmath89-independence of appropriate residual moment combinations also occurs in the more complicated models considered below .
then , we can use ( [ square_params_a])([square_params_b ] ) to obtain predictions for the two free parameters @xmath169 of the square pulse model ( [ square_pulse ] ) .
equation ( [ square_params_c ] ) is independent of our parameter estimates and therefore serves as an _ a posteriori _ consistency test . ) or ( [ single_mass ] ) to compute the mass @xmath126 ; however , as we shall see , both yield the same result .
we concentrate on ( [ square_params_a ] ) simply because the analysis techniques outlined here are easily generalized to the more complicated resonance models to follow . ]
inverting ( [ square_params_a])([square_params_b ] ) , we obtain @xmath170 and @xmath171 , where , in this and in all subsequent analyses , the uncertainties quoted stem from the error bounds on the dimension - four gluonic condensate ( [ dimfour ] ) and an estimated 15% uncertainty in each of @xmath149 and @xmath150 ( see ( [ dilparams ] ) ) . in figure
[ square_fig ] , we use @xmath172 and central values of @xmath126 and @xmath173 to plot the theoretical and phenomenological sides of ( [ gauss_sp ] ) versus @xmath105 for @xmath174 .
the excellent agreement between theory and phenomenology demonstrated by these plots is a vast improvement over the results of the @xmath6 single narrow resonance analysis ( see figure [ sum_rules_single ] ) .
however , closer quantitative scrutiny reveals that the fits depicted in figure [ square_fig ] are not quite as accurate as one might think .
the quantity @xmath175 involves higher order moments of the @xmath6 gaussian sum - rule and so is sensitive to an even finer level of detail than are @xmath176 and @xmath177 .
a qcd calculation of this quantity yields @xmath178 in significant violation of our consistency check ( [ square_params_c ] ) . for our second example of a single non - zero width model
, we would ideally like to consider a breit - wigner resonance .
however , closed - form expressions do not exist for the gaussian image of a breit - wigner shape , and so we instead employ a gaussian model to describe a well - defined resonance peak with a non - zero width @xmath179}\quad . \label{gauss_res}\ ] ] the quantity @xmath180 is a normalization constant related to the total resonance strength , and the gaussian width @xmath173 is related to an equivalent breit - wigner width @xmath181 through @xmath182 in this model , the @xmath6 normalized gaussian sum - rule ( [ ngsr ] ) becomes @xmath183 } \exp\left[-\frac{\left(\hat{s}-m^2\right)^2}{2\left(\gamma^2 + 2\tau\right)}\right ] \quad , \label{phenom_gauss}\ ] ] where @xmath184 as in prior analyses , we begin with an extraction of the optimized continuum threshold parameter @xmath79 , and again , we accomplish this by examining the behaviour or the peak position .
the phenomenological side of ( [ phenom_gauss ] ) admits a single peak ; however , in contrast to our previous two analyses , _ the position of this peak is dependent on @xmath89 .
_ differentiating the right - hand side of ( [ phenom_gauss ] ) and setting the result to zero gives @xmath185 \label{gauss_constraint}\ ] ] which , unfortunately can not be explicitly solved for @xmath105 .
consequently , in the absence of an exact solution , we approximate the phenomenological peak position by the expression @xmath186 where @xmath187 are to be considered unknown parameters . explicit numerical experiments in ( realistic ) worst - case scenarios show that , provided @xmath188 , the next term in the expansion ( [ peak_approx ] ) [ _ i.e. _ @xmath189 is negligible and can safely be ignored .
therefore , we are led to define the following @xmath133-function as a measure of the deviation of the theoretical peak position ( [ peakpos ] ) from the phenomenological peak position characterized by the expansion ( [ peak_approx ] ) : @xmath190 ^ 2 \quad.\ ] ] the @xmath133 minimizing conditions @xmath191 can then be used to write @xmath187 as functions ) is linear and inhomogeneous . while trivial to obtain , the solution is rather a mess and so is omitted for brevity . ] of @xmath79 leaving us with a one - dimensional minimization problem in @xmath79 : @xmath192 ^ 2 \quad.\ ] ] minimizing ( [ oned ] ) with respect to @xmath79 furnishes us with an optimized choice for the continuum threshold parameter .
once the optimized @xmath79 has been determined , the subsequent analysis of the gaussian resonance model proceeds in a fashion completely analogous to the analysis of the square pulse model .
for instance , the appropriate gaussian model equations corresponding to ( [ square_params_a])([square_params_c ] ) are @xmath193 where @xmath194\quad.\ ] ] again , the first two equations may be ( numerically ) inverted to yield predictions for the model s two free parameters @xmath169 while the third serves as an independent consistency check .
the quantity @xmath195 is small , and hence the terms in ( [ gauss_m1][gauss_a ] ) proportional to @xmath195 have a negligible effect on the extracted resonance parameters , a property which also simplifies our subsequent analysis .
note that since @xmath196 in ( [ gauss_a ] ) , we anticipate that this model will still underestimate the qcd value of this moment combination , a result which is confirmed in our detailed analysis .
carrying out the appropriate sequence of calculations first gives an optimized continuum threshold parameter @xmath172 which , in turn , yields a mass of @xmath170 and an equivalent breit - wigner width ( see ( [ gamma_bw ] ) ) of @xmath197 .
again , we test the validity of these values by plotting both the theoretical and phenomenological sides of ( [ phenom_gauss ] ) versus @xmath105 for @xmath198 , setting @xmath172 and employing the central values of the resonance parameters . the resulting graphs are , to the naked eye , indistinguishable from those corresponding to the square pulse model analysis . ]
( figure [ square_fig ] ) and so are omitted .
substitution of the central values of @xmath126 and @xmath173 into ( [ gauss_a ] ) yields @xmath199 which must again be compared with the qcd value of @xmath178 .
clearly , there exists a significant discrepancy between the two as the gaussian model can not even correctly predict the sign of this moment combination .
the quantity @xmath175 , however , is a measure of the asymmetry of @xmath200 with respect to @xmath105 about its average value defined by @xmath176 .
therefore , considering that both the gaussian and square pulse models represent resonance strength distributions that are symmetric ( about @xmath135 ) , it is perhaps not surprising that they fail to accurately predict @xmath175 .
hence , we are prompted to consider a skewed generalization of the gaussian resonance model : @xmath201\ ] ] where the factor @xmath202 introduces a degree of asymmetry and has been chosen to achieve consistency with known low - energy two pion decay rates @xcite . substituting ( [ skewed_gauss ] ) into ( [ ngsr ] ) gives @xmath203 ( \hat{s}^2\gamma^4 + 4\hat{s}\gamma^2m^2\tau + 4m^4\tau^2 + 2\tau\gamma^4 + 4\tau^2\gamma^2 ) \left [ 1+{\ensuremath{\mathrm{erf}}}\left(\frac{\hat{s}\gamma^2 + 2\tau m^2}{2\gamma\sqrt{\tau}\sqrt{\gamma^2 + 2\tau}}\right)\right ] \\ &
+ 4\exp\left[-\frac{\hat{s}^2\gamma^2 + 2\tau m^4}{4\tau\gamma^2}\right ] \gamma\sqrt{\tau}\sqrt{\gamma^2 + 2\tau}(\hat{s}\gamma^2 + 2\tau m^2 ) \end{aligned}\ ] ] where @xmath204 + 2m^2\gamma\exp\left(-\frac{m^4}{2\gamma^2}\right)\sqrt{\pi}(\gamma^2 + 2\tau)^{5/2 } \quad .\ ] ] the sum - rules analysis proceeds along the same lines as for the unskewed gaussian resonance model
. the appropriate generalizations of ( [ gauss_m1])([gauss_a ] ) are @xmath205 numerically inverting)([skew_sigma ] ) proportional to @xmath195 are found to have a negligible impact on the solutions . ]
( [ skew_m1])([skew_sigma ] ) , we find a mass @xmath206 and an equivalent breit - wigner width of @xmath207 .
a subsequent phenomenological prediction yields @xmath208again , completely inaccurate compared with the qcd value .
these results , as well as the results of the unskewed non - zero resonance width model analyses are summarized for convenience in table [ width_tab ] .
.the results of a @xmath6 gaussian sum - rules analysis of a variety of non - zero resonance width models using central values of the qcd parameters . for the gaussian resonance models ,
the given width is actually the equivalent breit - wigner width ( see ( ) ) .
the @xmath175 values should be compared with the qcd prediction @xmath178 . [ cols="^,>,>,>",options="header " , ] thus the moment consistency test and @xmath133 measure of the agreement between the phenomenological and qcd values of the gaussian sum - rules clearly favour the double resonance scenarios .
however , this @xmath133 does not clearly distinguish between the various two - resonance scenarios . despite this apparent difficulty in distinguishing between the scenarios ,
the mass of the two states in the various double resonance scenarios are relatively stable , lying in the range @xmath209 , with the greatest mass splittings occurring in the double narrow resonance model and in the narrow plus skewed gaussian resonance model .
in this paper , we used qcd gaussian sum - rules to analyze the scalar glueball sector in an effort to obtain predictions for the hadronic parameters ( _ i.e. _ mass , width , coupling strength ) of low - lying scalar glueball states . in our analysis , we incorporated instanton effects and employed a number of phenomenological models more general than the traditional single narrow resonance .
first , we demonstrated that the leading order @xmath4 gaussian sum - rule ( the only sum - rule which receives a contribution from the let ) leads to results which are unstable under moderate qcd uncertainties : single resonance mass extractions ranged anywhere from 1.0 gev to 1.8 gev .
this variation is too large to yield a definitive prediction for the mass of the lightest scalar glueball , but does , however , indicate a lower bound of roughly 1 gev on gluonium mass scales .
this observation supports similar results obtained previously in @xcite where it was shown that instanton effects serve to increase the scale of masses extracted from the @xmath4 sum - rule whereby reducing the discrepancy between masses predicted using this @xmath4 sum - rule and those extracted from @xmath210 sum - rules . due to the instability of the @xmath4 gaussian sum - rule analysis , we turned to the @xmath6 sum - rule .
we showed that hadronic parameters extracted using a single narrow resonance model led to poor agreement between theory and phenomenology as indicated by the plots of figures [ sigma ] and [ sum_rules_single ] .
consequently , we then considered phenomenological models which allowed for resonance strength to be distributed over an appreciable energy range .
we focused on three single non - zero width resonance models ( see section [ analysis ] ) , and the results of all the corresponding analyses were indicative of a ( @xmath211 gev ) , wide ( @xmath212 gev ) resonance ( see table [ width_tab ] ) .
the coincidence between plots of the theoretical and phenomenological gaussian sum - rules ( characterized by figure [ square_fig ] ) represented a vast improvement over the corresponding graphs obtained from a single narrow resonance analysis ( figure [ sum_rules_single ] ) .
as indicated in table [ width_tab ] , all three models failed completely to predict the value of the first independent moment combination @xmath175 , prompting us to move on to analyses of phenomenological models that allow for a second resonance . regarding both a @xmath133 measure of the discrepancy between theoretical and phenomenological @xmath6 gaussian sum - rules and the ability to accurately predict higher order , independent moments , the double resonance models represented a marked improvement compared to the single non - zero width resonance models . resulting hadronic parameter estimates from the various models analyzed
are summarized in table [ res_summary_table ] .
estimates of the larger of the two masses ( @xmath126 ) were remarkably insensitive to the particular width model used , approximately 1.4 gev in all cases considered .
mass predictions for the lighter of the two resonances ( @xmath213 ) ranged from 1.0 gev in the double narrow and narrow plus skewed gaussian models to roughly 1.25 gev in the narrow plus square and narrow plus gaussian models
again , a surprising degree of stability considering the variety of width models employed
. further , in those models which allowed for a non - zero resonance width , we made no _ a priori _ assumptions as to which of the two resonances was wide ; however , the width was consistently found to be associated with the lighter of the two states and was relatively large in magnitude , yielding an equivalent breit - wigner width of @xmath214 in the gaussian resonance models .
these hadronic parameter estimates were subsequently employed in predictions of higher order , independent moment combinations .
the results of this analysis are summarized in table [ moms_summary_table ] .
the deviation of the phenomenological predictions from the values computed using qcd was typically 2050% , with the best agreement occurring in the skewed gaussian resonance model . while certainly not perfect , this represented a vast improvement over analogous calculations performed in the various single non - zero width resonance models ( see table [ width_tab ] ) .
finally , in table [ chi_table ] , we have collected the results of a @xmath133 measure of the difference between the theoretical @xmath6 gaussian sum - rules and the various phenomenological sum - rules corresponding to each of the resonance models considered . for the double resonance models , the values quoted were consistently an order of magnitude lower than those pertaining to the single non - zero width models .
therefore , in addition to our previous conclusion concerning the necessity for distributed resonance strength , we note further that a single non - zero width resonance model is insufficient to account for this spreading and that far better consistency between theory and phenomenology ( as encapsulated in the gaussian sum - rules of scalar gluonium ) is achieved by employing models which allow for a second resonance . although certainly relevant to the nature of the spectrum of scalar states with masses under 2 gev , it is difficult to make a direct comparison between our results and the entries of the pdg @xcite since the typical mass separation between the various observed scalar states is relatively small : on the order of 0.2 gev .
unfortunately , this is approximately the same magnitude as the uncertainties we calculate in our mass estimates .
these error bars are the direct result of uncertainties in the qcd parameters ( [ dimfour],[dilparams ] ) .
therefore , in light of these uncertainties , we refrain from drawing definitive conclusions concerning the nature of any specific scalar state . instead , we merely note that certain classification schemes proposed by minkowski and ochs @xcite and narison @xcite require a relatively light ( @xmath215 1 gev ) , wide state a situation completely consistent with our results .
furthermore , the persistent @xmath216 state we find may indeed have ramifications regarding the nature of either the @xmath9 or the @xmath10 @xcite .
shifman , z. phys . ; + p. pascual and r. tarrach , phys .
; + s. narison , z. phys .
; + c.a .
dominguez and n. paver , z. phys . ; + j. bordes , v. gimnez and j.a .
pearrocha , phys . lett .
; + j.l .
liu and d. liu , j. phys .
; + l.s .
kisslinger , j. gardner and c. vanderstraeten , phys .
lett . ; + tao huang , hong ying jin and ai - lin zhang , phys .
rev . .
a. belavin , a. polyakov , a. schwartz and y. tyupkin , phys .
lett . ; + g. t hooft , phys .
; + c.g .
callan , r. dashen and d. gross , phys .
shifman , a.i .
vainshtein and v.i .
zakharov , nucl .
phys . .
r. shankar , phys .
; + r.g .
moorhouse , m.r .
pennington , g.g .
ross , nucl .
phys . ; + k.g .
chetyrkin , n.v .
krasnikov , n.n .
tavkhelidze phys .
+ e.g. floratos , s. narison , e. de rafael , nucl .
phys . .
used to calculate the gaussian sum - rule defined by ( [ finish ] ) . the inner circular segment @xmath218 has a radius of @xmath219 , and the circular segments @xmath220
, @xmath221 and @xmath222 have a radius @xmath71 .
the wavy line on the negative real axis denotes the branch cut of @xmath217 , and the linear segments of the contour above and below the branch cut are denoted by @xmath223 .
the contour @xmath69 is that portion of @xmath61 ( see figure [ cont_fig ] ) which lies in the interior of a circle of radius @xmath71 centered at @xmath224 , and similarly for @xmath70 .
[ deformed ] ] for the theoretical prediction ( dotted curve ) compared with @xmath225 for the single - resonance model ( solid curve ) for the @xmath6 sum - rule using the @xmath133-optimized value of the continuum @xmath226 .
[ sigma ] ] with the single narrow resonance phenomenological model ( [ phenom_single ] ) using the @xmath133-optimized values of the resonance mass ( @xmath227 ) and continuum ( @xmath226 ) .
the @xmath89 values used for the three pairs of curves , from top to bottom in the figure , are respectively @xmath228 , @xmath229 , and @xmath230 .
the phenomenological model is consistently _ larger _ than the theoretical prediction near the peak , but is consistently _ smaller _ than the theoretical prediction in the tails .
central values of the qcd parameters have been used.[sum_rules_single ] ] with the phenomenological side of the normalized gaussian sum - rule ( [ ngsr ] ) in the single non - zero width resonance models and the double resonance models .
individual plots for each model are indistinguishable from those shown , and are hence not repeated as additional figures .
the @xmath133-optimized value of the continuum ( @xmath226 ) has been used to extract the resonance parameters as outlined in the text .
the @xmath89 values used for the three pairs of curves , from top to bottom in the figure , are respectively @xmath228 , @xmath229 , and @xmath230 .
note the almost perfect overlap between the theoretical prediction and the phenomenological models .
central values of the qcd parameters have been used . ]
|
although marginally more complicated than the traditional laplace sum - rules , gaussian sum - rules have the advantage of being able to probe excited and ground states with similar sensitivity .
gaussian sum - rule analysis techniques are applied to the problematic scalar glueball channel to determine masses , widths and relative resonance strengths of low - lying scalar glueball states contributing to the hadronic spectral function .
a feature of our analysis is the inclusion of instanton contributions to the scalar gluonic correlation function .
compared with the next - to - leading gaussian sum - rule , the analysis of the lowest - weighted sum - rule ( which contains a large scale - independent contribution from the low energy theorem ) is shown to be unreliable because of instability under qcd uncertainties .
however , the presence of instanton effects leads to approximately consistent mass scales in the lowest weighted and next - lowest weighted sum - rules .
the analysis of the next - to - leading sum - rule demonstrates that a single narrow resonance model does _ not _ provide an adequate description of the hadronic spectral function .
consequently , we consider a wide variety of phenomenological models which distribute resonance strength over a broad region some of which lead to excellent agreement between the theoretical prediction and phenomenological models .
including qcd uncertainties , our results indicate that the hadronic contributions to the spectral function stem from a pair of resonances with masses in the range 0.81.6 gev , with the lighter of the two potentially having a large width .
|
the question of non - local quantum correlations versus local realism , first raised in the famous epr paper @xcite , has held the interest of the physics community since .
j. s. bell @xcite showed that the predictions of quantum mechanics are incompatible with any model based on local realism . the pioneering experimental work of a. aspect
_ @xcite and others @xcite supports the predictions of quantum mechanics and contradicts local realism : bell inequalities applicable to the various experimental arrangements have been shown to be violated .
it should be mentioned that some aspects of the experimental setups have been criticized and questioned @xcite .
problems of experimental bias or enhancement of particular polarization states by detection systems were experimentally checked by t. haji - hassan _
_ @xcite and found absent . and more recently kwiat _ et al .
_ @xcite have proposed and described an experimental arrangement that overcomes shortcomings of previous experiments .
while experiments are still open to criticism , it is generally accepted that local realism is untenable . in this letter
we assume that in nature there exist non - local correlations , as predicted by quantum mechanics , and we address the following question : can an experimenter _ non - locally tamper _ with non - local correlations , without violating relativistic causality ? quantum mechanics predicts non - local correlations ; however , it does not provide an explanation " about what creates them .
several theoretical models go beyond quantum mechanics and propose to explain the phenomenon of non - local correlations via a superluminal `` communication link '' @xcite .
if one accepts the possibility of a communication link , then a natural next step would be to probe whether it is possible to tamper with this link and _ jam _ the superluminal communication @xcite .
up to now , the possibility of jamming non - local correlations has not received due consideration , perhaps because of a tacit assumption that such tampering necessarily violates relativistic causality .
( the expression _
relativistic causality _ is used here to denote the principle that information can not be transferred at speeds exceeding the speed of light . ) in this letter we show that jamming of non - local correlations can be consistent with relativistic causality .
our results are independent of the model used to describe how the non - local quantum correlations arise , that is , the nature of the superluminal communication link , and they apply to any jamming mechanism .
jamming might take many forms .
the following discussion does not discuss a mechanism for jamming ; rather , it defines the constraints that any jamming mechanism must obey in order to be consistent with relativistic causality . in order to derive and illustrate the constraints ,
it is convenient to consider a particular experimental arrangement which can be subjected to jamming @xcite .
we will consider an epr - bohm experimental arrangement to study pairs of spin-@xmath0 particles entangled in a singlet state @xcite .
spacelike separated spin measurements on these pairs allow a test of the bell inequalities .
suppose that two experimenters , alice and bob , perform the spin measurements .
one particle of each entangled pair arrives at alice s analyzing station and the other particle arrives at bob s .
when alice and bob get together and combine the results of their measurements , they will find violations of the bell inequalities , as predicted by quantum mechanics @xcite .
we now introduce a third experimenter , jim , the jammer , who has access to a jamming device which he can activate , at will , and tamper with the communication link between each entangled pair of particles .
his action is spacelike separated from the measurements of alice or bob or from both of them .
jamming acts at a distance to modify the correlations between the particles ; it disturbs the conditions which make possible the phenomenon of non - local quantum correlations .
therefore , the correlations measured jointly by alice and bob will not agree with the predictions of quantum mechanics .
jamming is truly non - local and can not be carried out within the framework of quantum mechanics .
for example , consider three systems , @xmath1 , @xmath2 and @xmath3 , in a quantum state @xmath4 .
let experimenters near @xmath1 and @xmath2 measure @xmath5 and @xmath6 , with eigenstates denoted by @xmath7 and @xmath8 , respectively .
the only freedom available to an experimenter near @xmath3 is the choice of what local operator @xmath9 to measure .
but the probabilities @xmath10 for outcomes @xmath11 and @xmath12 , @xmath13 are _ independent _ of the choice of operator @xmath9 .
thus no measurement on @xmath3 can affect the results of the measurements performed on @xmath1 and @xmath2 , even if the three systems have interacted in the past @xcite . in general
, jamming would allow jim to send superluminal signals
. the constraints that must be satisfied in order to insure that jim can not send superluminal signals are embodied in two conditions .
the first condition , the _ unary condition _ , a necessary but not sufficient condition , requires that jim not be able to send signals to alice or bob _ separately_. in effect this condition demands that alice and bob , separately , measure zero average spin along any axis .
explicitly , let @xmath14 and @xmath15 tally the number of spin - up and spin - down results , respectively , found by alice for a given axis . for the same axis ,
let @xmath16 tally , in the absence of jamming , the joint results of alice and bob .
the parameters @xmath17 and @xmath18 denote , respectively , the results ( @xmath19 or @xmath20 ) of the polarization measurements carried out by alice and bob .
let @xmath21 tally , in the presence of jamming , the corresponding polarization measurements carried out by alice and bob .
the unary condition imposes the following relations between @xmath16 and @xmath21 : @xmath22 a similar set of relations holds for the results @xmath23 and @xmath24 found by bob . hence regardless of
whether jim has activated the jamming device , alice and bob will find that the average spin projection along any axis tends to zero , and jim can not send superluminal signals , separately , to either alice or bob .
the unary condition allows a range of possibilities for the jammed correlations : from correlations which are only slightly different from those predicted by quantum mechanics , down to completely random correlations .
in particular , the unary condition allows conservation of angular momentum , _ i.e. _ perfect anticorrelation of spin components along any parallel axes .
as stated in the previous section , the unary condition is a necessary but not sufficient condition . for jamming to respect relativistic causality
, we must also restrict the relationships in space and time among the three events @xmath25 , @xmath26 and @xmath27 generated , respectively , by alice , bob and jim .
1 shows the geometry of three different configurations of an epr - bohm experimental setup along with the corresponding minkowski diagrams of the events @xmath25 , @xmath26 and @xmath27 . in the configuration shown in fig .
1(a ) , jamming is _ not _ permitted . here
alice and bob are in close proximity while jim is far away .
if jamming were permitted , alice and bob could immediately after jim activates the jamming device measure the spin projections of their respective particles and combine their results to determine the spin correlations .
they would find spin correlations differing from the predictions of quantum mechanics and infer that jim activated the jamming device .
the corresponding minkowski diagram , fig .
1(b ) , shows that the future light cones of @xmath25 and @xmath26 overlap , in part , outside the future light cone of @xmath27 . a light signal originating at @xmath27
can not reach this overlap region of @xmath25 and @xmath26 , where alice and bob can combine their results .
were jamming possible here , it would violate relativistic causality .
1(c ) shows a configuration that would also permit superluminal signalling : jim obtains the results of alice s measurements prior to deciding whether to activate the jamming device .
bob is far from both alice and jim .
the corresponding minkowski diagram , fig .
1(d ) , shows that @xmath25 precedes @xmath27 by a timelike interval and both @xmath25 and @xmath27 are spacelike separated from @xmath26 . since jim has access to alice s results
, he can send a superluminal signal to bob by _ selectively _ jamming : for instance , suppose jim activates the jamming device only when alice obtains the value @xmath28 for the projection of the spin of a particle .
bob will , then , find that the average spin component along a given axis does _ not _ tend to zero .
the preceding can be demonstrated by comparing the results of the spin measurements @xmath23 and @xmath29 , carried out by bob in the absence of jamming , eqs .
( 3 ) , and in the presence of selective jamming , eqs . ( 4 ) .
the notation previously defined is used in eqs .
( 3 - 4 ) .
@xmath30 @xmath31 hence the results obtained by bob in the presence of selective jamming will be different from those obtained in the absence of jamming unless @xmath32 and @xmath33 .
however , the latter requirements imply that jamming , in this configuration , can not have any discernible effect , i.e. jamming in this configuration is impossible . to eliminate configurations which allow violations of relativistic causality , as shown in fig .
1(a ) to fig .
1(d ) , we further restrict jamming by imposing a second condition , the _ binary condition_. the binary condition , which is manifestly covariant , demands that the overlap of the future light cones of @xmath25 and @xmath26 lie entirely within the future light cone of @xmath27 and therefore a light signal emanating from @xmath27 can reach the overlap region . the configuration shown in fig . 1(a ) and 1(b ) , which allows an overlap of the future light cones of @xmath25 and @xmath26 outside of the future light cone of @xmath27 , is therefore forbidden .
the configuration shown in fig 1(c ) and 1(d ) , a configuration for selective jamming , violates the unary condition and it is also disallowed by the [ fig : hyp ] binary condition . a configuration which satisfies the binary condition
is shown in fig .
1(e ) and 1(f ) . the constraints to which a jamming configuration must conform , in order not to violate relativistic causality , are embodied in the unary and binary conditions .
these conditions are manifestly lorentz invariant .
however , the time sequence of the events @xmath25 , @xmath26 and @xmath27 is not .
a time sequence @xmath25 , @xmath27 and @xmath26 in one lorentz frame may transform into @xmath26 , @xmath27 and @xmath25 in another lorentz frame . hence while one observer will claim that alice completed her measurements before jim activated his jamming mechanism and thus jim affected only the results of bob s measurements
, another observer will claim that bob carried out his measurements first and jim affected only alice s results .
similar situations are encountered in quantum mechanics where different observers in different lorentz frames will give conflicting interpretations of the same set of events .
for example , with respect to an entangled pair of particles in an epr - bohm experiment , the question of which observer caused the collapse of the entangled state has no lorentz - invariant answer @xcite .
if jamming is possible then one must accept the possibility of reversal of the _ cause - effect _ sequence @xcite ; however , the allowed configuration which satisfies the _ unary _ and _ binary _ conditions does not lead to contradictory causal loops , i.e. no _ effect _ can send a signal to its _
cause_. indeed , consider one jammer , @xmath34 , who acts on the correlations between two spacelike separated events , @xmath25 and @xmath26 .
we first recall that the unary condition precludes signalling to @xmath25 and @xmath26 , separately , by @xmath27 ; therefore , only the combined results of the measurements of @xmath25 and @xmath26 can reveal whether @xmath34 activated a jamming mechanism . in order to complete a contradictory casual loop
one must gather the results of the measurements of @xmath25 and @xmath26 into the past light cone of @xmath27 and then send a signal to @xmath27 , the _ cause_.
but the binary condition requires that the overlap of the future light cones of @xmath25 and @xmath26 be completely contained in the future light cone of @xmath27 , so the only place where information from @xmath25 and @xmath26 can be put together by means of ordinary signals is the future of @xmath27 .
one might suppose that other jammers , using their non - local action , could somehow transmit the information from @xmath25 and @xmath26 into the past light - cone of @xmath27 .
such a scheme would require at least two more jammers .
since these jammers must have access to the results of @xmath25 and @xmath26 , we place @xmath35 and @xmath36 ( generated by @xmath37 and @xmath38 ) at timelight separations , respectively , from @xmath25 and @xmath26 .
events @xmath25 and @xmath26 are spacelike separated from each other and from @xmath27 , so @xmath35 and @xmath36 will either be spacelike separated from @xmath27 or in its future light cone .
the cases of @xmath37 and @xmath38 are similar , so we discuss only @xmath37 ; however , the conclusions reached apply equally to @xmath37 and @xmath38 .
the jammer , @xmath37 , can communicate the results of @xmath25 by jamming or not jamming the non - local correlations between pairs of entangled particles measured at events @xmath39 and @xmath40 .
notice that in order to communicate the result of a single measurement done at @xmath25 , @xmath37 must jam ( or not jam ) an ensemble of epr pairs .
the result of a single measurement carried out at @xmath25 is recovered from the correlations determined from the combined measurements done at @xmath39 and @xmath40 . for the jammer @xmath37 to gather the information at @xmath25 into the past light cone of @xmath27
requires that both @xmath39 and @xmath40 lie in the past light cone of @xmath27 , i.e. @xmath27 lies in the overlap of the future light cones of @xmath39 and @xmath40 .
this requirement , however , is incompatible with the binary condition when applied to the triplet of events , @xmath39 , @xmath40 and @xmath35 , which requires that the overlap of @xmath39 and @xmath40 be contained within the future light cone of @xmath35 .
this , in turn , implies that @xmath27 will lie in the future light cone of @xmath35 , contradicting the assumption that @xmath35 is either spacelike separated from @xmath27 or in @xmath27 s future light cone .
consequently , at least one event @xmath39 or @xmath40 must be spacelike separated from @xmath27
. therefore the introduction of @xmath37 does not help to gather the results of @xmath25 into the past light cone of @xmath27 .
then , by induction , we find that no scheme to close a contradictory causal loop , by introducing any number of jammers , can succeed .
in quantum mechanics non - local correlations are well established ; however , these correlations can not be used to send superluminal signals . in this letter
we have raised the question of whether a form of non - locality beyond quantum mechanics
non - local tampering with quantum correlations could also respect relativistic causality .
we find that jamming configurations which obey two conditions the _ unary _ condition , which forbids superluminal signalling to either of two experimenters , and the _ binary _ condition , which restricts the space - time configuration of the two experimenters and the jammer respect relativistic causality .
for these configurations , the cause - effect sequence might not be preserved in all lorentz frames ; however , they do not lead to contradictory causal loops .
hence , we find that a stronger form of non - locality than that arising in quantum mechanics action at a distance rather than non - local correlations is consistent with relativistic causality .
@xcite the results presented in this letter are independent of the model used to describe the nature of the non - local correlations and apply to any jamming mechanism . experimental studies , to date , have not tested the possibility of jamming .
we suggest that current and projected epr - bohm experiments test the possibility of jamming in configurations consistent with the constraints derived in this letter .
the constraints on the jamming configuration , however , because of their generality , do not themselves suggest a preferred mechanism for carrying out the jamming procedure .
|
we present a possible scheme to tamper with non - local quantum correlations in a way that is consistent with relativistic causality , but goes beyond quantum mechanics . a non - local jamming " mechanism , operating within a certain space - time window , would not violate relativistic causality and would not lead to contradictory causal loops .
the results presented in this letter do not depend on any model of how quantum correlations arise and apply to any jamming mechanism .
15.0 cm -1.0 in -42pt plus 2 mm minus 2 mm taup-2263 - 95 + ` june 21 , 1995 +
|
we thank d.m . brink for useful discussions . this research was supported by the monbusho scholarship and the international scientific research program : joint research : contract number 09044051 from the japanese ministry of education , science and culture , and by the u.s .
department of energy under grant no .
de - fg03 - 00-er41132 .
w. reisdorf , f.p .
hessberger , k.d .
hildenbrand , s. hofmann , g. mnzenberg , k.h .
schmidt , j.h.r .
schneider , w.f.w .
schneider , k. smmerer , g. wirth , j.v .
kratz and k. schlitt , nucl .
a * 438 * , 212 ( 1985 ) .
fig . 2 + comparison of the experimental data and the results of the orientation average ( the dashed line ) and the exact coupled channels ( the solid line ) calculations for the @xmath32o+@xmath0sm fusion reaction .
the top and bottom panels are the fusion excitation function and the fusion barrier distribution , respectively .
the data are taken from ref .
|
in heavy - ion fusion reactions involving a well deformed nucleus , one often assumes that the orientation of the target nucleus does not change during the reaction .
we discuss the accuracy of this procedure by analyzing the excitation function of the fusion cross section and the fusion barrier distribution in the reactions of @xmath0sm target with various projectiles ranging from @xmath1c to @xmath2ar .
it is shown that the approximation gradually looses its accuracy with increasing charge product of the projectile and target nuclei because of the effects of finite excitation energy of the target nucleus .
the relevance of such inaccuracy in analyzing the experimental data is also discussed .
it is now well established that nuclear intrinsic degrees of freedom strongly influence the fusion cross section in heavy - ion reactions at energies near and below the coulomb barrier @xcite .
typical examples include the rotational excitation of a deformed target nucleus which leads to a large enhancement of the fusion cross section at low energies .
a characteristic in this case is that the excitation energy of the rotational motion is often much smaller than the curvature of the fusion barrier , which determines the time scale of the fusion process .
this is the case when , e.g. , one of the deformed rare earth nuclei or actinides is the target nucleus .
one then often calculates the fusion probability for each partial wave @xmath3 following @xcite @xmath4 where @xmath5 is the orientation of the deformed target , which will be specified later more precisely .
we call this the orientation average formula . the upper index @xmath6 stands for the sudden tunneling approximation .
@xmath7 is the fusion probability for a given orientation .
it is determined by solving a one - dimensional schrdinger equation for the relative distance between the projectile and target @xmath8 , where the potential consists of the nuclear and coulomb components given by @xmath9 } \label{nucl } \\ v_c(r,\theta)&=&\frac{z_pz_te^2}{r}+\sum_{\lambda } \left[\beta_{\lambda}+ \frac{2}{7}\sqrt\frac{5}{\pi } \beta_{\lambda}^2\,\delta_{\lambda,2}\right ] \frac{3z_pz_te^2}{2\lambda+1}\frac{r_t^\lambda}{r^{\lambda+1 } } y_{\lambda 0}(\theta ) \label{coul}\end{aligned}\ ] ] with the angle dependent radius of the target nucleus @xmath10 . \label{radius}\end{aligned}\ ] ] in these equations @xmath11 is the depth parameter of the nuclear potential between the projectile and target , @xmath12 ( @xmath13 ) and @xmath14 ( @xmath15 ) are the radius and the atomic number of the projectile ( target ) , @xmath16 is the deformation parameter of the target nucleus of multipolarity @xmath17 , and @xmath18 the surface diffuseness parameter .
the second order term is retained only for the quadrupole coupling in the coulomb interaction . as in almost all analyses of heavy - ion fusion reactions at sub - barrier energies ,
we adopt the no - coriolis approximation and work in the rotating coordinate frame @xcite , where the z - axis is taken to be parallel to the radial vector of the relative motion between the projectile and the target nuclei .
also , we assume an axial symmetry for the target nucleus .
the parameter @xmath5 is then the angle between the symmetry axis and the z - axis in the rotating frame .
eq .
( [ orient ] ) is quite appealing and provides a simple understanding of the large enhancement of the fusion cross section at sub - barrier energies .
it states that the deformation and the associated rotational excitation of the target nucleus result in a distributed fusion barriers , where some of them will be lower than that of the bare potential in the absence of these degrees of freedom .
the same idea of the orientation average is still correct if one replaces the gauss integral in eq .
( [ orient ] ) by an appropriate summation even when the rotational band is truncated at a certain angular momentum @xmath19 as is always the case in nuclear physics .
if only @xmath20 is present , the corresponding sum is given by the @xmath21 points gauss quadrature @xcite .
the orientation average procedure is exact only in the degenerate spectrum limit , i.e. , when the excitation energy of the rotational motion is zero , which is not the case in actual nuclei .
it is , therefore , important to examine the applicability of this formula before one makes a quantitative analysis of the experimental data .
a step towards this direction has been undertaken in ref .
@xcite , where an analytic formula has been derived to modify the fusion probability in the sudden tunneling approximation by taking the finite excitation energy of the rotational motion into account as @xmath22p^s_j(e ) \label{analy}\end{aligned}\ ] ] where @xmath23 is the excitation energy of the first excited 2@xmath24 state of the rotational band .
@xmath25 , @xmath26 being the curvature of the fusion barrier , is the tunneling time and @xmath27 is a measure of the strength of the channel coupling which causes the rotational excitation of the target nucleus during the collision . in ref .
@xcite , it was identified with the coupling strength at the fusion barrier @xmath28 .
eq .
( [ analy ] ) is useful to qualitatively discuss the conditions to apply the orientation average formula .
it clearly shows that the excitation energy of the rotational motion should be much smaller than the curvature of the fusion barrier .
it also shows that the coupling strength also governs the validity of the formula . despite these advantages ,
as we see later , eq .
( [ analy ] ) is not tolerable for quantitative discussions .
this is partly because eq .
( [ analy ] ) has been derived in a perturbation theory , and also by ignoring the radial dependence of the coupling form factor . in heavy - ion fusion reactions ,
the latter is a very crude approximation , because the nuclear and coulomb couplings interfere leading to a strong radial dependence of the coupling form factor@xcite . nevertheless , for future reference
, we show in table i the exponential factor in eq .
( [ analy ] ) , which we call the dissipation factor , for several systems to be examined later .
table ii lists the bare nuclear potential parameters we used and the coulomb barrier properties , where the radius parameter @xmath29 has been introduced as @xmath30 . in this paper
we compare the results of the orientation average formula with the numerical solutions of the corresponding coupled channels equations , which are obtained using the computer code ccfull@xcite by keeping the finite excitation energy , i.e. 0.082 mev , of the target nucleus @xmath0sm .
we call the latter as the exact coupled channels calculations .
we keep only the quadrupole deformation in the target , i.e. , @xmath20 , for simplicity .
all the projectiles are treated as inert .
we truncate the rotational band of @xmath0sm at @xmath19=20@xmath31 member for all reactions . according to the table of isotopes @xcite ,
the highest member which has been so far experimentally observed is 16@xmath24 .
we have confirmed that the comparison between the exact coupled - channels and the orientation average calculations does not almost change beyond @xmath19=12@xmath24 .
we also remark that the coupled - channels calculations almost converge at @xmath19=12@xmath31 member@xcite for the @xmath32o and @xmath0sm reactions , but higher levels introduce non - negligible effects for heavy projectiles . both the orientation average and the exact coupled channels calculations give the fusion cross section which is a monotonically increasing function of energy . in order to facilitate to see the accuracy of the orientation average formula , we plot in fig .
1 the ratio of the fusion cross section calculated by solving the exact coupled channels equations to that obtained by the orientation average formula .
the figure clearly shows that the deviation of the results of the orientation average formula from those of the exact calculations gradually increases with the charge product of the projectile and target nuclei .
this behaviour is consistent with table i , which has been obtained based on eq .
( [ analy ] ) , though the actual deviation is much larger than what eq .
( [ analy ] ) predicts .
we also note that the deviation is significant even for light projectiles at energies below the coulomb barrier .
an important question is whether this significantly affects the understanding of the mechanism of heavy - ion fusion reactions and the structural informations such as the deformation parameters extracted from the data analyses . in this connection , we compare in fig .
2 the experimental data of the excitation function of the fusion cross section and the fusion barrier distribution @xcite with the theoretical results calculated by the orientation average formula ( the dashed line ) and by the exact treatment of the corresponding coupled - channels equations ( the solid line ) for @xmath32o+@xmath0sm reaction .
the second derivative of the cross section times the bombarding energy has been calculated with the point difference method with the interval @xmath33 mev .
similar comparison is done in fig .
3 for @xmath2ar+@xmath0sm reaction .
the bare coulomb barrier is at @xmath34 mev and 127.57 mev for @xmath32o+@xmath0sm and @xmath2ar+@xmath0sm reactions , respectively . in these calculations
the deformation parameter @xmath35 of @xmath0sm was determined to be 0.32 by fitting the data of @xmath32o+@xmath0sm fusion reactions .
the same deformation parameter has been used for @xmath2ar+@xmath0sm reactions as well , though the effective optimum values can differ in two reactions .
fig .
2 shows that the difference between two theoretical lines is much smaller than their deviation from the experimental data .
this indicates that one can safely use the orientation average formula to data analyses of @xmath32o+@xmath0sm fusion reactions , even though its deviation from the exact calculation can be noticeable at low energies as shown in fig .
1 . the top panel of fig .
3 shows that the deviation between the exact and the orientation average calculations is not so drastic even for @xmath2ar+@xmath0sm reaction in this semilogarithmic plot . on the other hand ,
the situation is different for the fusion barrier distribution shown in the lower panel .
the difference between the two calculations can be more easily recognizable than the case for @xmath32o+@xmath0sm fusion reactions .
the present theoretical calculations still underestimate the fusion cross sections compared with the experimental data for the @xmath2ar+@xmath0sm reaction at low energies
. this will be partly because we ignored the projectile excitations , whose excitation energy is as small as 1.46 mev . in summary
, we have studied the accuracy of the orientation average formula for fusion cross sections between a spherical projectile and a well deformed target by comparing its results with those of the exact coupled channels calculations .
we found that the results of the orientation average formula significantly differ from those of the exact numerical solution of the corresponding coupled channels equations for systems with a large charge product of the projectile and target nuclei .
this suggests the necessity of the proper coupled - channels calculations beyond the orientation average formula for these systems in order to properly identify the role of various channel coupling effects and to extract reliable informations on nuclear structure , especially through the analysis of the fusion barrier distribution of high precision data .
unexpectedly , we observed a significant deviation of the orientation average formula from the exact calculations even for light projectiles in the energy region well below the coulomb barrier .
this deviation is , however , much smaller than the deviation of these two calculations from the experimental data . in this sense ,
the orientation average formula is safely applicable to light systems .
|
for a function @xmath12 and a subset @xmath13 , we write @xmath14 to denote @xmath15 .
the _ incidence vector _ of @xmath16 in @xmath17 is denoted by @xmath18 , i.e. @xmath19 is 1 for @xmath20 and 0 for @xmath21 ( usually @xmath22 will be clear from the context ) .
when @xmath16 is a multiset , @xmath19 denotes the number of occurrences of @xmath23 in @xmath16 . in an undirected graph @xmath5 ,
the sets of nodes and edges are denoted by @xmath24 and @xmath25 , respectively .
when @xmath5 is a directed graph , we speak of arcs rather than edges and write @xmath26 instead of @xmath25 .
a similar notation is used for paths , cycles , and etc . a _ walk _ in @xmath5
is meant to be a sequence @xmath27 , where each @xmath28 is an edge ( or arc ) and @xmath29 are its endnodes ; when @xmath5 is a digraph , @xmath28 is directed from @xmath30 to @xmath31 .
edge - simple ( or arc - simple ) walks are called _
paths_. we consider an undirected graph @xmath5 and a distinguished subset @xmath1 of nodes , called _ terminals_. nodes in @xmath32 are called _ inner_. a _ @xmath6-path _ is a path @xmath8 in @xmath5 whose endnodes are distinct terminals ; we usually assume that all the other nodes of @xmath8 are inner .
the set of @xmath6-paths is denoted by @xmath33 .
a _ multiflow _ is a function @xmath34 .
equivalently , one may think of @xmath7 as a collection @xmath35 ( for some @xmath36 ) , where the @xmath37 are @xmath6-paths and the @xmath38 are nonnegative rationals , called
_ weights _ of paths . sometimes ( e.g. , in @xcite ) such a multiflow @xmath7 is called _ free _ to emphasize that _ all _ pairs of distinct terminals are allowed to be connected by flows .
@xmath39 of @xmath7 is @xmath40 . for a node @xmath10 ,
define @xmath41 the function @xmath42 on @xmath24 is called the _ ( node ) load function_. let @xmath43 be a nonnegative integer function of _ node capacities_. we say that @xmath7 is _ feasible _ if @xmath44 for all @xmath45 .
suppose we are given , in addition , a function @xmath46 of _ node costs_. then the _ cost _ @xmath47 of a multiflow @xmath7 is the sum @xmath48 , where @xmath49 stands for the cost @xmath50 of a path @xmath8 . in this paper
we consider the following problem : @xmath51 given @xmath52 as above , find a multiflow @xmath7 of minimum possible cost @xmath47 among all feasible multiflows of maximum possible value .
when @xmath53 , ( @xmath51 ) turns into the undirected min - cost max - flow problem under node capacities and costs , having a variety of applications ; see , e.g. , @xcite .
it admits integer optimal primal and dual solutions @xcite . in the special case @xmath54
, we are looking simply for a feasible multiflow of maximum value .
such a problem has half - integer optimal primal and dual solutions , due to results of pap @xcite and vazirani @xcite , respectively .
also it is shown in @xcite that the problem is solvable in strongly polynomial time by using the ellipsoid method .
an edge - capacitated version of ( @xmath51 ) has been well studied . in this version , denoted by ( @xmath55 ) ,
@xmath56 and @xmath11 are functions on @xmath25 rather than @xmath24 . for a multiflow
@xmath7 , its _ edge load function _ is defined similarly to ( [ eq : node_load ] ) : @xmath57 and its cost is defined to be @xmath58 .
problem ( @xmath55 ) is reduced to ( @xmath51 ) by adding an auxiliary node on each edge , but no converse reduction is known .
an old result is that ( @xmath55 ) has a half - integer optimal solution @xcite .
also it is shown in @xcite that ( @xmath55 ) has a half - integer optimal dual solution and that half - integer primal and dual optimal solutions can be found in strongly polynomial time by using the ellipsoid method .
a `` purely combinatorial '' weakly polynomial algorithm , based on cost and capacity scaling , is devised in @xcite . in the special case of ( @xmath55 ) with @xmath54 ,
the half - integrality results are due to lovsz @xcite and cherkassky @xcite , and a strongly polynomial combinatorial algorithm is given in @xcite ( see also @xcite for faster algorithms ) .
in this paper we prove that ( @xmath51 ) always admits a half - integer optimal primal and dual solutions .
in particular , this implies all half - integrality results mentioned in the previous subsection .
similar to @xcite , we introduce a _ parametric _ generalization of ( @xmath51 ) , study properties of geodesics ( shortest @xmath6-paths with respect to some length function ) , and reduce the parametric problem to a certain single - commodity flow problem .
however , the details of this construction are more involved .
in particular , the reduced problem concerns integer flows in a _ bidirected _ graph . the second goal is to explore the complexity of ( @xmath51 ) .
we show that half - integer optimal primal and dual solutions to the parametric problem ( and therefore to ( @xmath51 ) ) can be found in strongly polynomial time by using the ellipsoid method .
instead of ( @xmath51 ) , it is convenient to consider a more general problem , namely : @xmath59 given @xmath52 as in ( @xmath51 ) and , in addition , @xmath60 , find a feasible multiflow @xmath7 maximizing the objective function @xmath61
. we will prove the following [ th : primal_halfint ] for any @xmath60 , problem ( @xmath59 ) has a half - integer optimal solution .
( note that @xmath62 for any multiflow @xmath7 and @xmath63 .
therefore , the optimality of a multiflow in the parametric problem preserves when both @xmath11 and @xmath64 are multiplied by the same positive factor @xmath65 .
this implies that the theorem is generalized to arbitrary @xmath66 and @xmath67 ( but keeping the integrality of @xmath56 ) .
however , we prefer to deal with integer - valued @xmath11 and @xmath64 in what follows . ) by standard linear programming arguments , ( @xmath51 ) and ( @xmath59 ) become equivalent when @xmath64 is large enough ( moreover , the existence of a half - integer optimal solution for ( @xmath59 ) easily implies that taking @xmath68 is sufficient ) .
problem ( @xmath59 ) can be viewed as a linear program with variables @xmath69 assigned to @xmath6-paths @xmath8 .
assign to a node @xmath45 a variable @xmath70 .
then the linear program dual to ( @xmath59 ) is : @xmath71 minimize @xmath72 provided that the following holds for every @xmath6-path @xmath8 : @xmath73 the above dual problem ( @xmath71 ) involves lengths of paths ( namely , @xmath74 and @xmath49 ) determined by `` lengths '' of nodes ( @xmath75 and @xmath11 , respectively ) .
it is useful to transform lengths of nodes into lengths of edges .
to do so , for @xmath76 , we define the function @xmath77 on @xmath25 by @xmath78 where @xmath79 if @xmath80 , and @xmath81 if @xmath82 .
this provides the correspondence @xmath83 ( where @xmath84 stands for @xmath85 , and @xmath86 for @xmath87 ) . for @xmath88 as above , define @xmath89 let @xmath90 denote the @xmath91-distance between vertices @xmath92 and @xmath10 , i.e. the minimum @xmath91-length @xmath93 of a @xmath92@xmath10 path @xmath8 in @xmath5 .
then , in view of ( [ eq : node_to_edge ] ) and ( [ eq : ell ] ) , the constraints in ( @xmath71 ) can be rewritten as @xmath94 by the linear programming duality theorem applied to ( @xmath59 ) and ( @xmath71 ) , a feasible multiflow @xmath7 and a function @xmath95 satisfying ( [ eq : dual_dist ] ) are optimal solutions to ( @xmath59 ) and ( @xmath71 ) , respectively , if and only if the following ( complementary slackness conditions ) hold : [ eq : cs_geodesic ] if @xmath8 is a @xmath6-path and @xmath96 , then @xmath97 ; in particular , @xmath8 is @xmath91-shortest ; [ eq : cs_saturated ] if @xmath45 and @xmath98 , then @xmath10 is _ saturated _ by @xmath7 , i.e. @xmath99 . in the rest of the paper , to simplify technical details , we will always assume that the input costs @xmath11 of all nodes are _
strictly positive_. then the edge lengths @xmath91 defined by ( [ eq : ell ] ) are strictly positive as well .
this assumption will not lead to loss of generality in essence , since the desired results for a nonnegative input cost function @xmath11 can be obtained by applying a perturbation technique in spirit of @xcite ( by replacing @xmath11 by an appropriate strictly positive cost function ) .
condition ( [ eq : cs_geodesic ] ) motivates the study of the structure of @xmath91-shortest @xmath6-paths in @xmath5 . to this aim , set @xmath100 .
a @xmath6-path @xmath8 such that @xmath101 is called an _
@xmath91-geodesic _ ( or just _ geodesic _ if @xmath91 is clear form the context ) .
when a multiflow @xmath7 in @xmath5 is given as a collection ( [ eq : mf_coll ] ) in which all paths @xmath37 are @xmath91-geodesics , we say that @xmath7 is an _
@xmath91-geodesic _ multiflow .
next we utilize one construction from @xcite , with minor changes .
consider a node @xmath45 .
define the _ potential _ @xmath102 to be the @xmath91-distance from @xmath10 to the nearest terminal , i.e. @xmath103 . set @xmath104 ( in particular , @xmath105 ) . for @xmath106 , define @xmath107 . also define @xmath108 .
we refer to @xmath109 as the _ zone _ of a terminal @xmath110 , and to @xmath111 as the set of _ central _ nodes ( w.r.t .
@xmath91 ) . the sets @xmath109 ( @xmath106 ) and @xmath111 are pairwise disjoint and
give a partition of @xmath112 .
the following subset of edges is of importance : @xmath113 one can see that the subgraph @xmath114 of @xmath5 contains all @xmath91-geodesics .
moreover , a straightforward examination shows that the structure of @xmath91-geodesics possesses the properties as expressed in the following lemma ( which is , in fact , a summary of claims 13 from @xcite and uses the strict positivity of @xmath91 ) .
[ lm : geodesics ] let @xmath8 be an @xmath91-geodesic running from @xmath110 to @xmath115 .
then @xmath8 is contained in @xmath116 and exactly one of the following takes place : 1 .
@xmath8 contains no central nodes and can be represented as the concatenation @xmath117 , where @xmath118 , @xmath119 , @xmath120 , and @xmath121 .
2 . @xmath8 contains exactly one central node @xmath122 and can be represented as the concatenation @xmath123 , where @xmath124 , @xmath125 , @xmath120 , and @xmath126 . in both cases , parts @xmath127 and @xmath128 are contained in the induced subgraphs @xmath129 $ ] and @xmath130 $ ] , respectively .
the potentials @xmath131 are strictly increasing as we traverse @xmath127 from @xmath132 to @xmath92 , and strictly decreasing as we traverse @xmath128 from @xmath10 to @xmath133 .
conversely , any @xmath6-path in @xmath116 obeying the above properties is an @xmath91-geodesic .
in this subsection we introduce an auxiliary bidirected graph , which will be the cornerstone of our approach both for proving half - integrality results and for providing a polynomial - time algorithm . given @xmath5 , @xmath6 , @xmath56 , @xmath11 and @xmath64 as above , let @xmath75 be an optimal solution to ( @xmath71 ) . form the edge lengths @xmath134 , the potential @xmath131 , the subgraph @xmath116 , and the sets @xmath109 ( @xmath106 ) and @xmath111 , as in subsection [ ssec : geodesics ] .
one may assume that @xmath135 ( since @xmath136 , by ( [ eq : dual_dist ] ) , and if @xmath137 then @xmath138 is an optimal solution to ( @xmath59 ) , by ( [ eq : cs_geodesic ] ) ) .
for further needs , we reset @xmath139 , making all node capacities even integers . now our goal is to prove the existence of an _ integer _ optimal multiflow @xmath7 in problem ( @xmath59 ) ( which is equivalent to proving the half - integrality w.r.t . the initial @xmath56 ) .
recall that in a _ bidirected _ graph ( or a _ bd - graph _ for short ) edges of three types are allowed : a usual directed edge , or an _ arc _ , that leaves one node and enters another one ; an edge directed _ from both _ of its ends ; and an edge directed _ to both _ of its ends ( cf .
when both ends of an edge coincide , the edge becomes a loop . for our purposes we admit no loop entering and leaving its end node simultaneously .
sometimes , to specify the direction of an edge @xmath140 at one or both of its ends , we will draw arrows above the corresponding node characters .
for example , we may write @xmath141 if @xmath23 is directed from @xmath92 to @xmath10 ( a usual arc ) , @xmath142 if @xmath23 leaves both @xmath143 , @xmath144 if @xmath23 enters both @xmath143 , and @xmath145 if @xmath23 leaves @xmath92 ( and either leaves or enters @xmath10 ) .
a _ walk _ in a bd - graph is an alternating sequence @xmath146 of nodes and edges such that each edge @xmath28 connects nodes @xmath30 and @xmath31 , and for @xmath147 , the edges @xmath148 form a _ transit pair _ at @xmath31 , which means that one of @xmath148 enters and the other leaves @xmath31 .
as before , an edge - simple walk is referred to as a _ path_. now we associate to @xmath116 a bd - graph @xmath149 with edge capacities @xmath150 , as follows ( see fig . [
fig : aux_graph ] for an illustration ) .
each noncentral node @xmath151 generates two nodes @xmath152 in @xmath149 .
they are connected by edge ( arc ) @xmath153 going from @xmath154 to @xmath155 and having the capacity equal to @xmath2 .
we say that @xmath153 _ inherits the capacity _ of the node @xmath10 . for @xmath106 , the set @xmath156 in @xmath149
is called the _ zone _ of @xmath132 , similar to @xmath109 in @xmath5 .
consider an edge @xmath157 .
let @xmath158 for some @xmath106 and assume for definiteness that @xmath159 ( note that @xmath160 implies @xmath161 ; this is where the strict positivity of the cost function @xmath11 is important ) .
then @xmath23 generates in @xmath149 an edge ( arc ) going from @xmath162 to @xmath154 , and we assign infinite capacity to it .
( by `` infinite capacity '' we mean a sufficiently large positive integer . )
now let @xmath118 and @xmath119 for distinct @xmath163 .
then @xmath23 generates an infinite capacity edge @xmath164 ( leaving both @xmath162 and @xmath155 ) .
the transformation of central nodes is less straightforward .
each @xmath165 generates in @xmath149 a so - called _ gadget _ , denoted by @xmath166 .
it consists of @xmath167 nodes ; they correspond to @xmath168 and the elements of @xmath6 and are denoted as @xmath169 and @xmath170 , @xmath106 .
the edges of @xmath166 are : a loop @xmath171 leaving @xmath169 ( twice ) and , for each @xmath172 , an edge @xmath173 going from @xmath170 to @xmath169 , called the @xmath132-_leg _ in the gadget . each edge in @xmath166 is endowed with the capacity equal to @xmath174 .
each gadget @xmath166 is connected to the remaining part of @xmath149 as follows . for each edge of the form @xmath175 in @xmath116 , we know that @xmath176 for some @xmath106 ( by the construction of @xmath116 )
. then @xmath175 generates an infinite capacity edge ( arc ) going from @xmath155 to @xmath170 . finally , we add to @xmath149 an extra node @xmath65 , regarding it as the _ source _ , and for each @xmath106 , draw an infinite capacity edge ( arc ) from @xmath65 to @xmath177 .
the obtained bd - graph @xmath149 captures information about the @xmath91-geodesics in @xmath5 .
namely , each @xmath91-geodesic @xmath8 going from @xmath132 to @xmath133 induces a unique closed @xmath65@xmath65 walk @xmath178 in @xmath149 .
the first and the last edges of @xmath178 are @xmath179 and @xmath180 , respectively . for a noncentral node @xmath10 in @xmath8 ,
@xmath178 traverses the edge @xmath181 .
an edge @xmath182 with @xmath183 inside a zone induces the edge @xmath184 in @xmath178 .
an edge @xmath185 connecting different zones ( if any ) induces the edge @xmath164 in @xmath178 .
finally , suppose @xmath8 traverses a central node @xmath122 and let @xmath186 be the edges of @xmath8 incident to @xmath168 .
by lemma [ lm : geodesics ] , @xmath118 and @xmath119 for some @xmath187 . then the sequence of nodes @xmath188 in @xmath8 generates the subpath in @xmath178 with the sequence of edges @xmath189 .
the resulting walk @xmath178 is edge - simple , so it is a closed path .
conversely , let @xmath190 be a ( nontrivial ) @xmath65@xmath65 walk in @xmath149 .
one can see that @xmath190 with @xmath65 removed is concatenated as @xmath191 , where @xmath192 is a directed path within a zone @xmath193 , @xmath194 is reverse to a directed path within a zone @xmath195 ( with possibly @xmath196 ) , and @xmath197 either ( i ) is formed by an edge @xmath164 connecting these zones ( in which case @xmath187 ) , or ( ii ) is the walk with the sequence of edges @xmath198 , for some central node @xmath168 of @xmath116 .
moreover , the image in @xmath5 of each of @xmath199 is an @xmath91-shortest path .
when @xmath196 happens in case ( ii ) , @xmath190 traverses the edge @xmath173 twice . in all other cases ,
@xmath190 is edge - simple and its image in @xmath5 is an @xmath91-shortest @xmath6-path ( a @xmath64-geodesic ) .
these observations show that there is a natural bijection between the @xmath91-geodesics in @xmath5 and the ( nontrivial ) @xmath65@xmath65 paths in @xmath149 .
we will refer to the bd - graph @xmath149 described above as the _ compact _ bd - graph related to @xmath116 ; it will be essentially used to devise an efficient algorithm for solving ( @xmath59 ) in section [ sec : primal_alg ] .
besides , in the proof of the primal integrality ( with @xmath56 even ) in section [ sec : prim_h_int ] , we will deal with a modified bd - graph .
it is obtained from @xmath149 as above by replicating each gadget @xmath166 into @xmath174 copies @xmath200 , @xmath201 , called the _ 1-gadgets _ generated by @xmath168 . more precisely , to construct @xmath200 , we make @xmath202-th copy @xmath203 of the node @xmath169 , @xmath202-th copy @xmath204 of the loop @xmath171 leaving @xmath203 ( twice ) , and @xmath202-th copy @xmath205 of each leg @xmath173 , @xmath110 , where @xmath205 goes from @xmath170 to @xmath203 ( so @xmath170 , @xmath110 , are the common nodes of the created 1-gadgets ) .
all edges in these 1-gadgets are endowed with _
unit _ capacities .
we keep notation @xmath149 for the constructed graph and call it the _ expensive _ bd - graph related to @xmath116 .
also we keep notation @xmath56 for the edge capacities in @xmath149 .
there is a natural relationship between the @xmath65@xmath65 walks ( paths ) in both versions of @xmath149 .
the 1-gadgets created from the same central node @xmath168 of @xmath116 are isomorphic , and for any @xmath206 , there is an automorphism of @xmath149 which swaps @xmath203 and @xmath207 and is invariant on the other nodes .
let @xmath208 be a bidirected graph .
like in usual digraphs , @xmath209 and @xmath210 denote the sets of edges in @xmath208 entering and leaving @xmath211 , respectively . a loop @xmath23 at @xmath10 , if any , is counted twice in @xmath209 if @xmath23 enters @xmath10 , and twice in @xmath210 if @xmath23 leaves @xmath10 ; hence @xmath209 and @xmath210 are actually multisets .
( recall that we do not allow a loop which simultaneously enters and leaves a node . )
let @xmath65 be a distinguished node with @xmath212 in @xmath208 ( the _ source _ ) and let the edges of @xmath208 have integer capacities @xmath213 .
bidirected @xmath65-flow _ , or a _ bd - flow _ for short , is a function @xmath214 satisfying @xmath215 for all nodes @xmath216 ; and the _ value _ of @xmath217 is defined to be @xmath218 ( cf .
@xcite ) . here
@xmath219 is the _ divergence _ of @xmath217 at @xmath10 .
note that if @xmath23 is a loop at @xmath10 then @xmath23 contributes @xmath220 in @xmath221 .
if @xmath222 for all @xmath223 then @xmath217 is called _ feasible_. in addition , if @xmath217 is integer - valued on all edges then we refer to @xmath217 as an integer bidirected @xmath65-flow , or an _ ibd - flow_. one can see that finding a fractional ( resp .
integer ) bd - flow of the maximum value is equivalent to constructing a maximum fractional ( resp .
integer ) packing of closed @xmath65@xmath65 walks ( they leave @xmath65 twice ) .
return to an optimal solution @xmath75 to ( @xmath71 ) , and let @xmath224 . consider the ( expensive or compact ) bd - graph @xmath149 related to @xmath116 , and the capacity function @xmath56 on the edges of @xmath149 ( constructed from the node capacities @xmath56 of @xmath5 ) .
the above correspondence between @xmath91-geodesics in @xmath5 and @xmath65@xmath65 paths in @xmath149 is extended to @xmath91-geodesic multiflows in @xmath5 and certain @xmath65-flows in @xmath149 ( where @xmath65 is the source in @xmath149 as before ) .
more precisely , let @xmath7 be a ( fractional ) @xmath91-geodesic multiflow in @xmath5 represented by a collection of @xmath91-geodesics @xmath37 and weights @xmath225 , @xmath226 ( cf .
( [ eq : mf_coll ] ) ) .
then each @xmath37 determines a @xmath65@xmath65 path @xmath227 in @xmath149 , and @xmath228 is a bd - flow in @xmath149 ; we say that @xmath217 is _ generated _ by @xmath7 ( note that @xmath229 ) .
furthermore , @xmath217 is feasible if @xmath7 is such , and for each central node @xmath230 , the following relations hold : @xmath231 considering an arbitrary bd - flow @xmath217 in @xmath149 , we say that @xmath217 is _ good _ if it satisfies ( [ eq : good ] ) for all @xmath230 ( here the second relation in ( [ eq : good ] ) is important , while the first one obviously holds for any bd - flow ) .
the following assertion is of use .
[ lm : f - f ] let @xmath217 be a good bd - flow in @xmath149
. then @xmath217 is generated by an @xmath91-geodesic multiflow @xmath7 in @xmath5 .
moreover , if @xmath217 is integral , then it is generated by an integer @xmath91-geodesic multiflow @xmath7 . in both cases , @xmath7 can be found in @xmath232 time .
suppose there is a central node @xmath230 such that @xmath233 .
let us say that @xmath234 _ dominates _ at @xmath168 ( w.r.t .
@xmath217 ) if @xmath235 .
from ( [ eq : good ] ) it follows that there exist distinct @xmath236 such that @xmath237 and none of @xmath238 dominates at @xmath168 .
choose such @xmath239 .
build in @xmath149 a maximal walk @xmath190 starting with @xmath240 and such that @xmath241 for all edges @xmath23 of @xmath190 .
it is easily seen from the construction of @xmath149 that @xmath190 is edge - simple , terminates at @xmath65 , and have all vertices in @xmath193 , except for @xmath242 .
build a similar walk ( path ) @xmath197 starting with @xmath243 .
then the concatenation of the reverse to @xmath190 , the loop @xmath171 and the path @xmath197 is a @xmath65@xmath65 path and its image @xmath8 in @xmath5 is an @xmath91-geodesic ( from @xmath132 to @xmath133 ) .
assign the weight of @xmath8 to be the maximum number @xmath244 subject to two conditions : ( i ) @xmath245 for each @xmath246 , and ( ii ) the flow @xmath247 is still good . if @xmath244 is determined by ( i )
, we have @xmath248 ( where @xmath249 ) , whereas if @xmath244 is determined by ( ii ) , there appears @xmath234 dominating at @xmath168 ( w.r.t .
@xmath250 ) . if @xmath251 , repeat the procedure for @xmath250 and @xmath168 , otherwise apply the procedure to @xmath250 and another @xmath252 , and so on .
( note that if , in the process of handling @xmath168 , a current weight @xmath244 is determined by ( ii ) , then the weights of all subsequent paths through @xmath171 are already determined by ( i ) ; this will provide the desired complexity . )
eventually , we come to a good flow @xmath253 with @xmath254 for all @xmath230 . this @xmath253 is decomposed into a sum of flows along @xmath65@xmath65 paths in a straightforward way , in @xmath232 time ( like for usual flows in digraphs ) .
taking together the images in @xmath5 of the constructed weighted @xmath65@xmath65 paths , we obtain a required @xmath91-geodesic multiflow @xmath7 .
the running time of the whole process is @xmath232 , and if @xmath217 is integral , then the weights @xmath244 of all paths are integral as well .
( integrality of a current weight @xmath244 subject to integrality of a current flow @xmath217 is obvious when @xmath244 is determined by ( i ) , and follows from the fact , implied by ( [ eq : good ] ) , that for any @xmath234 , @xmath255 is even , when @xmath244 is determined by ( ii ) . )
[ rem : compact_vs_expensive ] in case of an expensive bd - graph @xmath149 , any feasible ibd - flow @xmath217 is good . indeed , for any 1-gadget @xmath200 in @xmath149
, we have @xmath256 and , therefore , @xmath257 . the second relation in ( [ eq : good ] )
is trivial when @xmath258 , and follows from the constraints @xmath259 ( @xmath110 ) when @xmath260 .
define the following subset of edges in @xmath149 : @xmath261 for an optimal ( possibly fractional ) solution @xmath7 to ( @xmath59 ) and a node @xmath4 with @xmath262 , we have @xmath99 ( by ( [ eq : cs_saturated ] ) ) ; so the edge @xmath153 of @xmath149 corresponding to @xmath10 is saturated by the bd - flow @xmath217 generated by @xmath7 , i.e. @xmath263 .
we call the edges in @xmath264 _ locked_. thus , the graph @xmath149 admits a ( fractional ) good feasible bd - flow saturating the locked edges .
the following strengthening is crucial .
[ prop : aux_bidir_flow ] there exists a good feasible ibd - flow in @xmath149 that saturates all locked edges .
a proof of this proposition involves an additional graph - theoretic machinery and will be given in section [ sec : prim_h_int ] .
assuming its validity , we immediately obtain theorem [ th : primal_halfint ] from lemma [ lm : f - f ] .
in this section we devise a strongly polynomial algorithm for solving the primal parametric problem ( @xmath59 ) . as before
, we assume that @xmath11 and @xmath64 are integral and that the node capacities @xmath56 are even , so our goal is to find an integer optimal multiflow .
the algorithm starts with computing a ( fractional ) optimal dual solution @xmath75 and constructing the bd - graph @xmath149 w.r.t .
the length function @xmath265 .
then it finds a good ibd - flow @xmath217 in @xmath149 saturating the locked edges ( assuming validity of proposition [ prop : aux_bidir_flow ] ) . applying the efficient procedure as in the proof of lemma [ lm : f - f ] to decompose @xmath217 into a collection of paths with integer weights , we will obtain an integer optimal solution to ( @xmath59 ) . to provide the desired complexity
, we shall work with @xmath149 given in the _ compact _ form ( defined in subsection [ ssec : aux_graph ] ) .
the core of our method consists in finding the load function of some integer optimal multiflow @xmath7 in @xmath5 ( without explicitly computing @xmath7 itself ) .
this function will just generate the desired ibd - flow in @xmath149 .
we describe the stages of the algorithm in the subsections below .
problem ( @xmath71 ) straightforwardly reduces to a `` compact '' linear program , as follows . besides variables @xmath266 ( @xmath4 ) , assign a variable @xmath267 to each terminal @xmath106 and node @xmath10 ( a sort of `` distance '' of @xmath10 from @xmath132 ) .
consider the following problem ( where @xmath268 and @xmath269 are defined according to ( [ eq : wbar ] ) ) : [ eq : compact_dualpprob ] minimize @xmath72 subject to the following constraints : @xmath270 programs ( @xmath71 ) and ( [ eq : compact_dualpprob ] ) are equivalent . indeed ,
if @xmath271 is a feasible solution to ( [ eq : compact_dualpprob ] ) then , obviously , @xmath75 is a feasible solution to ( @xmath71 ) .
conversely , let @xmath75 be a feasible solution to ( @xmath71 ) . for @xmath4 and @xmath110 ,
define @xmath272 , where @xmath134 .
it is easy to check that @xmath271 is a feasible solution to ( [ eq : compact_dualpprob ] ) .
the size of the constraint matrix in ( [ eq : compact_dualpprob ] ) ( written in binary notation ) is polynomial in @xmath273 .
therefore , ( @xmath71 ) is solvable in strongly polynomial time by use of tardos s version of the ellipsoid method .
( this remains valid when @xmath11 and @xmath64 are nonnegative rational numbers . ) the following fact is of importance .
[ lm : node_load ] one can find , in strongly polynomial time , a function @xmath274 such that @xmath275 for all @xmath45 , where @xmath7 is some integer optimal multiflow in ( @xmath59 ) .
we explain that in order to construct the desired @xmath276 , it suffices to compare two optimal objective values : one for the original ( integer ) costs @xmath11 and the other for certain perturbed costs @xmath277 .
these values are computed by solving the corresponding dual problems by the method described in the previous subsection .
more precisely , let @xmath278 be the nodes of @xmath5 .
let @xmath279 , define @xmath280 , @xmath281 , and define the cost function @xmath277 on @xmath24 to be @xmath282 .
then for any integer feasible multiflow @xmath7 , we have @xmath283 this and the fact that @xmath284 is an integer ( as @xmath285 are integral ) imply that if @xmath7 is optimal for @xmath277 , then @xmath7 is optimal for @xmath11 as well .
( an integer optimal multiflow for even capacities @xmath56 exists by theorem [ th : primal_halfint ] . ) moreover , for such an optimal @xmath7 , the number @xmath286 is computed in strongly polynomial time , since it is equal to @xmath287 , where @xmath75 and @xmath288 are optimal dual solutions for @xmath11 and @xmath277 , respectively . here
we use the lp duality equalities @xmath289 and @xmath290 .
also the size of binary encoding of @xmath277 is bounded by that of @xmath11 times a polynomial in @xmath291 , so the dual problem with @xmath277 is solved in strongly polynomial time w.r.t . the original data .
hence , we have @xmath292 .
the number @xmath293 is an integer and , in view of @xmath294 for each @xmath202 , the @xmath291 coefficients in its base @xmath295 decomposition ( the representation via degrees of @xmath295 ) are just @xmath296 , thus giving @xmath276 . recall that together with a node load function each multiflow @xmath7 also induces its edge counterpart ( see ( [ eq : edge_load ] ) ) .
lemma [ lm : node_load ] can be strengthened as follows .
[ lm : full_load ] one can find , in strongly polynomial time , a function @xmath297 such that @xmath275 for all @xmath45 and @xmath298 for all @xmath299 , where @xmath7 is some integer optimal multiflow @xmath7 in ( @xmath59 ) . split each edge @xmath300 of @xmath5 into two edges @xmath301 in series and assign to each new node @xmath302 the capacity @xmath303 and the cost @xmath304 .
clearly this transformation does not affect the problem in essence .
the node load function , which can be found in strongly polynomial time by lemma [ lm : node_load ] , yields the desired node and edge load functions in the original graph @xmath5 .
now we explain how to find , in strongly polynomial time , an integer optimal multiflow solving ( @xmath59 ) ( for a graph @xmath5 , even node capacites @xmath56 , rational node costs @xmath11 , and an integer parameter @xmath64 ) by using an optimal dual solution @xmath75 and a function @xmath276 as in lemma [ lm : full_load ] . for this @xmath276
, there exists an integer @xmath91-geodesic multiflow @xmath305 in @xmath5 satisfying @xmath306 for all @xmath45 and @xmath307 for all @xmath299 , where @xmath308 .
our goal is to construct one of such multiflows explicitly . to do this , we consider the _ compact _
bd - graph @xmath149 related to @xmath116 ( see subsection [ ssec : aux_graph ] ) and put @xmath217 to be the function on @xmath309 corresponding to @xmath276 .
more precisely , let @xmath310 be the subset of edges of @xmath149 neither incident to @xmath65 nor contained in the gadgets @xmath166 ( @xmath122 ) . by the construction of @xmath149 , there is a natural bijection @xmath311 of @xmath310 to the set @xmath312 .
for each @xmath313 , we set @xmath314 . in their turn , the values of @xmath217 on the edges of a gadget @xmath166 are assigned as follows : for the loop @xmath171 at @xmath169 , set @xmath315 , and for each @xmath110 , set @xmath316 , where @xmath317 is the set of edges in @xmath116 connecting @xmath109 and @xmath168 . finally , for each @xmath110 , we set @xmath318 .
using the fact that the function @xmath276 on @xmath319 is determined by some integer optimal ( @xmath91-geodesic ) multiflow @xmath305 , it follows that the obtained function @xmath217 on @xmath309 is integer - valued and has zero divergency at all nodes different from @xmath65 .
so @xmath217 is an ibd - flow in @xmath149 .
moreover , @xmath217 is generated by @xmath305 as above ; in particular , @xmath217 is good ( i.e. satisfies ( [ eq : good ] ) ) .
by lemma [ lm : f - f ] , we can find , in strongly polynomial time , an integer @xmath91-geodesic multiflow @xmath7 generating @xmath217 .
then @xmath7 and @xmath305 have the coinciding node and edge load functions , and the optimality of @xmath305 implies that @xmath7 is an integer optimal solution to ( @xmath59 ) as well , as required .
to complete the proof of theorem [ th : primal_halfint ] it remains to prove proposition [ prop : aux_bidir_flow ] , which claims the existence of an ibd - flow saturating the `` locked '' edges .
we eliminate the lower capacity constraints ( induced by the locked edges ) by reducing the claim to the existence of an ibd - flow with a certain prescribed value .
let @xmath208 be a bidirected graph with a distinguished source @xmath65 and edge capacities @xmath320 , as described in subsection [ ssec : bidir_flows ] .
the classic max - flow min - cut theorem states that the maximum flow value is equal to the minimum cut capacity .
a bidirected version of this theorem involves a somewhat more complicated object , called an _
odd barrier_. in this subsection we give its definition and state the crucial properties ( in theorems [ th : max_ibd_flow_min_odd_barrier],[th : bd_residual],[th : bd_barrier_from_flow ] ) that will be used in the upcoming proof of proposition [ prop : aux_bidir_flow ] .
these properties are nothing else than translations , to the language of bidirected graphs , of corresponding properties established for integer symmetric flows in skew - symmetric graphs , as we will explain in the appendix .
let @xmath321 .
the _ flip _ at ( the nodes of ) @xmath22 modifies @xmath208 as follows : for each node @xmath322 and each edge @xmath23 incident to @xmath10 , we reverse the direction of @xmath23 at @xmath10 ( while preserving the directions of edges at nodes in @xmath323 ) .
for example , if @xmath324 and @xmath325 then @xmath23 becomes @xmath144 , and if @xmath326 and @xmath327 then @xmath23 becomes @xmath328 .
bd - graphs @xmath208 and @xmath329 are called _ equivalent _ if one is obtained by a flip from the other .
note that flips do not affect bidirected walks in @xmath208 in essence and do not change the maximum value of an ibd - flow in it .
we will essentially use flips to simplify requirements in the definition of odd barriers below .
next , we employ a special notation to designate certain subsets of edges . for @xmath330 ,
let @xmath331 $ ] denote the set of edges with one endpoint in @xmath22 and the other in @xmath332 .
we will often add arrows above @xmath22 and/or @xmath332 to indicate the subset of edges in @xmath331 $ ] directed in one or another way .
for example , @xmath333 $ ] denotes the set of edges that enter both @xmath22 and @xmath332 , @xmath334 $ ] denotes the set of edges leaving @xmath22 and having the other endpoint in @xmath332 ( where the direction is arbitrary ) , and @xmath335 $ ] denotes the set of edges that leave @xmath22 at both endpoints ( including twice leaving loops ) .
when @xmath336 , the second term in the brackets may be omitted : @xmath337 $ ] , @xmath338 $ ] , and @xmath339 $ ] stand for @xmath340 $ ] , @xmath341 $ ] , and @xmath342 $ ] , respectively .
finally , for a function @xmath343 on the edges , we write @xmath344 $ ] ( rather than @xmath345)$ ] ) for @xmath346 } { \varphi}(e)$ ] . ) . ]
a tuple @xmath347 , where @xmath329 is some bd - graph equivalent to @xmath208 , is called an _
odd barrier _ for @xmath208 if the following conditions hold with respect to @xmath329 ( see fig .
[ fig : bd_barrier ] ) : [ eq : bd_odd_barrier ] * @xmath348 give a partition of @xmath349 , and @xmath350 .
* for each @xmath351 , @xmath352 $ ] is odd . * for distinct @xmath353 , @xmath354 = 0 $ ] .
* for each @xmath351 , @xmath355= 0 $ ] .
the _ capacity _ of @xmath356 is defined to be @xmath357 + c[{\overrightarrow{a } } ] - k.\ ] ] [ th : max_ibd_flow_min_odd_barrier ] for @xmath358 as above , the maximum ibd - flow value is equal to the minimum odd barrier capacity . a ( feasible ) ibd - flow @xmath276 and an odd barrier @xmath359 for @xmath208 have maximum value and minimum capacity , respectively , if and only if the following conditions hold with respect to @xmath329 : * @xmath360 = c[{\overrightarrow{a } } , { \overleftarrow}a]$ ] and @xmath361 = 0 $ ] ; * @xmath362 = c[{\overrightarrow{a}},m]$ ] and @xmath363 = 0 $ ] ; * for each @xmath351 , either @xmath364 = c[{\overrightarrow{a}},b_i ] - 1 $ ] and @xmath365 = 0 $ ] , or @xmath364 = c[{\overrightarrow{a}},b_i]$ ] and @xmath366 = 1 $ ] . note that there may exist many minimum capacity odd barriers for @xmath367 .
it is well - known that in a usual edge - capacitated digraph with a source @xmath132 and a sink @xmath133 , the set of nodes reachable by paths from @xmath132 in the residual digraph of a maximum @xmath132@xmath133 flow @xmath217 determines a minimum capacity @xmath132@xmath133 cut .
moreover , this minimum cut does not depend on the choice of @xmath217 and , therefore , may be regarded as the _ canonical _ one .
a similar phenomenon takes place for maximum ibd - flows and minimum odd barriers ( and we will essentially use this in the proof of proposition [ prop : aux_bidir_flow ] ) . to describe this ,
consider an ibd - flow @xmath276 in @xmath208 .
the _ residual _ bd - graph @xmath368 endowed with the _ residual _
capacity @xmath369 is constructed in a similar way as for usual flows .
more precisely , @xmath370 and the edges of @xmath368 are as follows : [ eq : residual_edges ] * each edge @xmath371 with @xmath372 whose residual capacity is defined to be @xmath373 , and * the _ reverse _ edge @xmath374 to each edge @xmath371 with @xmath375 ; the directions of @xmath374 at the endpoints are reverse to those of @xmath23 and the residual capacity is @xmath376 .
a bidirected walk @xmath8 in @xmath368 is called _ @xmath377-simple _ if @xmath8 passes each edge @xmath23 at most @xmath378 times .
if @xmath8 is a @xmath377-simple closed @xmath65@xmath65 walk leaving its end @xmath65 twice , we can increase the value of @xmath276 in @xmath208 by 2 , by sending one unit of flow along @xmath8 .
so the existence ( in @xmath368 ) of such a walk @xmath8 , which is called _
@xmath379-residual _ , implies that @xmath276 is not maximum .
a converse property holds as well .
[ th : bd_residual ] an ibd - flow @xmath276 in @xmath208 is maximum if and only if there is no @xmath379-residual walk . when we are given a maximum ibd - flow @xmath276 , a certain minimum odd barrier can be constructed by considering the residual graph @xmath368 .
namely , let @xmath380 ( resp .
@xmath381 ) be the set of nodes @xmath10 that are reachable by a @xmath379-residual @xmath65@xmath10 walk that leaves @xmath65 and enters @xmath10 ( resp . leaves both @xmath65 and @xmath10 ) .
then @xmath382 , by the maximality of @xmath276 .
[ th : bd_barrier_from_flow ] let @xmath276 be a maximum ibd - flow for @xmath383 . define @xmath384 and @xmath385 , where @xmath386 and @xmath387 .
let @xmath388 be the node sets of weakly connected components of the underlying undirected subgraph of @xmath368 induced by @xmath389 .
define @xmath329 to be the bd - graph obtained from @xmath208 by flipping the set @xmath390 ( contained in @xmath16 ) .
then @xmath391 is a minimum odd barrier .
an important fact is that the minimum odd barrier @xmath392 does not depend on @xmath276 , and we refer to it as the _ canonical _ odd barrier for @xmath383 .
[ th : canon_bd_barrier ] the sets @xmath381 are same for all maximum ibd - flows @xmath276 in @xmath208 , and similarly for the sets @xmath380 , the minimum odd barriers @xmath392 , and the graphs @xmath329 obtained from @xmath208 by flipping @xmath393 . in fact
, we have freedom of choosing any of the two forms ( expensive or compact ) of @xmath149 to prove proposition [ prop : aux_bidir_flow ] in full , as it is easy to see that the claims in these cases are reduced to each other .
we prefer to deal with the expensive form , taking advantage from certain nice structural features arising in this case .
one reason for our choice is that any ibd - flow in the expensive @xmath149 is automatically good , as explained in remark [ rem : compact_vs_expensive ] .
we know that there exists a good fractional bidirected @xmath65-flow @xmath217 in @xmath149 that saturates the set @xmath264 of locked edges , and our goal is to show the existence of an ibd - flow saturating @xmath264 . recall that any edge @xmath394 is generated by some node @xmath10 of @xmath5 , i.e. @xmath395
. it will be convenient for us to construct the desired ibd - flow without explicitly imposing the `` lower capacities '' on the locked edges .
for this purpose , we modify @xmath149 as follows .
first , we add a loop @xmath396 with infinite capacity ( entering @xmath65 twice ) .
also we add to @xmath149 a node @xmath397 , which is regarded as a new source .
second , let @xmath264 contain an edge @xmath398 generated by a vertex @xmath399 in some zone @xmath109 , @xmath106 .
we delete @xmath153 from @xmath149 and , instead , add two edges @xmath400 and @xmath401 with capacity @xmath2 each .
third , let @xmath264 contain the loops @xmath204 ( @xmath201 ) for some central node @xmath168 of @xmath116 .
we replace each @xmath204 ( having unit capacity ) by edge @xmath402 with capacity 2 ; we call it the _ root edge _ at @xmath203 .
we denote the resulting bd - graph by @xmath403 and keep the previous notation @xmath56 for its edge capacities .
the above @xmath65-flow @xmath217 is transformed , in an obvious way , into a feasible @xmath397-flow in @xmath403 , denoted by @xmath217 as before .
note that this @xmath217 saturates all edges created from those in @xmath264 ( i.e. from @xmath153 and @xmath204 as above ) ; these edges leave @xmath397 and the value of @xmath217 is maximum among the feasible @xmath397-flows in @xmath403 and is equal to @xmath404 .
let @xmath276 be a maximum ibd - flow in @xmath403 .
we are going to prove that @xmath405 .
this would imply that the corresponding ibd - flow in @xmath149 saturates @xmath264 as required . to this aim , consider the _ canonical _
odd barrier @xmath406 for @xmath407 ( see theorem [ th : canon_bd_barrier ] ) . here
@xmath408 is the bd - graph ( with the source @xmath397 ) obtained from @xmath403 according to theorem [ th : bd_barrier_from_flow ] ( i.e. @xmath409 for @xmath410 ) . from now on , speaking of edge directions , the capacities @xmath56 and the flow @xmath276 , we mean those in @xmath408 , unless explicitly stated otherwise .
we have ( cf .
( [ eq : barrier_cap ] ) ) @xmath411 + c[{\overrightarrow{a } } ] - k.\ ] ] the following assertion is crucial .
[ lm : b_sets ] for each @xmath412 : * @xmath413 for some @xmath230 and @xmath414 ; * @xmath204 is not locked ( so @xmath403 contains the loop @xmath204 but not the root edge at @xmath203 ) ; * among the edges ( legs ) connecting @xmath16 and @xmath415 , one edge leaves @xmath16 and the other edges enter @xmath16 .
by the constructions of @xmath149 and @xmath403 , for any @xmath230 and distinct @xmath416 , there is an automorphism @xmath417 of @xmath403 that swaps @xmath203 and @xmath207 and is invariant on the other nodes .
also @xmath131 respects the capacities in @xmath403 , and the function @xmath418 induced by @xmath276 under @xmath131 ( i.e. @xmath419 ) is again a maximum ibd - flow in @xmath403 .
since @xmath356 is canonical , it follows from theorem [ th : canon_bd_barrier ] that @xmath420 the nodes in @xmath421 are _ flipped _ when constructing @xmath408 from @xmath403 . then ( [ eq : r_preserved ] ) and theorem [ th : bd_barrier_from_flow ] imply that for @xmath422 as above , [ eq : theta_preserved ] * @xmath203 is flipped if and only if @xmath207 is flipped ; * @xmath423 if and only if @xmath424 . *
@xmath425 if and only if @xmath426 .
let @xmath427 .
since the capacity @xmath428 $ ] ( in @xmath408 ) is odd , the set @xmath429 $ ] contains an edge @xmath23 with @xmath430 odd .
any edge in @xmath408 having an odd capacity is either a loop @xmath204 or a leg @xmath205 ( regarding `` infinite '' capacities as even ones ) .
obviously , no loop can be `` responsible '' for the oddness of @xmath428 $ ] .
so @xmath431 for some @xmath230 , @xmath432 and @xmath110 .
let @xmath433 denote the edge of @xmath403 corresponding to @xmath23 .
then ( see fig .
[ fig : aux_graph](b ) ) @xmath433 leaves @xmath170 and enters @xmath203 . due to flips
, however , this may not be the case for @xmath23 in @xmath408 .
suppose @xmath434 ( and @xmath435 ) .
then @xmath23 leaves @xmath203 , whence @xmath203 is a flipped node in @xmath16 .
now ( [ eq : theta_preserved])(a , b ) imply that all @xmath207 are flipped nodes belonging to @xmath16 and that @xmath436 $ ] for all @xmath437 .
but then @xmath23 can not be `` responsible '' for the oddness of @xmath438 $ ] since @xmath174 is even .
so we have @xmath439 and @xmath440 .
then @xmath23 leaves @xmath170 .
the edge @xmath433 leaves @xmath170 as well .
hence @xmath170 is not flipped .
since @xmath174 is even , there must be @xmath441 such that the leg @xmath442 is not in @xmath429 $ ] ( for otherwise one may pick another pair @xmath443 ) .
then @xmath207 is not in @xmath415 . in view of ( [ eq : theta_preserved])(c )
, @xmath207 belongs to a @xmath444-set in @xmath356 different from @xmath415 . considering the automorphisms @xmath445 for all distinct @xmath446 and using the fact that the canonical barrier @xmath356 preserves under @xmath131 ( in view of ( [ eq : r_preserved ] ) )
, we can conclude that the nodes @xmath447 belong to different @xmath444-sets in @xmath356 .
since these @xmath444-sets are pairwise disjoint and each automorphism @xmath131 swaps two copies of @xmath448 and do not move the remaining nodes in @xmath408 , each of these @xmath444-sets can contain only a single node .
thus , @xmath413 , yielding ( i ) in the lemma . next we show ( ii ) . from the construction of @xmath408 it follows that @xmath449 by the first equality , any @xmath450-residual walk ending at a node @xmath451 enters @xmath10 , and by the second equality , there exist an @xmath450-residual walk @xmath8 to @xmath452 that enters @xmath453 and an @xmath450-residual walk @xmath190 to @xmath453 that leaves @xmath453 . recall that the residual walks leave the source @xmath397 .
let @xmath454 be the last edge of @xmath8 , and @xmath455 the last edge of @xmath190 .
define @xmath310 ( @xmath456 ) to be the set of legs @xmath457 with @xmath458 ( resp .
@xmath459 ) . note that ( cf .
( [ eq : residual_edges ] ) ) if @xmath313 then @xmath460 and @xmath23 enters @xmath453 in @xmath461 , and if @xmath462 then @xmath463 and @xmath374 leaves @xmath453 in @xmath461 .
[ eq : epepp ] supposing the existence of the root edge @xmath464 ( in @xmath403 and @xmath408 ) , we can come to a contradiction as follows . since there is no loop at @xmath453 , both nodes @xmath143 are in @xmath16 .
note that the edge @xmath11 is different from @xmath465 ( which leaves @xmath453 ) and from @xmath466 ( which enters @xmath397 ) .
then ( [ eq : epepp ] ) implies that @xmath467 .
furthermore , @xmath11 is of the form @xmath468 . for
if @xmath11 enters @xmath92 then the edge of @xmath8 preceding @xmath11 leaves @xmath92 , whence the part of @xmath8 from @xmath397 to @xmath92 forms an @xmath450-residual walk leaving @xmath92 , which is impossible since @xmath469 .
so @xmath470 $ ] and @xmath471 . then @xmath472=0 $ ] , by theorem [ th : max_ibd_flow_min_odd_barrier](iii ) .
this implies that @xmath473 $ ] .
but then the last edge @xmath474 of the walk @xmath190 as above can not be reverse to any edge in @xmath456 ; for otherwise @xmath475 enters @xmath10 , implying that the part of @xmath190 from @xmath397 to @xmath10 leaves @xmath10 .
also @xmath475 is neither reverse to an edge in @xmath310 ( cf .
( [ eq : epepp ] ) ) , nor equal to @xmath465 .
the latter is because @xmath476 $ ] , and therefore , @xmath477 implies @xmath478 ( cf .
theorem [ th : max_ibd_flow_min_odd_barrier](iii ) ) , whence @xmath479 .
thus , @xmath190 does not exist .
this contradiction yields ( ii ) .
it remains to show ( iii ) . by ( ii )
, we have @xmath480 , and @xmath481 $ ] is exactly the set of legs at @xmath452 .
suppose @xmath482|\ne 1 $ ] .
then @xmath483 , since @xmath484=d$ ] is odd .
hence @xmath485\ge d-1\ge 2 $ ] ( by theorem [ th : max_ibd_flow_min_odd_barrier](iii ) ) .
also the fact that all legs enter @xmath453 together with @xmath486 and @xmath487 implies that the only possible case is when @xmath488 , @xmath485=2 $ ] and @xmath472=0 $ ] .
now take an @xmath450-residual walk @xmath190 to @xmath453 that leaves @xmath453 , and let @xmath475 be its last edge .
then @xmath475 is neither the loop @xmath204 ( which is saturated ) , nor reverse to a leg @xmath489 with @xmath490 . indeed ,
if @xmath491 then @xmath475 enters @xmath10 ( in view of @xmath451 and @xmath492 $ ] ) , and hence the part of @xmath190 from @xmath397 to @xmath10 leaves @xmath10 , which is impossible since @xmath451 .
this contradiction yields ( iii ) and completes the proof of the lemma .
based on lemma [ lm : b_sets ] , we now finish the proof of proposition [ prop : aux_bidir_flow ] .
consider @xmath413 and let @xmath205 be the unique edge in @xmath429 $ ] .
then @xmath493 = { \ { e_{w^i , t } \mid t\in t-\{s\ } \}}$ ] .
consider the maximum fractional bd - flow @xmath217 as before . by the goodness of @xmath217 ( see ( [ eq : good ] ) ) , we have @xmath494 - f[{\overleftarrow}a , b_p ] = f(e_{w^i , s } ) - \sum\nolimits_{t\in t-\{s\ } } f(e_{w^i , t } ) \\
= f(e_{w^i , s } ) - \left ( 2f(e_{w^i } ) - f(e_{w^i , s } ) \right ) = 2 \left ( f(e_{w^i , s } ) - f(e_{w^i } ) \right ) \le 0 = c[{\overrightarrow{a } } , b_p ] - 1
. \end{gathered}\ ] ] using this and ( [ eq : val_g ] ) , we have @xmath495 - 2f[{\overleftarrow}a , { \overrightarrow{a } } ] \right ) + \left ( f[{\overrightarrow{a } } ] - f[{\overleftarrow}a ] \right ) \\ = \left ( 2f[{\overrightarrow{a } } , { \overleftarrow}a ] - 2f[{\overleftarrow}a , { \overrightarrow{a } } ] \right ) + \left ( f[{\overrightarrow{a}},m ] - f[{\overleftarrow}a , m ] \right ) + \sum\nolimits_{p=1}^k \left ( f[{\overrightarrow{a } } , b_p ] - f[{\overleftarrow}a , b_p ] \right ) \\
\le 2c[{\overrightarrow{a } } , { \overleftarrow}a ] + c[{\overrightarrow{a } } , m ] + \sum\nolimits_{p=1}^k \left ( c[{\overrightarrow{a}},b_p ] - 1 \right ) = c[{\overrightarrow{a } } ] + 2c[{\overrightarrow{a } } , { \overleftarrow}a ] - k = c({\mathcal{b } } ) = { \mathop{\rm val}\nolimits ( g ) } .
\end{gathered}\ ] ] thus , we obtain the desired relation @xmath496 ( which , in fact , holds with equality ) .
this completes the proof of proposition [ prop : aux_bidir_flow ] .
[ th : dual_halfint ] let @xmath497 and @xmath498
. then problem ( @xmath71 ) has a half - integer optimal solution .
the proof follows easily from theorem [ th : primal_halfint ] and the general fact that the `` totally dual @xmath499-integrality '' implies the `` totally primal @xmath499-integrality '' , which is a natural generalization of a well - known result on tdi - systems due to edmonds and giles @xcite .
more precisely , we utilize the following simple fact ( see , e.g. , ( * ? ? ?
* statement 1.1 ) ) : [ lm : dual - primal ] let @xmath16 be a nonnegative @xmath500-matrix , @xmath475 an integral @xmath36-vector , and @xmath501 a positive integer .
suppose that the program @xmath502 has a @xmath499-integer optimal solution for every nonnegative integral @xmath291-vector @xmath56 such that @xmath503 has an optimal solution .
then for every nonnegative integral @xmath291-vector @xmath56 , the program @xmath504 has a @xmath499-integer optimal solution whenever it has an optimal solution . in our case
, we set @xmath505 and take as @xmath16 ( resp .
@xmath475 ) the constraint matrix ( resp . the right hand side vector ) of ( @xmath71 ) .
then @xmath475 is integral , @xmath503 becomes ( @xmath59 ) , @xmath506 becomes ( @xmath71 ) , and the half - integrality for the former implies that for the latter .
this proof is not `` constructive '' and does not lead directly to an efficient method for finding a half - integer optimal solution @xmath75 to ( @xmath71 ) .
below we devise a strongly polynomial algorithm .
it has as the input arbitrary ( rational - valued ) optimal solutions @xmath75 and @xmath7 to ( @xmath71 ) and ( @xmath59 ) , respectively , and outputs a half - integer optimal solution @xmath507 to ( @xmath71 ) .
( such @xmath75 and @xmath7 can be found in strongly polynomial time as described in section [ sec : primal_alg ] . ) as before , we set @xmath265 , and in what follows , speaking of a geodesic , we always mean an @xmath91-geodesic in @xmath5 , i.e. a @xmath6-path @xmath8 with @xmath508 . our goal is to construct @xmath509 satisfying the following conditions : [ eq : hat_l ] * @xmath510 for any @xmath6-path @xmath8 in @xmath5 ; * @xmath511 for each geodesic @xmath8 ; * for @xmath45 , if @xmath512 then @xmath513 .
then ( [ eq : hat_l ] ) and the complementary slackness conditions ( [ eq : cs_geodesic])([eq : cs_saturated ] ) imply that the node length @xmath507 forms an optimal solution to ( @xmath71 ) .
we construct an undirected graph @xmath208 and endow it with _ edge _ lengths @xmath514 as follows .
we first include in @xmath208 the terminal set @xmath6 and all nodes and edges of @xmath5 contained in geodesics .
also we add to @xmath208 the edges of @xmath5 with both ends lying on geodesics or @xmath6 .
the edges @xmath23 of the current @xmath208 are called _ regular _ and we define @xmath515 . next we add to @xmath208 additional edges , which are related to constraints due to parts of @xmath5 outside @xmath208 .
more precisely , we scan all pairs of nodes @xmath516 not connected by a ( regular ) edge and such that there exists a path @xmath190 in @xmath5 having all nodes in @xmath517 and whose first node is adjacent to @xmath92 , and the last node to @xmath10 .
we add to @xmath208 edge @xmath300 , referring to it as a _ virtual _ edge , and define its length @xmath518 to be the minimum value of @xmath519 among such paths @xmath190 .
the construction of @xmath520 reduces to a polynomial number of usual shortest paths problems in @xmath5 .
note that @xmath512 holds for each node @xmath521 ( by ( [ eq : cs_geodesic ] ) and ( [ eq : cs_saturated ] ) ) .
we assign @xmath522 for these nodes @xmath10 and will further focus on finding values of @xmath507 on the nodes in @xmath208 . for a path @xmath8 in @xmath208 ,
let @xmath523 denote its _ full length _ @xmath524 .
clearly @xmath525 holds for any @xmath6-path @xmath8 in @xmath208 , and for each @xmath6-path @xmath190 in @xmath5 , there exists a _ shortcut _
path @xmath8 in @xmath208 such that @xmath526 .
the desired lengths @xmath507 on @xmath527 will be extracted from a system of linear constraints described below . for a node @xmath528 ,
let @xmath529 ( @xmath530 ) denote the set of terminals @xmath106 ( resp .
pairs @xmath236 ) such that @xmath10 belongs to a geodesic from @xmath132 ( resp .
connecting @xmath132 and @xmath133 ) .
when a terminal @xmath132 belongs to no geodesic , we set by definition @xmath531 . for each @xmath528 and @xmath532
, we introduce two variables @xmath533 and @xmath534 and impose the following constraints : [ eq : rho_system ] * for each @xmath106 , @xmath535 .
* for each @xmath528 and @xmath532 , @xmath536 * for each @xmath528 and @xmath537 , @xmath538 ( and @xmath539 ) .
* if @xmath540 and @xmath541 , then @xmath542 moreover , if there exists a geodesic from @xmath132 containing both @xmath143 in this order ( resp .
in the order @xmath543 ) , then the former ( resp .
the latter ) inequality is replaced by equality .
( note that in this case @xmath544 . ) * if @xmath540 , @xmath545 , @xmath546 , and @xmath120 , then @xmath547 .
the meaning of these variables becomes evident from the proof of the next statement .
system ( [ eq : rho_system ] ) has a solution . for a @xmath548@xmath65 path @xmath8 in @xmath208 ,
define its _ pre - length _ to be @xmath549 ( i.e. compared with the full length , we do not count the last node ) . for @xmath528 and @xmath532 , define @xmath533 ( resp .
@xmath534 ) to be the minimum pre - length ( resp .
the minimum full length ) of an @xmath132@xmath10 path in @xmath208
. then ( [ eq : rho_system])(i)(iii ) follow from the construction . condition ( [ eq : rho_system])(iv ) represents a sort of triangle inequalities ( giving one equality if @xmath23 belongs to a geodesic from @xmath132 ) .
finally , condition ( [ eq : rho_system])(v ) holds since the full length of any @xmath132@xmath133 path in @xmath208 is at least @xmath64 .
we observe that in linear system ( [ eq : rho_system ] ) , each constraint contains at most two variables , each occurring with the coefficient 1 or 1 , and that the r.h.s . in it
is an integer .
a well - known fact is that a linear system with such features is totally dual half - integral ; therefore , it has a half - integer basis solution ( whenever it has a solution at all ) , and such a solution can be found in strongly polynomial time ( cf . , e.g. , @xcite ) . given a half - integer solution @xmath550 to ( [ eq : rho_system ] ) , we define half - integer node lengths @xmath507 as follows : @xmath551 now the desired algorithmic result is provided by the following @xmath507 is well - defined and satisfies ( [ eq : hat_l ] ) .
we first show that for any @xmath528 and @xmath552 , @xmath553 this is trivial when @xmath554 ( since in this case @xmath555 and @xmath556 ) .
let @xmath557 .
if @xmath558 , then ( [ eq : same_hat_l ] ) follows from ( [ eq : rho_system])(iii ) .
now ( [ eq : same_hat_l ] ) with any two @xmath559 is implied by the fact that the graph whose nodes and edges are the elements of @xmath529 and @xmath530 , respectively , is connected ( as it is easy to see that for @xmath560 , at least one of @xmath561 is in @xmath530 as well ) .
so @xmath507 is well - defined .
property ( [ eq : hat_l])(iii ) is immediate from ( [ eq : rho_system])(ii ) . to see ( [ eq : hat_l])(ii ) , consider an @xmath132@xmath133 geodesic @xmath8 .
going along @xmath8 step by step and applying ( [ eq : rho_system])(ii),(iv ) , we observe that for each node @xmath10 on @xmath8 , the @xmath132@xmath10 part @xmath562 of @xmath8 satisfies @xmath563 . when reaching @xmath133 , we obtain @xmath564 , and now ( [ eq : hat_l])(ii ) follows from ( [ eq : rho_system])(iii ) and @xmath565 ( by ( [ eq : rho_system])(i ) ) .
finally , consider an arbitrary @xmath6-path @xmath190 in @xmath208 , from @xmath548 to @xmath65 say . to conclude with ( [ eq : hat_l])(i )
, it suffices to show that @xmath566 represent @xmath190 as the concatenation @xmath567 , where @xmath197 is a part of a geodesic from @xmath548 .
we prove ( [ eq : hatlambda ] ) by induction on the number @xmath568 of edges in @xmath569 . when @xmath570 , @xmath190 is a geodesic , and we are done .
assuming this is not the case , take the first edge @xmath140 of @xmath569 , where @xmath92 is the end of @xmath197 . by reasonings above ,
if @xmath555 ( and therefore , @xmath572 ) , ( [ eq : hatlambda ] ) immediately follows from ( [ eq : rho_system])(v ) ( with @xmath573 and @xmath574 ) . and
if @xmath575 , then @xmath10 belongs to some @xmath132@xmath133 geodesic @xmath576 .
, one may assume that @xmath577 and @xmath578 . applying ( [ eq : rho_system])(v ) to @xmath579 , we have @xmath580 comparing this with @xmath581 and using @xmath582
, one can conclude that @xmath583 , where @xmath584 is the @xmath133@xmath65 path being the concatenation of the @xmath133@xmath10 part of ( the reverse of ) @xmath576 and the @xmath10@xmath65 part @xmath585 of @xmath190 .
since @xmath586 , we can apply induction and obtain @xmath587 , as required .
in this section we recall the notions of skew - symmetric graphs and integer skew - symmetric flows , review known results on such graphs and flows , and then use them to derive necessary results on bidirected graphs and flows to which we appealed in section [ sec : prim_h_int ] . a _ skew - symmetric graph _ , or an _ sk - graph _ for short , is a digraph @xmath588 , with possible parallel arcs , endowed with two bijections @xmath589 such that : @xmath590 is an involution on the nodes ( i.e. @xmath591 and @xmath592 for each node @xmath10 ) ; @xmath593 is an involution on the arcs ; and for each arc @xmath11 from @xmath92 to @xmath10 , @xmath594 is an arc from @xmath595 to @xmath596
. for relevant results on sk - graphs and a relationship between sk- and bd - graphs , see @xcite . for brevity @xmath590 and @xmath593
are combined into one mapping @xmath597 on @xmath598 , which is called the _ symmetry _ ( or skew - symmetry , to be precise ) of @xmath5 . for a node ( arc )
@xmath599 , its symmetric node ( arc ) @xmath600 is also called the _ mate _ of @xmath599 , and we usually use notation with primes for mates , denoting @xmath600 by @xmath601 . although @xmath5 is allowed to contain parallel arcs , when it is not confusing , an arc from @xmath92 to @xmath10 may be denoted as @xmath602 or @xmath141 .
the symmetry @xmath597 is extended in a natural way to walks , subgraphs and other objects in @xmath5 .
in particular , two walks are symmetric to each other if the elements of one of them are symmetric to those of the other and go in the reverse order : for a walk @xmath605 , the symmetric walk @xmath606 is @xmath607 .
next we explain a relationship between skew - symmetric and bidirected graphs . given an sk - graph @xmath5 ,
choose an arbitrary partition @xmath608 of @xmath24 such that @xmath609 .
then @xmath5 and @xmath131 determine the bd - graph @xmath610 with @xmath611 whose edges correspond to the pairs of symmetric arcs in @xmath5 .
more precisely , arc mates @xmath612 of @xmath5 generate one edge @xmath23 of @xmath610 connecting nodes @xmath613 such that : ( i ) @xmath23 goes from @xmath92 to @xmath10 if one of @xmath614 goes from @xmath92 to @xmath10 ( and the other goes from @xmath603 to @xmath615 in @xmath616 ) ; ( ii ) @xmath23 leaves both @xmath143 if one of @xmath612 goes from @xmath92 to @xmath603 ( and the other from @xmath10 to @xmath615 ) ; ( iii ) @xmath23 enters both @xmath617 if one of @xmath612 goes from @xmath615 to @xmath10 ( and the other from @xmath603 to @xmath92 ) .
note that @xmath23 becomes a loop if @xmath614 connect a pair of symmetric nodes .
conversely , a bd - graph @xmath610 determines an sk - graph @xmath5 with symmetry @xmath597 as follows .
make a copy @xmath618 of each element @xmath10 of @xmath619 , forming the set @xmath620 .
put @xmath621 . for each edge @xmath23 of @xmath610 connecting nodes @xmath92 and @xmath10 , assign two `` symmetric '' arcs @xmath614 in @xmath5 so as to satisfy ( i)(iii ) above ( where @xmath622 and @xmath623 ) .
an example is depicted in fig .
[ fig : bd - sk ] .
[ rem : bd_ss ] note that one bd - graph generates one sk - graph , by the second construction . on the other hand ,
one sk - graph generates a set of bd - graphs , depending on the partition @xmath131 of @xmath624 , by the first construction .
namely , for each pair of symmetric mates @xmath625 in @xmath5 one may distribute @xmath626 between @xmath627 so that either @xmath628 , @xmath629 or , reversely , @xmath630 , @xmath631 .
the resulting bd - graphs are obtained from one other by making corresponding flips ( defined in subsection [ ssec : ibd - flows ] ) .
there is essentially a one - to - one correspondence between the walks in @xmath610 and @xmath5 .
more precisely , let @xmath632 be the natural mapping of @xmath633 to @xmath634 .
each walk @xmath605 in @xmath5 ( where @xmath635 ) induces the sequence @xmath636 of nodes and edges in @xmath610 .
one can see that @xmath637 is a walk in @xmath610 ( i.e. @xmath638 form a transit pair at @xmath639 , for each @xmath202 ) and that @xmath640 is the walk reverse to @xmath637 .
moreover , for any walk @xmath641 in @xmath610 , there is exactly one walk @xmath8 in @xmath5 such that @xmath642 ( considering @xmath8 up to replacing an arc @xmath643 by its mate @xmath604 when @xmath612 are parallel , i.e. correspond to a loop in @xmath610 ) .
we call a function @xmath343 on the arcs of an sk - graph @xmath5 _ ( self-)symmetric _ if @xmath644 for all @xmath645 .
let @xmath646 be a designated _ source _ ; its mate @xmath647 is regarded as the _
sink_. an _ integer skew - symmetric @xmath132@xmath647 flow _ , or an _ isk - flow _ for short , is a symmetric function @xmath648 being an @xmath132@xmath647 flow in a usual sense : @xmath215 for all @xmath649 , and @xmath650 .
the _ value _ of @xmath217 is @xmath651 . here
@xmath652 denotes the usual divergence ( given by ( [ eq : div ] ) , where @xmath209 and @xmath210 are the sets of arcs entering and leaving @xmath10 , respectively ) .
the above correspondence between bd- and sk - graphs is naturally extended to flows .
more precisely , if @xmath217 is a symmetric @xmath132@xmath647 flow in @xmath5 , then transferring the values of @xmath217 from the pairs of arc mates of @xmath5 to the edges of the bd - graph @xmath655 , we obtain a @xmath656-flow in @xmath610 , denoted as @xmath657 .
the converse correspondence is evident as well . for @xmath658 ,
let @xmath659 denote the set of arcs going from @xmath22 to @xmath332 . also ( accommodating notation from section [ sec : prim_h_int ] to digraphs ) we denote by @xmath338 $ ] the set of arcs leaving @xmath22 .
let @xmath660 be a symmetric capacity function .
then a tuple @xmath661 of subsets of @xmath24 is called a ( skew - symmetric ) _ odd barrier _
the source @xmath132 ) if the following conditions hold ( see fig . [
fig : odd_barrier ] ) : * the sets @xmath662 give a partition of @xmath24 , each @xmath663 is self - symmetric ( @xmath664 ) , and @xmath665 ; * for each @xmath202 , @xmath666 is odd . * for distinct @xmath422 , @xmath667 .
* for each @xmath202 , @xmath668 .
odd barriers in skew - symmetric graphs are related to their bidirected counterparts introduced in section [ sec : prim_h_int ] .
indeed , consider a bd - graph @xmath610 with integer edge capacities @xmath670 and a source @xmath132 .
construct the related sk - graph @xmath5 with @xmath671 , where @xmath672 .
edge capacities @xmath56 in @xmath610 induce symmetric arc capacities in @xmath5 , also denoted by @xmath56 .
the source @xmath132 in @xmath610 gives the source @xmath132 and the sink @xmath647 in @xmath5 .
consider a skew - symmetric odd barrier @xmath673 in @xmath5 .
this barrier gives rise to the following odd bd - barrier @xmath674 in @xmath610 obeying @xmath675 .
we first construct a new bd - graph from @xmath5 by taking a bipartition @xmath676 of @xmath24 such that @xmath677 and @xmath678 ; cf .
remark [ rem : bd_ss ] .
the resulting bd - graph @xmath679 is equivalent to @xmath610 .
moreover , @xmath679 is obtained from @xmath610 by flipping a subset of nodes within @xmath16 .
the node subsets @xmath680 in @xmath5 are self - symmetric and induce subsets @xmath681 in @xmath610 and @xmath679 in a natural way ; namely , @xmath682 and similarly for @xmath683 .
define @xmath684 , where @xmath685 .
a straightforward examination shows that the properties in ( [ eq : sk_odd_barrier ] ) imply their bidirected counterparts in ( [ eq : bd_odd_barrier ] ) . to see that @xmath675 , define @xmath686 .
note that @xmath687 = c(a , a ' ) + c(a , z)$ ] .
the capacity @xmath688 is equal to @xmath689 $ ] ( in @xmath679 ) since @xmath690 consists of pairs of arc mates , each pair corresponding to an edge in @xmath691 $ ] . and
the capacity @xmath692 is equal to @xmath693 $ ] since the ( symmetric ) set @xmath694 corresponds to @xmath695 in @xmath679 .
[ th : max_isk_flow_min_odd_barrier ] for @xmath696 as above , the maximum isk - flow value is equal to the minimum odd barrier capacity .
an isk - flow @xmath217 and an odd barrier @xmath697 have maximum value and minimum capacity , respectively , if and only if the following hold : next we establish additional correspondences .
consider an isk - flow @xmath217 in @xmath5 .
the _ residual _ sk - graph @xmath704 endowed with the _ residual _ capacities @xmath705 is constructed in a standard fashion : @xmath706 , and the arcs of @xmath704 are : ( cf .
( [ eq : residual_edges ] ) ) .
a path @xmath8 in @xmath704 is called _ @xmath713-regular _ if @xmath714 holds for each pair of arc mates @xmath614 occurring in @xmath8 .
( in other words , the bidirected image of @xmath8 in @xmath715 is a @xmath713-simple walk . )
if @xmath8 is a @xmath713-regular @xmath132@xmath647 path , we can increase the value of @xmath217 by 2 ( by sending one unit of flow along @xmath8 and one unit of flow along @xmath562 ) .
so the existence of such a @xmath8 implies the non - maximality of @xmath217 . a converse property is valid as well .
given a maximum isk - flow @xmath217 , a certain minimum odd barrier can be constructed by considering the residual graph @xmath704 .
the construction described in the proof of theorem 3.5 in @xcite ( relying on lemma 2.2 in @xcite ) is as follows .
[ th : sk_barrier_from_flow ] let @xmath217 be a maximum isk - flow .
let @xmath716 be the set of nodes reachable from @xmath132 by @xmath713-regular paths in @xmath704 .
define @xmath717 and @xmath718 .
let @xmath388 be the node sets of weakly connected components of the subgraph @xmath5 induced by @xmath719 .
then @xmath720 is a minimum odd barrier .
@xmath721 this subset @xmath584 of nodes in @xmath5 corresponds to two sets @xmath722 and @xmath723 in @xmath610 ( defined just before theorem [ th : bd_barrier_from_flow ] ; here @xmath724 and @xmath725 ) .
more precisely , assuming that each node @xmath726 corresponds to node mates @xmath727 in @xmath5 ( cf .
subsection [ ssec : sk_graph ] ) , one can realize that @xmath728 ( resp .
@xmath729 ) is the set of nodes @xmath730 such that @xmath731 ( resp .
@xmath732 ) . let @xmath217 be a maximum isk - flow such that the set @xmath733 is inclusion - wise minimal and let @xmath734 .
consider another maximum isk - flow @xmath276 ( if any ) .
then @xmath735 - k = c({\mathcal{b } } ) = { \mathop{\rm val}\nolimits ( g ) } = \sum\nolimits_{v \in a } { \mathop{\rm div}\nolimits}_{g}(v ) \\
= g\langle a , a'\rangle + g\langle a , m\rangle + g\langle a , b_1\rangle + \ldots + g\langle a , b_k\rangle , \end{gathered}\ ] ] where for disjoint subsets @xmath736 , @xmath737 denotes @xmath738 . for @xmath351
, we have : @xmath739 ; @xmath740 is odd ; and @xmath741 is even ( the latter is due to a result in @xcite ; see also ( * ? ? ? * corollary 3.2 ) ) . therefore , @xmath742 .
also @xmath743 and @xmath744 . comparing these relations with ( [ eq : canon_value ] )
, we conclude that : in terms of the residual graph @xmath749 , ( i ) and ( ii ) mean that the sets @xmath16 and @xmath750 are connected in @xmath749 by exactly @xmath501 arcs @xmath751 , each @xmath752 going from @xmath16 to @xmath663 and having the residual capacity 1 . by symmetry , @xmath753 and @xmath754
are connected by only the arcs @xmath755 ( each @xmath756 goes from @xmath663 to @xmath753 , and @xmath757 ) . also by ( [ eq : sk_odd_barrier])(iii),(iv ) , no arc in @xmath749
connects different sets among @xmath680 .
therefore , the set @xmath758 of nodes in @xmath749 reachable from @xmath132 by @xmath377-regular paths is contained in @xmath733 . by the minimality of @xmath733 , we have @xmath759 , as required .
@xmath721 j. edmonds and e. l. johnson . matching : a well - solved class of integer linear programs . in r. guy ,
h. hanani , n. sauer , and j. schnhein , eds , _ combinatorial structures and their applications _ ,
gordon and breach , ny , 1970 , pp .
8992 . g. pap . some new results on node - capacitated packing of a - paths . in _
stoc 07 : proceedings of the thirty - ninth annual acm symposium on theory of computing _ , pages 599604 , new york , ny , usa , 2007 .
|
we consider an undirected graph @xmath0 with a set @xmath1 of terminals , and with nonnegative integer capacities @xmath2 and costs @xmath3 of nodes @xmath4 .
a path in @xmath5 is a _
@xmath6-path _ if its ends are distinct terminals . by a _ multiflow _
we mean a function @xmath7 assigning to each @xmath6-path @xmath8 a nonnegative rational _ weight _ @xmath9 , and a multiflow is called _ feasible _ if the sum of weights of @xmath6-paths through each node @xmath10 does not exceed @xmath2 .
the _ value _ of @xmath7 is the sum of weights @xmath9 , and the _ cost _ of @xmath7 is the sum of @xmath9 times the cost of @xmath8 w.r.t .
@xmath11 , over all @xmath6-paths @xmath8 .
generalizing known results on edge - capacitated multiflows , we show that the problem of finding a minimum cost multiflow among the feasible multiflows of maximum possible value admits _ half - integer _ optimal primal and dual solutions .
moreover , we devise a strongly polynomial algorithm for finding such optimal solutions
. _ keywords _ : minimum cost multiflow , bidirected graph , skew - symmetric graph
|
the derivation of semiclassical periodic orbit formulas for the trace of the quantum green function led to a deeper understanding of the influence of classical dynamics on quantum spectra .
closed periodic orbit expressions have been given by gutzwiller @xcite and balian and bloch @xcite for `` hard chaos '' systems and by berry and tabor @xcite for integrable dynamics .
integrability and hard chaos represent the two extremes on the scale of possible hamiltonian dynamics .
the term `` hard chaos '' introduced by gutzwiller @xcite is , however , not well defined .
it implies , that all periodic orbits are unstable and `` sufficiently '' isolated to allow for the stationary phase approximations in the derivation of the trace formula .
hence , for lack of a better definition , one might say that a system exhibits `` hard chaos '' if the gutzwiller trace formula as it stands is the leading term in a semiclassical expansion in @xmath0 .
neither integrability nor hard chaos is generic in low dimensional bounded hamiltonian dynamics . the variety of classical systems proposed as testing models for the gutzwiller trace formula in the last decade exemplify the problem of finding ideal chaos in the sense described above .
typical hamiltonian systems show a mixture of the two extremes .
chaotic regions in phase space ( containing unstable periodic orbits only ) are interspersed by stable islands , which themselves may have a complicated inner structure consisting of chaotic bands and invariant tori .
the `` stable regions '' reflect locally almost integrable behavior in phase space and can be treated semiclassically by an einstein brillouin
keller ( ebk ) approximation @xcite .
semiclassical contributions to the trace of the green function from periodic orbits inside the stable islands are of modified berry tabor type as has been discussed in @xcite .
much less is known about a semiclassical treatment of the transition region from the stable island to the outer chaotic neighborhood .
the outer boundary acts as a classically impenetrable wall in two degrees of freedom . a coupling between dynamically separated regions can only be described by complex trajectories @xcite .
although the outer chaotic region contains ( per definition ) only unstable periodic orbits , it shows almost regular behavior near the stable components .
the characteristics for this kind of regularity , also called intermittency in the chaos literature @xcite , is a vanishing liapunov exponent @xmath1 for unstable periodic orbits approaching the island , i.e.@xmath2 where @xmath3 denotes the largest eigenvalue of the stability matrix ( describing the linearized dynamics in the neighborhood of the periodic orbit ) and @xmath4 is the period .
i will show that intermittency in bounded systems leads to leading order corrections in the gutzwiller trace formula .
semiclassical contributions from ( unstable ) periodic orbits approaching the boundary of a stable island show a @xmath0dependent transition from gutzwiller to berry tabor like behavior .
this allows one to extend the concept of ebk quantization from a stable region over the marginal stable boundary into its `` chaotic '' neighborhood . + we will derive these results for a specific example , the quarter stadium billiard , ( see fig . [
fig : stadium ] ) .
this might appear surprising on first sight , because this billiard has been introduced as an example of a dynamical system which is completely ergodic with positive liapunov exponent @xcite .
the stadium billiard has thus been regarded as an ideal model for a `` chaotic '' system .
first indications for the validity of the random matrix conjecture relating the level statistics of individual `` chaotic '' systems to those of an ensemble of random gaussian orthogonal matrices have been found in this billiard @xcite . also the discovery of scarred wave functions @xcite was first made in this system . from then on , the classical and quantum aspects of the stadium billiard were studied intensively both theoretically and in microwave experiments @xcite , making it a standard model in quantum chaology .
the billiard dynamics possesses , however , regularities , which prevents it from being a hard chaos system . the so
called bouncing ball orbits , i.e. the continuous family of periodic orbits running back and forth in the rectangle , are marginally stable .
another peculiarity of the stadium is the existence of whispering gallery orbits accumulating at the boundary of the billiard .
these orbits have a strong , non generic influence on the spectral statistics which causes deviations from the goe
predictions @xcite .
there also exist quantum eigenstates , which are localized along the bouncing ball orbits ( the bouncing ball states ) , or along the boundary ( the whispering gallery states ) , in a much more pronounced way than the scars found along other periodic orbits . as a consequence ,
the stadium billiard is not an example of a hard chaos billiard , but may serve as an ideal model for systems exhibiting both regular and chaotic dynamics .
though there is no stable island , the bouncing ball orbits act as the boundary of a torus enclosing an area with zero volume in phase space .
this allows us to study pure boundary effects without dealing with the inner structure of the island .
semiclassical results obtained for hard chaos systems suffer considerable changes when dealing with intermittency as will be shown here .
first of all , the contributions of the marginal stable family to the trace of the green function need a special treatment @xcite .
this is true , however , also for unstable periodic orbits close to the bouncing ball family .
they can no longer be treated as isolated stationary phase points when approximating the trace of the green function at finite @xmath0 . as a consequence ,
the gutzwiller periodic orbit formula is not valid for the dynamics in a whole phase space volume surrounding the marginal stable family .
the size of this volume is explicitly @xmath0 dependent and shrinks to zero in the semiclassical limit .
furthermore , standard semiclassical arguments referring to a cut off in periodic orbit sums at a period given by the heisenberg time are not valid for intermittent systems .
intermittency is related to a loss of internal time scales in the dynamics and infinitely long orbits thus contribute dominantly to the semiclassical zeta function at finite @xmath0 . in particular ,
the regular bouncing ball states follow scaling laws for a cut off in the orbit summation different from the estimate given by the berry keating resummation technique @xcite . + the article is organized as follows : in section [ sec : bb ] , the contribution of the bouncing ball family to the trace of the green function is discussed .
the derivation follows mainly the ideas developed in @xcite although here we do not deal with diffraction effects .
it will be shown that the bouncing ball orbits give no contribution to individual eigenvalues of the stadium billiard . in section
[ sec : t - oper ] , the bogomolny transfer operator method @xcite is introduced .
a poincar surface of section is defined , which allows us to study the near bouncing ball dynamics , but explicitly excludes the bouncing ball family .
the poincar map in the bouncing ball limit is derived , which allows us to obtain the leading contributions to the transfer operator for the bouncing ball spectrum .
an approximate ebk
quantization of the bouncing ball states can be derived .
an analysis of the trace of the transfer operator unveils the breakdown of the gutzwiller periodic orbit formula for unstable periodic orbits close to the bouncing ball family . in section [ sec : bbstates ] , we develop the concept of a `` semiclassical island of stability '' surrounding the marginally stable bouncing ball family in phase space and discuss the semiclassical limit of the bouncing ball spectrum .
in this section , the bouncing ball contribution to the quantum spectral determinant @xmath5 is derived .
the \{@xmath6 } denote the real quantum eigenvalues of the system . following ref .
@xcite , the green function for the quarter stadium billiard , fig .
[ fig : stadium ] , is divided into a bouncing ball part and the rest , @xmath7 where @xmath8 .
the bouncing ball contribution is restricted to @xmath9 , @xmath10 inside the rectangle and contains in semiclassical approximation all paths from @xmath9 to @xmath10 not reaching the circle boundary or bouncing off the vertical line in the rectangle .
the classical paths correspond essentially to free motion , and the bouncing ball green function @xmath11 can be written as an infinite sum over hankel functions ( plus phases due to hard wall reflections ) .
the trace then has the form @xcite @xmath12 where @xmath13 is the length of the rectangle , @xmath14 denotes the radius of the circle and @xmath15 is the hankel function of the first kind .
the real part of the trace diverges due to the logarithmic singularity of the neumann function @xmath16 at the origin .
this kind of divergence is a well known phenomenon for the trace of green functions and can be regularised e.g. by defining @xmath17 $ ] for @xmath18 fixed @xcite .
note , that the same regularisation techniques have to be applied for the true quantum trace @xmath19 and are thus not an artifact of the semiclassical approximation .
we now move directly to the integrated trace , which reads @xmath20 and the asymptotic form of the hankel function has been used in the last step .
the sum in ( [ inttrap ] ) is of the same kind as the contribution of a periodic orbit family in an integrable system ( here the rectangle ) as derived by berry and tabor @xcite .
the real part of the non oscillating contribution is a function of @xmath21 only , @xmath22 the real parameter @xmath23 originates from the regularisation and has the form @xmath24 the imaginary part is the bouncing ball contribution to the mean level staircase function @xmath25 the term @xmath26 is of the general form as derived in @xcite for arbitrary billiards with compact domain , i.e. @xmath27 where @xmath28 corresponds to the volume of the billiard and @xmath23 , @xmath29 , @xmath30 are real constants .
note , that the logarithmic singularity for @xmath31 in ( [ b2 g ] ) is absent in ( [ b2 ] ) .
the sum in ( [ inttr ] ) is convergent on the whole complex @xmath32plane ( after analytic continuation for @xmath33 ) .
the function @xmath34 is free of singularities and analytic everywhere apart from the lines @xmath35 , @xmath36 integer , where it exhibits square
root cusps .
the dominant oscillation introduced through the bouncing ball orbits is clearly visible in the oscillatory part of the level staircase function @xcite , and agrees with the prediction ( [ inttr ] ) .
the bouncing ball part does , however , not contribute to individual eigenvalues .
the spectral determinant ( [ specdet ] ) can be factorised according to @xmath37 and the bouncing ball contribution @xmath38 is a function which has no zeros in the whole complex @xmath32 plane .
it gives rise to oscillations in the modulus of the spectral determinant with an amplitude growing like @xmath39 and a period @xmath40 , but it does not influence the individual zeros .
the bouncing ball contributions to the spectral determinant can thus be divided out without losing information about the spectrum .
this can be demonstrated by considering the boundary integral determinant @xmath41 where @xmath42 is the boundary integral kernel defined for @xmath9 and @xmath10 on the boundary of the billiard , ( for details see @xcite ) .
the function @xmath43 has the same zeros as the exact spectral determinant @xmath44 , but may differ otherwise .
especially , a smooth part @xmath45 is absent in the boundary integral determinant .
this term originates from the zero length limit in the green function and is thus a pure volume effect .
the modulation due to the bouncing ball contribution is clearly present in @xmath43 , see fig .
[ fig : det]a .
the large overall oscillation of the amplitude covers 7 order of magnitudes for @xmath46 but is completely removed by factorising out the bouncing ball part @xmath38 , see fig .
[ fig : det]b . note , that the boundary integral determinant after factorization exhibits cusps at @xmath47 , and is thus non
analytic there .
+ as a somewhat paradoxical result , we obtain that the bouncing ball family is responsible for strong modulations in the trace as well as in the spectral determinant .
it does not , however , contribute to individual eigenvalues , and does not explain the regular modulation in the level spacing itself .
the bouncing ball family covers a volume of measure zero in phase space , and is not sufficient to form a support for an eigenfunction alone .
( this is of course true for any invariant family e.g. in an integrable system . ) as a consequence , the neighborhood of the bouncing ball family in phase space must be responsible for the existence of the bouncing ball states and for the periodic change in the level spacings leading to the modulation in the level staircase function .
the next section is devoted to a proper semiclassical analysis of the near bouncing ball dynamics .
so far , it has been widely believed , that the remainder term in the trace of the green function ( [ green ] ) as well as in the spectral determinant ( [ specfak ] ) can in leading semiclassical approximation be described by the gutzwiller periodic orbit formula @xcite or equivalently by the semiclassical spectral determinant @xcite .
this belief is confirmed by studies of the fourier transform of the ( smoothed ) quantum spectrum , which agrees reasonably well with periodic orbit predictions @xcite . the semiclassical expression for the determinant , valid for systems with unstable , isolated periodic orbits only , has the form @xmath48 the sum is taken here over all ( single repeats of ) periodic orbits of the system .
the classical action @xmath49 is taken along the periodic orbit , and @xmath50 denotes the reduced monodromy matrix , which describes the linearized dynamics in phase space perpendicular to the periodic orbit on the energy manifold .
the maslov index @xmath51 counts twice the full rotations of the ( real ) eigenvectors of @xmath50 around the orbit ( plus two times the number of hard wall reflections ) .
the prefactors in front of the periodic orbit product are due to the zero - length limit in the green function in a similar way as derived in ( [ b2 ] ) , ( [ nmean ] ) and do not contribute to the spectrum .
( [ detgut ] ) is formal in the sense , that it is not convergent for real energies @xcite and suitable resummation techniques have to be applied @xcite .
thereby , the exponential function containing the periodic orbit sum is expanded and the resulting terms are regrouped by ordering them with respect to the total action or the total symbol length ( after choosing a suitable symbolic dynamics ) .
such techniques have been shown to work successfully for hard chaos systems . in the following
, i will show that formula ( [ detgut ] ) is not valid for periodic orbits within a phase space region surrounding the bouncing ball family and will give explicit bounds for this area .
our starting point is the bogomolny transfer operator @xcite which is a semiclassical propagator for a classical poincar map .
it has the form @xmath52 where @xmath9 , @xmath10 are points on an appropriate poincar surface of section in coordinate space and @xmath53 denotes the number of degree of freedom .
the sum has to be taken over all classical paths from @xmath9 to @xmath10 crossing the poincar surface only once with momentum pointing in the direction of the normal to the surface .
again , @xmath54 denotes the classical action along the path for fixed energy @xmath55 . for billiard systems
, it equals @xmath32 times the length of the classical trajectory .
the integer number @xmath56 counts the number of caustics in @xmath9space ( plus again twice the number of hard wall reflections ) .
the semiclassical eigenvalues are given by the zeros of the determinant @xmath57 .
evaluating the determinant in a cumulant expansion using the plemelj
smithies formula @xcite @xmath58 with @xmath59 provides the connection to the expanded periodic orbit formula ( [ detgut ] ) using the iterates of the map as an expansion parameter .
periodic orbits appear as stationary phase points in the various traces and the amplitudes are recovered using the relation @xmath60 the stationary phase approximation demands periodic orbits to be unstable and sufficiently isolated .
the last condition will be discussed in detail later .
note that the determinant and the cumulant expansion ( [ cumexp ] ) are well defined only if the operator @xmath61 is _ trace class _ , which means essentially , that the trace of @xmath61 exists and is finite in any basis .
( for further details see @xcite , other expansion methods using fredholm theory are discussed in @xcite ) .
the bogomolny transfer matrix method has been shown to work satisfactorily for hard chaos @xcite , mixed @xcite as well as integrable @xcite systems , and has also been applied successfully to the stadium taking the billiard boundary as the poincar surface of section @xcite .
we seek now a poincar map which reflects the whole dynamics but excludes the bouncing ball orbits in order to obtain a semiclassical expression for the remainder @xmath62 in ( [ specfak ] ) . choosing the intersection between the rectangle and the circle ( the vertical dotted line in fig .
[ fig : stadium ] ) fulfills this criteria .
the transfer operator corresponding to this poincar section takes on a particularly simple form for both the near bouncing ball limit and the whispering gallery limit .
it is thus preferable to other possible choices such as e.g. the circle boundary .
= 13.0 cm in the following , i will restrict attention to the stadium with @xmath14 = 1 .
the spectrum for general @xmath14 is obtained by simple scaling relations .
i will consider the classical poincar map @xmath63 with @xmath64 $ ] being the coordinate on the poincar plane starting from the bottom line .
the angle @xmath65-\pi/2,\pi/2[$ ] corresponds to the momentum vector pointing away from the circle measured in the clockwise direction , see fig .
[ fig : stadium ] .
( the corresponding energy dependent area preserving map is obtained in the coordinates @xmath66 = @xmath67 . )
the map can be written as @xmath68 \\
\nonumber \overline{\phi } & = & ( -1)^m \phi \quad \qquad \qquad \qquad \mbox{with } \quad m\le x+2 a \tan\phi < m+1 \\\nonumber \theta & = & \arcsin(\overline{x}\cos\overline{\phi})\\ \nonumber \phi ' & = & \overline{\phi } + 2(l+1)\theta - k\pi \qquad \mbox{with } \quad l\le\frac{\overline{\phi } + \theta}{\pi - 2 \theta } < l+1 \\ \nonumber x ' & = & \overline{x}\ ; \frac{\cos\overline{\phi}}{\cos\phi'}.\end{aligned}\ ] ] the coordinates @xmath69 correspond to the first return at the poincar plane with momentum pointing toward the circle . the angle @xmath70 is the angle of incidence for reflections on the circular section of the boundary ; see fig . [ fig : stadium ] . the length of a trajectory for one iteration of the map is @xmath71 the general expression for the monodromy matrix is given in appendix a. the integer numbers @xmath72 $ ] correspond to @xmath73reflections on the bottom or top line in the rectangle , the sign of @xmath36 equals the sign of @xmath74 .
the index @xmath75 $ ] counts the number of free flights in the circle and corresponds to @xmath76 bounces with the circle boundary .
the total number of bounces @xmath77 with the billiard boundary is thus @xmath78 the map provides automatically a symbolic coding .
it should be noted that the code does not form a `` good '' symbolic dynamics in the sense of markov partition theory .
multiple iterates of the map are not uniquely encoded by symbol strings @xmath79 , i.e. the partition is not generating .
in addition , there is strong pruning , i.e. some of the possible symbols strings are not realized by trajectories of the map .
the symbols defined by the map ( [ pm ] ) do , however , reflect the important contributions of the dynamics to a semiclassical description as will be seen in the next section .
the problem of finding a `` good '' symbolic dynamics for the stadium @xcite might be a consequence of this fundamental dilemma .
the bouncing ball limit @xmath80 can be reached only for @xmath81 or 1 , but for all @xmath82 on the poincar surface of section .
the opposite limit @xmath83 , ( the whispering gallery limit ) , is possible only for @xmath84 or @xmath85 and a decreasing @xmath82 interval of starting points .
together , there are four infinite series of fixed points of the map which are present for all parameter values @xmath13 . in the @xmath86 case , @xmath36 must be even and positive .
the corresponding periodic orbit family @xmath87 , see fig .
[ fig : po]a , has initial conditions @xmath88 and @xmath89 note that @xmath90 grows linearly with the length @xmath91 and thus with the order parameter @xmath92 which is in contrast to exponential growth for strictly hyperbolic dynamics .
= 15.0 cm the family with @xmath93 approaching the bouncing ball orbits are formed by periodic orbits which start in a corner of the rectangle and bounce off perpendicular to the bottom line in the circle , see fig .
[ fig : po]b . in this case
, @xmath36 must be odd and negative , i.e. @xmath94 .
the angle of incidence @xmath70 for the bounce in the circle fulfills the condition @xmath95 with approximate solution @xmath96 the starting conditions on the poincar surface are @xmath97 one obtains for the length of the periodic orbits @xmath98 and for the weight @xmath99 ( see also appendix [ app : a ] ) . again , the determinant increases linearly with the length of the periodic orbits .
an analysis of the periodic orbits @xmath100 and @xmath101 approaching the whispering gallery limit @xmath102 is provided in appendix [ app : b ] . the map ( [ pm ] ) can be considerably simplified both in the bouncing ball limit and in the whispering gallery limit . the latter is postponed to appendix [ app : b ] . for trajectories
@xmath103 , one has to distinguish between four different cases : ( @xmath104 , even ) , ( @xmath105 , even ) , ( @xmath106 , odd ) and ( @xmath107 , odd ) .
one obtains for ( @xmath104 , even ) @xmath108 + { \cal o}(m^{-4})\\ \label{phibb } \phi'_m & = & \frac{\pi}{2 } - \frac{2 a}{d_m } ( 1 - 2 \overline{x } ) \left [ 1 - \frac{4}{3 } \frac{a^2}{d_m^2 } \frac{1 - \overline{x}(3 - \overline{x}^2)}{1 - 2 \overline{x}}\right ] + { \cal o}(m^{-5})\\ \nonumber \mbox{with } & & \\
\nonumber d_m & = & 2a \tan\phi = m - x + \overline{x}. \end{aligned}\ ] ] the length of a trajectory is @xmath109 with @xmath110 .
note that @xmath111 depends at leading order on @xmath112 only ( and not independently on both @xmath82 and @xmath74 ) . for the transfer operator ( [ bot ] ) , we need the length of a trajectory as function of the initial and final points @xmath82 and @xmath111 .
one obtains @xmath113 + { \cal o}(m^{-4 } ) \\
\label{lxx2 } & = & \sqrt{d_m^2 + ( 2 a)^2 } + x ' - \frac{x'}{1 + 2 x ' } + 2 + { \cal o}(m^{-4 } ) \qquad x , x ' \in [ 0,1]\\ \label{dxx } \mbox{with } & & d_m(x , x ' ) = m - x + \frac{x'}{1 + 2 x ' } .\end{aligned}\ ] ] the length of a trajectory after one iteration of the map is thus given by its length in the rectangle plus corrections which depend on @xmath111 only ( up to @xmath114 ) .
the mixed second derivatives of @xmath115 showing up in the transfer operator ( [ bot ] ) are @xmath116 the length spectrum for the poincar map with @xmath36 odd or negative is obtained by the following replacements in ( [ lxx1 ] ) ( [ dlxx ] ) : + [ cols= " < , < , < , < , < " , ] the determinant @xmath117 is analytic in each strip @xmath118 m\pi , ( m+1)\pi[$ ] , with @xmath36 integer , but has a cusp at @xmath119 .
the non analytic behavior is introduced through infinitely long trajectories contributing in leading order at @xmath120 , see eqn .
( [ stsol ] ) .
these cusps are a consequence of omitting the bouncing ball contributions , which itself exhibit a cusp at integer multiples of @xmath121 , see eqn .
( [ inttr ] ) . the full spectral determinant @xmath44 ( [ specfak ] ) is analytic and nonanalytic behavior in the bouncing ball contributions is canceled by the near bouncing ball dynamics
. of special interest is , however , the maximal length of trajectories necessary to resolve the regular part of the quantum spectrum at a given wave number @xmath32 . inserting ( [ ebklim ] ) in ( [ stsol ] ) leads to an estimate for the trajectory lengths contributing dominantly to the ground state @xmath122 in each @xmath123 series , i.e. to the state located next to the cusp @xmath124 .
we obtain @xmath125 this is in contrast to general semiclassical arguments for bound systems leading to a cut
off for periodic orbits sums at half the heisenberg time @xcite .
this transforms for billiard into a cut off in the length spectrum according to @xmath126 where @xmath127 is the mean level density and @xmath28 the area of the billiard .
the estimate ( [ l_est ] ) scales differently with the billiard parameter @xmath13 and deviates thus especially in the `` integrable '' limit @xmath128 . the cut off ( [ t_h ] ) as derived in ref .
@xcite is based on two main assumptions : the overall validity of the gutzwiller periodic orbit formula and the analyticity of the semiclassical spectral determinant in a strip containing the real energy axis . both assumptions which might be intimately related for bound systems
fail here .
note , that this is not necessarily true for scattering systems .
a semiclassical quantization of two examples showing intermittency , the helium atom @xcite and hydrogen in a constant magnetic field for positive energy @xcite , could be achieved within the gutzwiller approach .
however , the intermittent part of the dynamics introduces in these cases as well regular structures in the resonance spectrum and non analyticity in the spectral determinant .
the threshold ( [ thres0 ] ) can be interpreted as the boundary of a region in phase space surrounding the marginal stable bouncing ball family .
semiclassical contributions from the dynamics in this region have a form similar to the one obtained for stable islands @xcite or in integrable systems @xcite .
the threshold values can be directly translated into momenta in phase space , i.e. @xmath129 with @xmath130 , see eqn .
( [ pm ] ) . a _ semiclassically stable island _ can thus be defined covering a phase space volume @xmath131 with @xmath132 the size of the semiclassical island approaches zero in the limit @xmath133 ( compared to the volume of the full phase space ) .
the number of quantum states associated with this island thus increases like @xmath134 on average .
this estimate coincides with the average increase of regular states given by the ebk quantization condition ( [ ebk ] ) .
it exceeds previous results by oconnor and heller @xcite , who obtained a linear increase in @xmath32 for the number of localized bouncing ball states up to the semiclassical limit @xmath135 .
note , that a @xmath136 increase in the number of bouncing ball states is still consistent with the schnirelman theorem @xcite , as the fraction of regular states compared to all eigenstates approaches zero in the semiclassical limit .
the oscillating part of the level staircase function @xmath137 shows a strong periodic modulation in the stadium , see fig .
[ fig : nosc]a .
( @xmath138 denotes here the quantum level staircase function and @xmath139 its mean part given by the weyl formula @xcite ) .
this modulation coincides with the oscillating part of the bouncing ball contributions @xcite , see eqn .
( [ inttr ] ) .
it was shown in section [ sec : bb ] , that the bouncing ball family does not contribute to individual eigenvalues . from the point of view of individual states in the spectrum , the oscillatory behavior in the level staircase function
is caused by a periodic change in the spacings between neighboring eigenvalues .
the spacings are unaffected by contributions coming from the classical bouncing ball family .
taking the concept of a semiclassical stable island seriously , we expect the spectrum of the stadium billiard to be divided into two different subspectra .
the majority of states belong to the `` chaotic '' subspace formed by non localized eigenstates ( leaving aside the phenomenon of scaring along short unstable trajectories ) .
their number increases on average like @xmath140 and the volume of the billiard is @xmath141 .
the level statistics of this subspectrum is expected to follow the goe prediction of random matrix theory .
the statistics as well as the average level repulsion is then stationary , i.e. independent of @xmath32 , in the unfolded spectrum .
the level staircase function is expected to be structureless showing only `` statistical '' fluctuations around the mean value @xmath142 .
the regular subspectrum contains the eigenstates originating from a quantization of the bouncing ball island ( [ vbb ] ) .
the coupling of the dynamics in the `` semiclassically integrable '' region to the outer classical motion is weak compared to the mixing in the outer region itself .
the level repulsion between bouncing ball states and chaotic eigenstates is thus small compared to the coupling among non bouncing ball states itself . a possible structure in the level staircase function for the regular states which can be obtained from the ebk quantization condition ( [ ebk ] )
can therefore survive in the full spectrum .
this is indeed the case . the oscillating part of the ebk
level staircase function can be defined with the help of eqn .
( [ nbars ] ) , i.e. @xmath143 here , @xmath144 is given by the number of levels obtained from the quantization condition ( [ ebk ] ) up to a certain @xmath32 value .
the result is shown in fig .
[ fig : nosc]b , ( where the next to leading terms in @xmath145 have been fitted numerically ) .
the function @xmath146 shows exactly the same modulations as the full spectrum and coincides also with the oscillating part of the bouncing ball contributions .
we conclude , that _ the dominant oscillation in the level staircase function of the full spectrum is caused by a modulation in the density of bouncing ball states only_.
i have shown that the marginally stable bouncing ball family in the stadium billiard does not contribute to individual quantum eigenvalues .
the result is confirmed by a semiclassical quantization of the quarter stadium using bogomolny s transfer matrix technique in a representation which excludes the bouncing ball orbits explicitly .
the regular bouncing ball quantum states in the spectrum derive semiclassically from the near bouncing ball dynamics alone , and these states follow a simple quantization rule .
the trace of the transfer operator can be approximated by periodic orbit contributions , which , however , show a @xmath0dependent transition from gutzwiller to berry
tabor like behavior when approaching the bouncing ball family .
this leads to the concept of a semiclassical island of stability surrounding the marginal stable family in phase space .
the boundary of this region is explicitly @xmath0dependent and the phase space volume of the island shrinks to zero ( compared to the total volume ) in the semiclassical limes @xmath147 ( or @xmath148 ) .
the quasi ebk quantization formula can be associated with a quantization of the semiclassical stable island .
the periodic modulations in the level staircase function can be related to a periodic change in the density of bouncing ball eigenstates .
the results demonstrate that averaged dynamical properties like ergodicity and positive liapunov exponent are not sufficient to ensure the applicability of the gutzwiller trace formula .
the `` chaoticity '' of the stadium is in a semiclassical sense indeed @xmath0dependent , and the billiard is for small @xmath32 closer to an integrable system than to a hard chaos one .
the gutzwiller periodic orbit weights have to be modified in the whispering gallery limit as well , see appendix [ app : b ] .
an accumulation of periodic orbits towards a limiting cycle of finite length has also been found in other systems as e.g. in the cardioid billiard @xcite , in the wedge billiard @xcite and in the anisotropic kepler problem @xcite .
we expect that a careful treatment as outlined in the appendix will solve problems concerning a semiclassical quantization of these systems .
+ the results obtained here for the stadium billiard are expected to be generic for systems with mixed phase space structure ( however complicated by the existence of island chains and can tori surrounding the stable island itself ) .
the stable island influences the classical dynamics in the outer chaotic region by creating intermittency .
the regular regions appear to be larger than the actual size of the stable island due to the finite phase space resolution of quantum mechanics .
the findings explain in a natural way the existence of ebk quantum states associated with regions outside a stable island as found in @xcite .
the width of this semiclassically integrable layer is @xmath0 dependent .
a semiclassical quantization of stable islands as well as the behavior of localized wave functions on classical boundaries and the description of tunneling through dynamical separatrices @xcite will be sensitive to this behavior . the results derived here indicate a failure of the berry keating periodic orbit resummation @xcite for mixed systems due to the intermittency introduced by the stable regions .
the work presented here is restricted to semiclassical aspects .
the influence of intermittency on frobenius perron and related classical operators @xcite is so far best described by a so called ber
approximation @xcite .
the spectra of classical operators are directly related to classical @xcite and semiclassical sum rules @xcite as well as to spectral statistics @xcite in hard chaos systems .
the influence of intermittency on these results is still an open issue .
+ + when completing this article , i became aware of a recent work carried out by primack et .
@xcite . the authors discus diffraction in the sinai billiard by analyzing the fourier transformation of the true quantum spectrum in detail .
they could indeed relate all deviations from the gutzwiller trace formula to diffraction effects ( due to the concave boundaries in this billiard ) for some of the near bouncing ball orbits .
the results in @xcite indicate clearly , that the influence of intermittency as dicussed in this article can also be seen directly in the fourier spectrum .
+ : + i would like to thank debabrata biswas , predrag cvitanovi , bertrand georgeot , kai hansen , jon keating and mark oxborrow for stimulating discussions and useful comments on the manuscript .
the work was supported by the deutsche forschungsgemeinschaft .
the monodromy matrix used in section [ sec : t - oper ] describes the linearized motion near a classical trajectory in phase space in a plane perpendicular to the orbit and on the energy manifold . for billiard systems ,
this plane is spanned by the local displacement vectors @xmath149 , @xmath150 pointing perpendicular to the actual momentum of the trajectory .
an initial displacement is thus propagated according to @xmath151 where @xmath152 depends on the path of the underlying trajectory . for 2dimensional billiards , one obtains @xmath153 where @xmath154 denotes the time between two bounces at the boundary , @xmath91 is the length of the path and @xmath155 .
the contribution from a reflection at the boundary is @xmath156 where @xmath157 denotes the local curvature of the boundary , and @xmath70 is the angle between the orbit and the normal to the boundary . the @xmath32 dependence is scale invariant due to the transformation @xmath158 , @xmath159 and we may set @xmath160 in what follows
. bounces on straight lines in the stadium billiard are treated as free flights .
the angle of incidence @xmath70 is the same for all successive bounces of an orbit in the circle .
the length between two of these bounces is @xmath161 where @xmath14 denotes the radius of the circle . the monodromy matrix for @xmath162 free flights in the circle interrupted by @xmath76 successive reflections on the circle boundary is thus given by @xmath163
the whispering gallery limit of the poincar map ( [ pm ] ) is formed by trajectories @xmath164 with @xmath36 fixed and @xmath102 in the notation of section [ sec : classpm ] .
these orbits approach the boundary of the billiard with an increasing number of bounces in the circle . only @xmath84 and @xmath165 is possible in the limit @xmath102 .
i will consider here the case @xmath166 and @xmath167 .
the approximate whispering gallery map for @xmath168 follows by analogy .
there is one fixed point for each symbol pair @xmath169 with starting conditions @xmath170 the angle of incidence @xmath70 for reflections in the circle is @xmath171 the length of the corresponding periodic orbit approaches a constant , i.e.@xmath172 the monodromy matrix along the orbits starting on the poincar surface of section can be deduced from appendix [ app : a ] and is @xmath173 the periodic orbit weighting factor @xmath174 is thus @xmath175 the determinant and thus the largest eigenvalue @xmath176 of the monodromy matrix increase quadratically with the symbol index .
the main difference compared to the bouncing ball limit is the convergence of the period @xmath177 towards a finite value : the length of the billiard boundary .
the whispering gallery limit therefore introduces no intermittency .
the liapunov exponent of periodic orbits , @xmath178 diverges logarithmically and the boundary orbit @xmath179 is infinitely unstable .
i will show below that the isolated orbit condition is again violated when taking the trace of the transfer operator .
a naive summation of gutzwiller periodic orbit weights would overestimate the whispering gallery contributions considerably .
the breakdown of the stationary phase condition is caused here by the accumulation of periodic orbits , i.e. stationary phase points , near the billiard boundary .
+ the phase space area of starting conditions on the poincar surface for trajectories with the same @xmath162value is centered on the corresponding periodic orbit .
its size shrinks to zero in the limit @xmath102 both in the @xmath82 and @xmath74 coordinate .
the transfer operator in the whispering gallery limit is thus directly given by the jacobian matrix ( [ mwg ] ) in local coordinates @xmath180 , @xmath181 centered on the fixed points ( [ powg ] ) .
the orbit length is approximately given by the quadratic form @xmath182 with @xmath183 where @xmath177 denotes the length of the periodic orbit .
when taking the trace of @xmath61 we are interested in trajectories satisfying @xmath184 which demands @xmath185 .
more important is , however , the restrictions on the @xmath186 interval , i.e.@xmath187 the interval length and thus the integration range decreases like @xmath188 .
the whispering gallery contribution to the trace of the @xmath61 operator are @xmath189 where the upper and lower limits in ( [ bwg ] ) have been approximated by their mean value @xmath190 the interval length decreases faster than ( the square root of ) the prefactor in the exponent for increasing @xmath162 .
a stationary phase approximation is not justified for @xmath102 , and we obtain ( as in the bouncing ball limit ) a transition behavior @xmath191 note , that the number of periodic orbits which can be treated as isolated stationary phase points , increases only at a rate proportional to @xmath192 .
the contributions for large @xmath162 fall off like @xmath188 and thus faster than in the bouncing ball case .
in addition , there is a cancelation between the @xmath100 and the @xmath101 family which show the same overall behavior but appear with opposite signs .
the whispering gallery contributions are thus small compared to near bouncing ball contributions , at least for @xmath193 . 99 m. c. gutzwiller , _ chaos in classical and quantum mechanics _
( springer , new york , 1990 ) .
r. balian and c. bloch , ann
. pays .
( n.y . ) * 69 * , 75 ( 1972 ) ; * 85 * , 514 ( 1974 ) . m. v. berry and m. tabor , proc .
london a * 349 * , 101 ( 1976 ) .
m. v. berry and m. tabor , j. phys a * 10 * , 371 ( 1977 ) .
o. bohigas , s. tomsovic and d. ullmo , phys . rep .
* 223 * , 44 ( 1993 ) .
d. wintgen and k. richter , comm . at
. mol .
* 29 * , 261 ( 1994 ) .
a. m. ozorio de almeida and j. h. hannay , j. phys .
a * 20 * , 5873 ( 1987 ) .
s. tomsovic , m. grinberg and d. ullmo , phys .
, 4346 ( 1995 ) .
k. richter , d. ulmo and r. jalabert phys .
( in press ) ; k. richter , d. ulmo and r. jalabert , _ disorder effects in ballistic microstructures _ , submitted to phys .
b ( 1996 ) .
m. sieber , j. phys .
a , * 29 * , 4715 ( 1996 ) . s. c. creagh , ann .
* 248 * , 60 ( 1996 ) . preprint ( 1996 ) .
a. shudo and k. s. ikeda , phys .
* 74 * , 682 ( 1995 ) .
e. doron and s. d. frischat , phys .
lett . * 75 * , 3661 ( 1995 ) .
h. g. schuster , _ deterministic chaos _ ( vch , weinheim , 1989 ) .
l. a. bunimovich , comm .
phys . * 65 * , 295 ( 1979 ) ; see also l. a. bunimovich chaos * 1 * , 187 ( 1991 ) . s. w. mcdonald and a. k. kaufman , phys .
lett . , 1189 ( 1979 ) ; phys . rev .
a * 37 * , 3067 ( 1988 ) .
g. casati and f. valz gris , and i. guarnieri , lett .
nuovo cimento * 28 * , 279 ( 1980 ) .
o. bohigas , m. j. giannoni and c. schmit , phys .
52 * , 1 ( 1984 ) .
e. j. heller , phys .
lett . * 53 * , 1515 ( 1984 ) .
stckmann and j. stein , phys .
lett . , 2215 ( 1990 ) ; * 68 * , 2867 ( 1992 ) . h.d .
, h. l. harney , h. lengeler , c. h. lewenkopf , c. rangacharyulu , a. richter , p. schardt and h. weidenmller , phys . rev .
lett . * 69 * , 1296 ( 1992 ) .
m. sieber , u. smilansky , s. c. creagh and r. g. littlejohn , j. phys .
a * 26 * , 6217 ( 1993 ) .
d. alonso and p. gaspard , j. phys .
a * 27 * , 1599 ( 1994 ) .
m. v. berry and j. p. keating , j. phys .
a * 23 * , 4839 ( 1990 ) ; m. v. berry and j. p. keating , proc .
r. soc .
a * 437 * 151 ( 1992 ) .
a. voros , j. phys .
a * 21 * , 685 ( 1988 ) .
e. b. bogomolny , nonlinearity * 5 * , 805 ( 1992 ) .
a , voros , commun .
phys . * 110 * , 439 ( 1987 ) .
j. p. keating and m. sieber , proc .
r. soc .
a * 447 * , 413 ( 1994 ) .
j. p. keating and m. v. berry , j. phys .
a * 20 * , l1139 ( 1987 ) .
p. a. boasman , nonlinearity * 7 * , 485 ( 1994 ) ; phd thesis _ semiclassical accuracy for billiards _ , university of bristol ( 1992 ) .
b. eckhardt und e. aurell , europhys .
* 9 * , 509 ( 1989 ) .
p. cvitanovi , phys .
* 61 * , 2729 ( 1988 ) ; p. cvitanovi and b. eckhardt , phys .
lett . , 823 ( 1989 ) .
g. tanner , p. scherer , e. bogomolny , b. eckhardt and d. wintgen , phys .
lett . * 67 * , 2410 ( 1991 ) .
m. sieber and f. steiner , phys .
lett . * 67 * , 1941 ( 1991 ) . g. tanner and d. wintgen , phys
lett . * 75 * , 2928 ( 1995 ) .
g. tanner , k. t. hansen and j. main , _ the semiclassical resonance spectrum of hydrogen in a constant magnetic field _
, to be published in nonlinearity ( 1996 ) .
a. wirzba and m. henseler , unpublished , darmstadt , 1995 ; + http://crunch.ikp.physik.th-darmstadt.de/~wirzba/missing_link.html ; further details can be found in m. reed and b. simon , _ methods of modern mathematical physics _ ,
i , chap vi , academic press ( new york ) , 1972 ; m. reed and b. simon , _ methods of modern mathematical physics _ , vol .
iv , chap xiii , academic press ( new york ) , 1976 .
b. lauritzen , chaos * 2 * , 409 ( 1992 ) .
o. biham and m. kvale , phys .
a * 46 * , 6334 ( 1992 ) .
k. t. hansen , phd thesis , university of oslo , ( 1993 ) .
b. georgeot and r. e. prange , phys .
* 74 * , 2851 ( 1995 ) .
t. szeredi , j. h. lefebvre and d. a. goodings , nonlinearity * 7 * , 1463 ( 1994 ) .
m. r. haggerty , phys .
e * 52 * , 389 ( 1995 ) .
o. bohigas , s. tomsovic and d. ullmo , phys .
lett . * 64 * , 1479 ( 1990 ) .
h. p. baltes and e. r. hilf , _ spectra of finite systems _ , mannheim : bibliographisches institut ( 1976 ) .
p. w. oconnor and e. j. heller , phys .
* 61 * , 2288 ( 1988 ) .
a. i. schnirelman , usp
. mat .
* 29 * , 181 ( 1974 ) ; y. colin de verdiere , commun .
102 * , 497 ( 1985 ) ; s. zelditch , duke math . j. * 55 * , 919 ( 1987 ) .
a. bcker and h. r. dullin , _ symbolic dynamics and periodic orbits for the cardiod billiard _ , preprint desy report desy 95 - 198 ( november 1995 ) .
h. bruus and n. d. whelan , nonlinearity * 9 * , 1023 ( 1996 ) .
g. tanner and d. wintgen , chaos * 2 * , 53 ( 1992 ) ; g. tanner and d. wintgen , chaos , solitons & fractals * 5 * , 1325 ( 1995 ) .
cvitanovi und b. eckhardt , j. phys .
a * 24 * , l237 ( 1991 ) .
p. cvitanovi and g. vattay , phys .
lett . * 71 * , 4138 ( 1993 ) .
p. dahlqvist , j. phys .
a , * 27 * , 763 ( 1994 ) ; p. dahlqvist , nonlinearity , * 8 * , 11 ( 1995 ) . j. h. hannay and a. m. ozorio de almeida , j. phys a * 17 * , 3429 ( 1984 ) .
p. cvitanovi , k. hansen and g. vattay , _ beyond the periodic orbit theory _ , preprint ( 1996 ) .
m. v. berry , proc .
r. soc .
* 400 * , 229 ( 1985 ) .
a. v. andreev and b. l. altshuler , phys .
, 902 ( 1995 ) ; o. agam , b. l. altshuler and a. v. andreev , phys . rev .
lett . * 75 * , 4389 ( 1995 ) ; a. v. andreev , o. agam , b. d. simon and b. l. altshuler , phys .
lett . * 76 * , 3947 ( 1996 ) .
e. b. bogomolny and j. p. keating , phys .
* 77 * , 1472 ( 1996 ) .
h. primak , h. schanz , u. smilansky and i. ussishkin , _ diffraction effects in the quantization of concave billiards _ , preprint , ( 1996 ) .
|
the impression gained from the literature published to date is that the spectrum of the stadium billiard can be adequately described , semiclassically , by the gutzwiller periodic orbit trace formula together with a modified treatment of the marginally stable family of bouncing ball orbits .
i show that this belief is erroneous .
the gutzwiller trace formula is not applicable for the phase space dynamics the bouncing ball orbits .
unstable periodic orbits close to the marginally stable family in phase space can not be treated as isolated stationary phase points when approximating the trace of the green function .
semiclassical contributions to the trace show an @xmath0 dependent transition from hard chaos to integrable behavior for trajectories approaching the bouncing ball orbits .
a whole region in phase space surrounding the marginal stable family acts , semiclassically , like a stable island with boundaries being explicitly @xmath0dependent .
the localized bouncing ball states found in the billiard derive from this semiclassically stable island .
the bouncing ball orbits themselves , however , do not contribute to individual eigenvalues in the spectrum . an ebk like quantization of the regular bouncing ball eigenstates in the stadium can be derived .
the stadium billiard is thus an ideal model for studying the influence of almost regular dynamics near marginally stable boundaries on quantum mechanics .
this behavior is generically found at the border of classically stable islands in systems with a mixed phase space structure .
+ pacs numbers : 05.45 , 03.65.sq .3 cm .3 cm .0 cm .0 cm 23.0 cm epsf
|
the dynamics of random walkers diffusing in the presence of a finite density of perfect absorbers is a rich problem which has been widely discussed in the physical and mathematical literature @xcite . at very large times
, the density of surviving walkers does not decay exponentially as a simple mean - field argument would predict , but rather behaves as @xmath8 , \label{don}\ ] ] where @xmath1 is the absorber density , and @xmath9 is a numerical constant .
the physical interpretation is that the process is dominated by particles starting in very large absorber - free regions ( voids ) , of linear size @xmath10 . in @xmath2 dimensions ,
these regions have a probability of order @xmath11 for small @xmath1 . in a void of size @xmath10 , solving the diffusion equation with absorbing conditions on its surface shows that the density typically decays as @xmath12 . a saddle - point argument
then leads to the result of eq .
( [ don ] ) , with the relevant regions being of typical size @xmath13 , at time @xmath14 .
another important question is the determination of the penetration or screening length @xmath0 , which measures the average distance between the starting point and the absorption point . in the limit of a small density of
uniformly distributed perfect absorbers of radius @xmath15 , and for @xmath3 , a classic result @xcite states that @xmath16 however , a simple heuristic argument casts doubts on the validity of eq .
( [ faux ] ) . in @xmath2 dimensions , let us consider a hypercubic box of linear size @xmath17 , which is the typical distance between absorbers .
we place one absorber of radius @xmath15 in this box , and impose periodic boundary conditions , which is equivalent to copying the box periodically .
we now release a random walker at time @xmath18 . to estimate the typical time when the walker will hit the absorber , we partition our box in smaller boxes of size @xmath19 .
the random walker will be absorbed with a finite probability once it has visited most of these small cells , including in particular the one containing the absorber .
we define @xmath20 as the number of different sites visited by a discrete random walker after a time @xmath14 .
hence , a fair estimate of the absorbing time @xmath21 is given by @xmath22 during this time , the random walker has traveled a typical distance @xmath0 given by @xmath23 where @xmath24 is the diffusion constant .
we can use the classical estimates for @xmath20 @xcite , which read @xmath25 finally , combining eqs .
( [ eu1],[eu2],[eu31],[eu32],[eu33 ] ) , we obtain the qualitative estimates @xmath26 in the present work , we will justify on more solid theoretical grounds the results of eqs .
( [ eu41],[eu42],[eu43 ] ) . in section [ sec2 ]
, we introduce a general formalism in order to compute @xmath0 and the distribution of distances traveled before absorption . in section [ sec3 ] , we define an exact renormalization group for the green s function on a lattice .
this recursion is solved using a mean - field ( or cavity - like ) approximation and our result confirms the estimates of eqs .
( [ eu41],[eu42],[eu43 ] ) . in the limit @xmath27
, we also address the effect of imperfect absorbers , which becomes relevant in @xmath3 . in section [ sec4 ] , we extend this approach to the continuum .
we compute the distribution of the distances of absorption and its first moments .
the theory is found to be in excellent agreement with numerical simulations in @xmath7 and @xmath28 . in section [ correla ] , we apply our renormalization approach to the case of a strongly correlated distribution of absorbers , characterized by a power law decay of the absorber density correlation function , @xmath29 , up to the correlation length @xmath5 . for @xmath30 and @xmath31 ( and for @xmath32 , in @xmath6 ) , the penetration length is found to be of the same order as @xmath5 , @xmath33 .
however , for @xmath34 , correlation are weak enough , so that the uncorrelated result of eqs .
( [ eu41],[eu42],[eu43 ] ) is recovered .
this result is illustrated by the exact solution of the problem in @xmath6 , and exactly solvable toy models in higher dimensions .
in addition , we test our ideas on the strongly correlated distribution of active sites in directed percolation simulations performed at the critical point , in @xmath7 and @xmath28 .
random walkers absorbed by the active sites are found to have a screening length @xmath35 , whereas the uncorrelated result is recovered above the critical dimension @xmath36 , when correlations become irrelevant . as a conclusion of this section
, we extend the result of eq .
( [ don ] ) to the case of a strongly correlated density of absorbers . in section [ dla ] , we discuss the determination of the screening length @xmath0 for diffusion limited aggregation ( dla ) clusters .
we emphasize that the most common method to measure @xmath0 seems inappropriate and we propose an improved scheme .
finally , we give some heuristic arguments inspired by the results of the previous sections , leading to the estimate @xmath37 , for dla clusters of gyration radius @xmath38 .
in a @xmath2-dimensional space , we consider a brownian particle with diffusion constant @xmath24 .
let @xmath39 be a positive killing field such that if the brownian particle is a the position @xmath40 , then in the next time interval @xmath41 , it is killed with probability @xmath42 .
if the particle is not killed , it just keeps on diffusing .
we define @xmath43 as the probability density that , starting from @xmath40 , that the particle s last resting place , _ i.e. _ where it is killed or absorbed , is at @xmath44 .
the quantity @xmath45 can be calculated by standard backward fokker - planck techniques .
we consider what happens in the first time step @xmath41 .
one possibility , occurring with probability @xmath42 , is that the particle is killed where it starts .
the other possibility is that it is not killed , this with probability @xmath46 , and so can then make a brownian jump @xmath47 having a component in the spatial direction @xmath48 , @xmath49 , obeying @xmath50 . putting this together gives @xmath51 expanding to order @xmath41 , and taking the expectation value over @xmath47 , we obtain @xmath52 the solution to eq .
( [ eqp ] ) is given by @xmath53 where @xmath54 is the green s function obeying @xmath55 note that the derivation of a probability density rather than a probability always requires a bit of care and our above derivation can be made more rigorous by defining an interval @xmath56 around the point @xmath44 then computing the probability that the particle is killed in @xmath56 then taking the limit @xmath57 .
one can check that @xmath58 is normalized as follows .
clearly the operator acting on @xmath54 in its defining equation eq .
( [ eqg ] ) is self - adjoint , which means that @xmath59 . integrating eq .
( [ eqg ] ) over all @xmath40 , and if the potential @xmath60 is sufficiently strong , the first term of eq .
( [ eqg ] ) will give an irrelevant surface term .
we thus obtain @xmath61 however @xmath54 is symmetric which ensures that @xmath62 and which demonstrates the correct normalization of @xmath58 . to further simplify this problem and reduce everything to the study of the green s function
, we write @xmath63 and consider @xmath60 as an operator , @xmath64 . using operator notation ,
@xmath58 is given by @xmath65^{-1 } v = [ h+v]^{-1 } [ h + v -h ] = i - [ h+v]^{-1 } h,\ ] ] which reads , in coordinate notation , @xmath66 now defining the disorder - averaged values of @xmath58 and @xmath54 by @xmath67 and @xmath68 respectively , we can write the averaged form of the above equation for a translational invariant distribution of absorbers @xmath69 if the disorder is isotropic , we will have @xmath70 , where @xmath71 , and likewise for @xmath68 .
the disordered averaged moments of the distance traveled before the particle is killed are more suitably obtained from @xmath72 , the fourier transform of @xmath67 defined as @xmath73 if we write the fourier transform of @xmath68 , @xmath74 as @xmath75 where @xmath76 is given by the one particle irreducible diagrams , we then obtain @xmath77 now , for small @xmath78 and to leading order , we expect that @xmath79 where @xmath80 is the inverse effective screening length of the averaged green s function and @xmath81 is the renormalization of the diffusion constant .
the resulting small @xmath78 expansion for @xmath72 is @xmath82 so if @xmath83 is the position where a particle released at the origin is observed , then the disorder averaged second moment of @xmath83 is simply given by @xmath84 the characteristic distance from the starting position , @xmath0 , at which the particle gets absorbed is therefore the same as the screening length for the green s function .
we thus define @xmath85 on a discrete lattice , the same arguments apply and we find that @xmath86 where @xmath87 denotes the lattice laplacian .
the discrete fourier transform of @xmath67 is defined as @xmath88 and is given by @xmath89 { \tilde g}({\bf k}),\ ] ] on a @xmath2-dimensional cubic lattice of lattice spacing @xmath90 .
now , if we write @xmath91 we find the lattice result , analogous to eq .
( [ eqfg ] ) @xmath92 the leading order behavior of @xmath93 must be of the form @xmath94 thus yielding the lattice result @xmath95 which takes the same form as in the continuum .
interestingly , we find that @xmath0 is not affected explicitly by the renormalization of the diffusion constant @xmath81 .
in this section , we will consider a model on the lattice @xmath96 ( hence the lattice constant is @xmath90 ) with a potential @xmath97 given by @xmath98 where the @xmath99 s are absorbing sites , with strength @xmath100 , which are uniformly and independently ( in this first instance ) distributed among all lattice sites .
we denote by @xmath101 the total number of lattice sites .
we now estimate the renormalization of the green s function @xmath102 for a system with @xmath103 absorbing sites by the addition of an @xmath104-th absorbing site . a similar method has been introduced in @xcite to calculate the effective diffusion constant of a tracer particle in a medium composed of randomly placed scatterers .
we denote by @xmath105 the position of the newly added absorber . in operator notation , and introducing the discrete laplacian @xmath106 , the discrete version of eq .
( [ eqg ] ) for @xmath102 and @xmath107 leads to , @xmath108 eliminating @xmath109 , we find @xmath110 after multiplying by @xmath102 and using the explicit form of @xmath111 , we obtain in coordinates notation , @xmath112 if we set @xmath113 , we obtain a closed form expression for @xmath114 , which can be substituted into eq .
( [ recnew ] ) .
this leads to the following _ exact _ recurrence : @xmath115 let us first consider the limit where the absorbing sites kill the particle on contact with probability @xmath116 .
this is achieved by taking the limit @xmath117 and yields @xmath118 which describes the exact renormalization of the green s function due to the addition of a perfect absorber . if the perfect absorbers are independently distributed , the position @xmath105 is completely uncorrelated with the @xmath119 s , for @xmath120 .
averaging eq .
( [ eqg1 ] ) over the position @xmath105 , but ignoring the correlation between the numerator and the denominator in the second term ( a mean - field - like approximation ) , we obtain @xmath121 we now perform the average over the remaining particle positions . using the statistical translational invariance of the system ,
we find @xmath122 taking the fourier transform of this equation yields @xmath123 defining the density @xmath124 , we obtain a differential equation for the evolution of @xmath125 in this approximation @xmath126 where @xmath127 is self - consistently given by @xmath128 the solution to eq .
( [ eqdif],[eqdifsc ] ) with the correct boundary conditions is @xmath129 where @xmath130 obeys @xmath131^{-1}.\label{des}\ ] ] as expected , the defining equation for @xmath130 does not depend on @xmath24 for perfect absorbers .
now using the fact that @xmath132 , we can integrate eq .
( [ des ] ) to obtain @xmath133 for small @xmath134 , we can use standard results for the integral in eq .
( [ des ] ) @xcite .
the results depend on the dimensionality and the lattice structure . * * @xmath6 * : substituting in the small @xmath134 behavior of the integral in eq .
( [ des ] ) @xcite , we find @xmath135 for small @xmath1 , integrating this gives , @xmath136 which then leads to @xmath137 this is clearly the correct scaling in @xmath1 , which will be recovered when solving exactly the one dimensional case ( see section [ exact1d ] ) .
formally , the result @xmath138 holds for any dimension @xmath139 .
the physical interpretation is clear : the screening length is simply proportional to the mean distance between absorbers , a result which will only remain true in the absence of strong correlation between them . * * @xmath7 * : here we find @xcite @xmath140 and integrating this for small @xmath1 , we obtain @xmath141 hence , the screening length is @xmath142 * * @xmath3 * : here , the leading order is analytic in @xmath1 @xcite , and we find @xmath143 giving @xmath144 the term @xmath145 depends explicitly on the dimension and the lattice structure .
for example , for the three dimensional cubic lattice , one has @xmath146 @xcite .
note that we expect these results to be strongly modified if the absorbers positions are spatially correlated , a problem which will be addressed in section [ correla ] .
let us comment on the disagreement between the present results and the one of ref .
@xcite , presented in eq .
( [ faux ] ) . in @xcite ,
the author first treats the effect of a single absorber on the free green s function .
he then assumes that the total correction to @xmath147 is simply proportional to the number of absorbers .
this statement is in fact incorrect , as absorbers far away from the introduced random walker should have a negligible contribution .
indeed , the walker should be absorbed well before being able to visit the regions where they stand .
this is actually the effect of screening , that our renormalization approach effectively captures . in addition , eq .
( 18 ) in ref .
@xcite , which leads to the final result of eq .
( [ faux ] ) , does not make any sense in the small @xmath1 limit , and it seems that the opposite non physical limit was in fact taken . finally , we can extend our formalism to the case of imperfect absorbers , corresponding to a finite value of @xmath100 .
the penetration length @xmath0 is now expected to depend explicitly on the diffusion constant @xmath24 . using eq .
( [ eqg1l ] ) , we find that the green s function still takes the form of eq .
( [ propag ] ) , but with @xmath130 now satisfying @xmath148^{-1},\label{desnewl}\ ] ] with @xmath149 in one and two dimensions a random walk is recurrent and visits any position a large number of times .
if the density of absorbers is small enough , a particle will visit the first absorber encountered a large number of times before making an excursion sufficiently far away from this first absorber and visiting a region occupied by a different absorber .
this means that most particles are absorbed by the first absorber encountered and thus the effect of a finite @xmath100 should become irrelevant in @xmath150 , when @xmath27 . in @xmath6
, we find the explicit result @xmath151 which leads to @xmath152 hence , for @xmath153 , we recover the result of eq .
( [ resul1ds ] ) . in @xmath7 ,
( [ desnewl ] ) leads to @xmath154=\rho,\ ] ] or @xmath155 for @xmath156 , we recover the result of eq .
( [ resul2ds ] ) .
finally for @xmath3 , considering imperfect absorbers deeply affects the result of eq .
( [ resul3ds ] ) .
indeed , we obtain @xmath157 note that in all dimensions , we find that the penetration length is an increasing function of @xmath158 , as physically expected .
we consider a system where the particle performs continuous brownian motion in one dimension .
it makes perfect sense to take an absorbing potential of the form @xmath159 corresponding to point - like absorbers . here
, we consider the case where the random walker is absorbed with probability one at each absorbing site , that is to say the limit @xmath117 . without loss of generality ,
we set @xmath160 , since the final expression of @xmath0 for perfect absorbers can not depend on the value of @xmath24 , whatever the spatial dimension . as in the previous section ,
we apply the cavity approach to calculate the renormalization of the green s function by the addition of an extra absorber at @xmath161 .
we apply again the method of section [ sec3 ] , which leads to the recurrence equation @xmath162 where the continuous fourier transform of @xmath68 is defined as @xmath163 the solution of eq .
( [ dif1d ] ) is given by @xmath164 where @xmath134 satisfies @xmath165 which leads to @xmath166 .
hence , we recover the lattice result @xmath167 we also obtain the explicit form @xmath168 which gives the distribution of @xmath169 , the displacement from the starting position to the point of absorption , to be @xmath170 in fact , in this one dimensional case , the distribution of @xmath169 can be computed exactly . a random walker starting from @xmath171 ,
we denote by @xmath172 the closest absorbing site to the right and by @xmath173 the closest absorbing site to the left .
standard results on brownian motion @xcite tell us that the probability of hitting @xmath172 before @xmath173 is @xmath174 this means that the probability density function for @xmath169 before averaging over the disorder is simply @xmath175 the disordered averaged density @xmath176 is now given by @xmath177 where the angled brackets denote the average over the absorber positions @xmath15 and @xmath172 . if the absorbers are placed as a poisson point process with rate @xmath1 , then the probability density function of @xmath15 and @xmath172 is poissonian @xmath178 this yields @xmath179 showing that in this case the relevant length scale in indeed @xmath180 .
however , we note that the probability density function given by the mean - field renormalization method is not exact in one dimension . a generalization of this one dimensional model can be constructed as follows .
we take the distribution of lengths @xmath10 between the absorbers to be given by @xmath181 .
the average interval length is simply related to the density by @xmath182 the probability that the particle starts within an interval of length @xmath10 has the probability distribution function @xmath183 given that one is in this interval , the position within it is uniformly distributed .
this means that we can write @xmath184 and @xmath185 , where @xmath186 is uniformly distributed over @xmath187 $ ] .
we thus have @xmath188 now performing the disorder average and using the symmetry of the problem , we find @xmath189 using the explicit form for @xmath190 in eq .
( [ qq ] ) , we obtain @xmath191 as a check , we set @xmath192 and recover the result eq .
( [ eqpo ] ) , obtained for the memoryless distribution of poissonian absorbers .
the above method of calculation also allows for a straightforward computation of the moments of @xmath193 which can be conveniently expressed in terms of the moments of the distribution @xmath181 .
we find @xmath194 in particular , we have @xmath195 which gives @xmath196 for uniformly distributed absorbers associated to a poissonian distribution of intervals . in this section
, we compute the behavior of the distance traveled before absorption in two dimensions and above . in this case , we take the absorbers to be spheres of radius @xmath15 . as
before , we restrict ourselves to the limit where the density of absorbers @xmath124 is small . as in the case of the lattice system , we denote by @xmath102 the green s function in the presence of @xmath103 absorbers and by @xmath107 the green s function obtained when an extra absorber is placed at the point @xmath105 which is uniformly distributed in the volume @xmath198 , independently of the other absorbers .
we denote by @xmath199 the potential due to the @xmath104-th absorber which takes the form @xmath200 where @xmath100 has the same dimension as @xmath24 divided by a distance .
@xmath111 is absorbing with strength @xmath100 on the _ surface _ of the sphere of radius @xmath15 centered at the point @xmath105 . in the limit @xmath201
, the absorber is a perfect absorber . combining eq .
( [ eqg ] ) written for @xmath102 and @xmath107 , we can find the equation for @xmath107 in operator notation as @xmath202 which can be rewritten as @xmath203 the corresponding equation on the lattice was easy to solve as the potential was given by a delta function .
however , here we must resort to a further approximation : we assume that we can set @xmath204 and @xmath205 in the above integral over the surface of a sphere centered at @xmath105 . we obtain @xmath206 where @xmath207 is the area of a sphere of unit radius in @xmath2 dimensions .
we can not make the same approximation for the integral in the denominator as the resulting term would be proportional to @xmath208 which is finite on a lattice , but diverges in the continuous case for @xmath197 .
the finite radius of the absorbing spheres regularizes the result .
now , in the limit of large @xmath100 , we obtain the expression for the renormalization group flow @xmath209 where @xmath210 following the same line of arguments as in the lattice case , we find that the disorder averaged green s function obeys @xmath211 where we have used the spherical symmetry of the disorder averaged green s function . for @xmath160 ,
the solution to this equation is @xmath212 where @xmath134 obeys @xmath213 and where @xmath214 has to be computed self - consistently . * @xmath7 : in two dimensions
, we have @xmath215 where @xmath216 is the bessel function of the second kind of order 0 @xcite .
we thus have @xmath217 assuming that @xmath134 is sufficiently small , _
i.e. _ for sufficiently small @xmath1 , we can use the small argument asymptotic form of @xmath218 @xcite , @xmath219 , to obtain @xmath220 where a cut - off of order @xmath221 naturally arises .
this equation can be integrated up to the leading order , yielding @xmath222 we thus find the same functional form as in the lattice case .
after some algebra , we find the following results for the lowest order moments : @xmath223 and @xmath224 the above calculation actually gives the full distribution for @xmath83 . if @xmath225 , then the probability density function for @xmath38 is given by @xmath226 where @xmath80 is given by eq .
( [ eqm3d ] ) . in the case of imperfect absorbers ( finite @xmath100 )
, the present results are not affected provided that @xmath1 is small enough ( @xmath227 ) .
+ in fig .
[ fig2d ] , we plot the two first moments of @xmath83 , as found from numerical simulations . we find a perfect agreement with eqs .
( [ y12d],[y22d ] ) , indicating that the theoretically predicted constant prefactors of the @xmath1 dependence are in fact probably exact in the small density limit . numerically ,
in order to access to the low density regime , we use the following algorithm which has been introduced in the context of dla @xcite . before performing the next move of our brownian walker
, we look for the nearest absorber ( by inspection of a grid keeping track of the coarse - grained absorber density ) , say found at the distance @xmath228 .
we then deposit the random walker anywhere on the circle of radius @xmath229 centered at its current position .
indeed , the first position where the walker would cross the perimeter is uniformly distributed on the circle .
the random walker is absorbed when the new distance to its nearest absorber is @xmath230 , where @xmath231 is small enough ( typically , we took @xmath232 , and @xmath90 ) .
note that if we were to actually simulate the random walker trajectory before it reaches the perimeter , the program would take a running time larger by a factor @xmath233 , where @xmath41 is the time increment .
when accessing to density of order @xmath234 , and using a time step of order @xmath235 , this factor is of order @xmath236 , which gives an idea of the huge gain achieved by using this algorithm .
all the simulations performed in this paper use variants of this algorithm . *
@xmath28 : in this case , we find @xmath237 for small @xmath134 , eq .
( [ eqmas ] ) yields @xmath238 and we again see that the functional dependence is the same as for the lattice case .
the lowest order moments are given by @xmath239 and @xmath240 for imperfect absorbers , the result now depends on the diffusion constant @xmath241 the radial distribution function for @xmath83 takes the explicit form @xmath242 in fig .
[ fig3d ] , we plot the two first moments of @xmath83 , as found from numerical simulations .
we find a perfect agreement with eqs .
( [ y13d],[y23d ] ) , indicating that these expressions are again probably exact in the small density limit . + in fig .
[ figpr3d ] , we plot the numerical probability density distribution @xmath243 for @xmath244 , which compares very well with our analytical result of eq .
( [ prth ] ) .
* @xmath245 : in higher dimensions , we find the same qualitative behavior as in @xmath28 .
in particular , the screening length is @xmath246 and is amplified by a factor @xmath247 for imperfect absorbers .
in this section , we consider the case of a non uniform distribution of absorbers displaying long range density - density correlations . in one dimension , such a situation arises naturally when the distribution of intervals @xmath181 between absorbers decays as a power law up to a distance @xmath5 , which is much larger than the mean distance between absorbers @xmath248 .
hence , we take the typical form @xmath249 for @xmath10 larger than a small scale cut - off @xmath15 ( @xmath15 should not be confused with the size of absorbers , which is irrelevant for perfect absorbers in @xmath6 ; here we consider point - like absorbers ) .
we only consider @xmath250 the condition @xmath32 ensures that @xmath181 can be normalized even in the limit @xmath251 , whereas @xmath252 implies that the first moment diverges in this limit : @xmath253 since @xmath32 , @xmath5 is indeed much larger than the mean distance between absorbers for small @xmath1 .
if the intervals between absorbers are drawn independently with the distribution @xmath181 , the density correlation function @xmath254 can be computed exactly , through its laplace transform @xmath255 , which is simply related to the laplace transform of @xmath181 by the relation @xcite @xmath256 hence in real space , the connected density correlation behaves like @xmath257 now applying the general result of eq .
( [ lexact ] ) , we obtain @xmath258 in one dimension , we find that the screening length scales as the correlation length , which is much larger than the screening length obtained in the uniform case . in the next subsections
, we shall illustrate the fact that this very same result should apply in higher dimensions .
we consider a system where the absorbers are distributed via a physical process which leads to long - range spatial correlations between them .
typical examples are given by directed percolation ( see next subsection ) , percolation or dla , the latter being briefly discussed in section [ dla ] . in general , the presence of correlations makes the problem much more difficult to treat analytically . here
, we present a semi - phenomenological approach based on our mean - field renormalization method , which applies in the case where correlations manifest themselves as a clustering phenomena .
we define the spatial correlation function as @xmath259 where the normalized connected correlation function @xmath260 is assumed to behave qualitatively as @xmath261 where @xmath5 is the correlation length of the system .
we also assume that the system is isotropic .
a relationship between the density and @xmath5 is found by associating @xmath5 as the characteristic distance where the connected part of the above correlation function , the first term in eq .
( [ cf ] ) , is of the same order as the second term .
this gives @xmath262 we implicitly consider the case where @xmath5 is much bigger than the mean distance between absorbers , which scales as @xmath263 .
hence , we will assume from now that @xmath264 the regime of small density thus obviously corresponds to a regime where the correlation length is large , for instance near a continuous transition .
if the correlation is manifested by the formation of clusters , then the typical number of absorbers in a cluster is given by @xmath265 which is divergent as @xmath266 , since @xmath267 . if @xmath268 is the number of clusters in the volume @xmath198 , the total number of absorbers is given by @xmath269 , which leads to the cluster density @xmath270 this last result expresses the fact that the typical distance between clusters of absorbers is of order @xmath5 .
we thus expect to find large empty regions in the system , whose linear size is of order @xmath5 , which is , again , much bigger than the mean distance between absorbers . in what follows
, we will repeat our renormalization calculation but in terms of the cluster number . as a starting point
, we will use the approximation of eq .
( [ approx ] ) , where the potential @xmath60 is concentrated at the center of the clusters and will take the form @xmath271 which is simply proportional to the mean density of absorbers for a cluster whose center is at @xmath272 . for large @xmath100 ,
the flow equation in @xmath273 leads to the same functional form for @xmath68 as before , but the corresponding equation for @xmath274 is now @xmath275 we first consider the case @xmath276 . in the limit of small @xmath80 and large @xmath5
, we find @xmath277 where we have used @xmath278 .
using the relation @xmath279 , we finally obtain @xmath280 this homogeneous equation admits the obvious solution @xmath281 leading to @xmath282 where we have reintroduced the dependence on the absorber radius @xmath15 . in the case
where @xmath34 , the integral in the denominator of the right - hand side of eq .
( [ eqtr ] ) converges and we simply get @xmath283 or @xmath284 which is the result for uncorrelated absorbers . for @xmath285 or @xmath7 , the calculation above leads to logarithmic corrections in the expression of @xmath0 . considering the crudeness of our argument
, we do not believe it is worth detailing the nature of these corrections . in conclusion
, the present results suggest that the screening length for a correlated system is either the correlation length @xmath5 ( @xmath30 ) or the screening length found in the uncorrelated case ( @xmath34 ) .
this result can be expressed in a synthetic way by @xmath286 with logarithmic corrections for the uncorrelated result in @xmath7 , which were obtained analytically in section [ sec4 ] . in this section
, we will test the ideas presented above by considering a correlated distribution of absorbers generated by the active sites remaining at time @xmath14 , at the critical point of directed percolation @xcite .
let us briefly introduce directed percolation on a @xmath2-dimensional hypercubic lattice .
lattice sites are empty ( inactive ) or occupied by a particle ( active ) . at time @xmath14 , each site is visited in a parallel dynamics . if the site is occupied , the particle is copied on its @xmath287 neighbors with probability @xmath67 , or removed with probability @xmath288 . if at least one particle has been copied on a given site , this site is simply considered as occupied , and empty otherwise .
if @xmath67 is large enough , a finite stationary density of particles @xmath1 establishes itself for large time . on the contrary , for small enough @xmath67 , the density decays exponentially with time .
in fact , there exists a critical value @xmath289 for which the system is critical : the density decreases algebraically , and the spatial ( and temporal ) correlation length diverges with time .
this defines the critical exponents @xmath290 and @xmath291 @xmath292 and the density correlation function has the typical form introduced in eq .
( [ cf],[cfp ] ) , with @xmath293 we have performed extensive numerical simulations of directed percolation at the critical point and considered the active sites present at several fixed times .
when any of these times is reached , we stop the simulation and launch a large number of random walkers which are absorbed by the active sites .
we then measure @xmath294 and the screening length @xmath295 , which are found to be proportional .
finally , the directed percolation dynamics is resumed until the next sampling time is reached , permitting us to explore systems with smaller and smaller densities , and increasing correlation length .
we have performed our simulations in @xmath7 and @xmath28 , since our @xmath6 result being exact , there is no doubt that the relation @xmath35 should be satisfied in this case . for @xmath296 ( @xmath36 is the critical dimension above which mean - field theory becomes exact ) , we have @xmath30 . interestingly ,
the equality @xmath285 holds exactly in @xmath297 and above ( @xmath298 and @xmath299 ) , so that for @xmath300 , we recover the result @xmath301 , identical to the uncorrelated case .
this is comforting , as correlations are known to become irrelevant above the upper critical dimension .
for the dimensions of interest here , one has @xcite @xmath302 in fig .
[ figlperco ] , we plot @xmath303 as a function of @xmath1 , and find a fair agreement with our prediction @xmath304 .
the numerical data are consistent with a subleading correction of order @xmath305 ( the uncorrelated result ) , although the rather strong curvature observed in @xmath7 could be as well ascribed to subleading logarithmic corrections mentioned at the end of subsection [ gencor ] .
we have also measured the average void size @xmath100 .
we pick a point at random in space and determine the radius of the largest disk ( in @xmath7 ) or sphere ( in @xmath28 ) which does not contain any absorber . in @xmath6 ( see eq .
( [ lexactmom1 ] ) ) , this is exactly a measure of @xmath5 since @xmath306 in higher dimension , we expect this property to hold , since @xmath100 measures the typical distance between absorber clusters , as discussed in subsection [ gencor ] , which was found to be of order @xmath5 .
in fact , we postulate the more general result @xmath307 fig .
[ figlxiperco ] illustrates the very convincing linear relation found numerically between @xmath0 and @xmath100 , implying @xmath35 .
note that we have the obvious bound @xmath308 as a particle can not be absorbed in empty regions . in this section
, we consider a specific geometry in @xmath197 where the screening length can be computed exactly , confirming our general result of eq .
( [ lfinal ] ) .
let us start by describing the model in @xmath7 .
we consider one dimensional semi - infinite half - lines ( @xmath309 ) of absorbers separated by a distance @xmath10 drawn independently with the distribution @xmath181 .
random walkers start from @xmath310 , and the screening length @xmath0 is defined as the average depth reached by the walkers before being absorbed . in this context , @xmath0 can be also viewed as a penetration length , similar to the one defined for dla ( see section [ dla ] ) . a random walker penetrating a channel of width @xmath10
will be absorbed at a depth of order @xmath10 , the only relevant length scale .
actually , the random walker density @xmath311 in the tube can be computed exactly , by solving the system @xmath312 with the absorbing condition for @xmath309 @xmath313 and a constant input of walkers at the entrance of the tube @xmath314 using discrete fourier transform along the direction @xmath315 , we arrive at the result @xmath316 which can also be written as a cumbersome expression involving standard @xmath317 and @xmath318 functions .
the flux of particles deposited at the depth @xmath319 is @xmath320 which decays exponentially over the scale @xmath10 .
the average deposition depth is then @xmath321 if we have an array of such channels , the probability that a walker first enters a channel of width @xmath10 is @xmath322 if we neglect the process where a random walker leaves this first channel to be absorbed in an other one , we find @xmath323 where @xmath324 is the inverse of the absorber density @xmath1 .
if @xmath181 decreases rapidly , @xmath325 , and we find @xmath326 in this geometry , the absorbers are highly correlated since they accumulate on lines .
note that the correlation function averaged over angle initially decreases as @xmath327 , so that eq .
( [ tubular ] ) is fully consistent with our general result for @xmath328 . finally , if the width distribution has a power law decay as in eq .
( [ distrib1d ] ) , eq .
( [ ltubu ] ) leads to @xmath329 again in perfect agreement with our general result . in fig .
[ figtub ] , we present numerical simulations of this two - dimensional tubular system .
the data are in perfect agreement with eq .
( [ finalrez ] ) , showing that the processes involving particles leaving a tube to be absorbed in an other one do not affect our general result .
this model can be generalized in higher dimensions in the following way .
we consider linear tubes whose @xmath330-dimensional surface is perfectly absorbing .
a @xmath330-dimensional cut of the system has the structure of a network of @xmath330-dimensional `` rectangular '' cells of edge length @xmath331 , @xmath332 , ... ,
@xmath333 , which are drawn independently using the same probability distribution @xmath181 .
the absorbers are placed on the @xmath334-dimensional surface of these cells ( see fig .
[ figprod ] ) .
the density of absorbers is still @xmath335 the solution of the diffusion problem in each tube is simply @xmath336 which decays exponentially on the scale @xmath337 , with @xmath338 hence , neglecting again processes where a walker leaves the first tube it enters to be absorbed elsewhere , the average penetration length is @xmath339^{d-2}.\label{newl0}\ ] ] if @xmath181 decays rapidly for @xmath340 , this integral leads to @xmath341 however , if @xmath181 takes the form of eq .
( [ distrib1d ] ) , we find @xmath342^{d-2},\\ & \sim & \langle l\rangle^{-(d-1)}\xi^{(d-1)\alpha+1}\sim\xi,\end{aligned}\ ] ] where we have used the fact that @xmath343 .
we again find @xmath33 , in agreement with our general argument for strongly correlated absorbers .
in this section , we consider a system for which absorbers are placed at the intersections of the product of @xmath2 linear chains ( see fig . [ figprod ] ) .
each chain has its intervals drawn from the distribution @xmath181 .
the density of absorbers reads @xmath344 if the distribution @xmath181 decreases rapidly on the scale of @xmath324 , it is clear that this system will behave similarly to a uniform distribution of absorbers , leading to @xmath345 .
hence , we assume that the distribution of intervals is of the form @xmath346 we shall see below that @xmath30 will lead to @xmath33 , whereas the uncorrelated result is recovered for @xmath34 .
the correlation function can be exactly computed , since the density is the cross product of @xmath2 independent densities @xmath347 where @xmath348 is the one dimensional correlation function , which can be exactly computed in terms of the laplace transform of @xmath181 ( see eq .
( [ laplace ] ) ) . after performing an average over angle
, the correlation is found to qualitatively behave as @xmath349 in the present model , the average void size introduced in section [ secperco ] is defined as the radius of the largest hypercube that does not contain any absorbers , averaged over the position of the center of the hypercube .
a simple calculation leads to the generalization of eq .
( [ voids ] ) .
if the center @xmath350 ( @xmath351 $ ] ) is drawn randomly in a cell of size @xmath352 with all @xmath353 corners occupied by an absorber , the largest hypercube not containing any absorber has a radius @xmath354 in order to simplify our calculation , we replace the actual expression of @xmath355 , by @xmath356 which behaves essentially in the same manner as the original expression . since the average of @xmath355 over the position @xmath40 of the center is @xmath357 , we finally get @xmath358^{d-1}.\label{newl}\ ] ] interestingly , eq . ( [ newl ] ) is the same ( up to the @xmath359 factor ) as the expression obtained for @xmath0 in the previous subsection ( see eq .
( [ newl0 ] ) ) , except that @xmath360 is now replaced by @xmath2 .
hence , if @xmath30 , we conclude that @xmath361 in fact , our previous results suggest that @xmath362 again . finally , note that in an homogeneous correlated system , eq . ( [ newl ] ) shows that the distribution of the void linear sizes is typically of the form @xmath363 which is identical to the void size distribution @xmath364 obtained in @xmath6 ( see eq .
( [ qq ] ) )
. note that in @xmath3 , we can obtain a regime where the correlation length is large compared to the average distance between absorbers , but is still smaller than the uncorrelated screening length , @xmath365 this regime corresponds to values of @xmath366 satisfying , @xmath367 in this case , although there are long - range correlations in the system , our argument of section [ gencor ] predicts that the uncorrelated result @xmath368 should hold . in fig .
[ figlprod ] , we measure numerically @xmath0 as a function of @xmath5 for the model studied in the present section , and in @xmath28 . in perfect agreement with our general result of eq .
( [ lfinal ] ) , we find that @xmath33 for @xmath369 , whereas @xmath370 for @xmath371 .
note that in both cases , we find @xmath372 ( not shown ) . in this subsection , we address the generalization of eq .
( [ don ] ) for a strongly correlated distribution of absorbers in @xmath2 dimensions .
we shall adapt to the present problem the usual variational argument @xcite leading to eq .
( [ don ] ) .
note that the effect of short range clustering on the density of surviving random walkers has been studied quantitatively in @xcite .
imagine that one releases a random walker of diffusion constant @xmath24 in an hypercubic region of linear size @xmath10 not containing any absorber ( a void ) .
its survival probability is larger than the survival probability computed assuming that the boundary of the void is perfectly absorbing .
this probability behaves like @xmath373 , where @xmath374 is the lowest mode of the diffusion equation in the hypercubic domain delimited by the void , with absorbing boundary conditions on the surface .
all the @xmath2 coordinates of @xmath374 are equal to @xmath375 . by actually solving the diffusion equation mentioned above with a uniform initial position of the random walker ,
we obtain an exact bound of the form @xmath376 where @xmath377 is some constant which does not depend on @xmath10 finally , and after averaging over the void size distribution @xmath364 , we find an exact bound for the density of surviving walkers @xmath378 it is commonly conjectured that this kind of bound actually captures the correct asymptotic behavior of @xmath379 @xcite .
let us now assume a distribution of hypercubic void sizes of the form found in eq .
( [ voiddist ] ) .
@xmath380 in @xmath6 , or in the preceding section , we considered a cut - off function @xmath381 . here
, we will consider the more general case @xmath382 where the value @xmath383 is a natural example : @xmath384 is the volume of the void divided by the correlation volume , an equivalent of the term @xmath385 obtained in the usual uncorrelated case ( see eq .
( [ volimp ] ) ) .
dropping all unimportant numerical constants for the sake of clarity , we obtain @xmath386 applying a saddle - point argument , and dropping subleading corrections , we finally find that @xmath387 we obtain a stretched exponential decay like in the uncorrelated case , but more importantly , we find that the time scale @xmath388 over which the density decays is now controlled by the correlation length @xmath389 instead of the density @xmath1 , as obtained in the uncorrelated case @xmath390
there have been several attempts to measure the screening length of dla clusters @xcite of gyration radius @xmath38 @xcite . in this section ,
we show that the methods used so far _ do not _ effectively measure @xmath0 .
we will propose a theoretical estimate for @xmath0 as well as a possible numerical method in order to measure it properly .
first of all , let us briefly mention how dla clusters are grown in dimension @xmath197 .
one first places a seed at the origin , for instance a @xmath2-dimensional sphere of radius @xmath15 .
a random walker ( of same radius @xmath15 ) is launched from far away and wanders until it touches the seed , whereupon it sticks to it .
then , another random walker is released , which sticks upon contact with any of the quenched particles .
this process goes on , leading to the formation of fractal clusters of dimension @xmath391 ( and presumably , @xmath392 @xcite ) , defined as @xmath393 where @xmath103 is the number of particles in the cluster .
we immediately note that for an incoming random walker , the points at a distance @xmath394 from the already formed structure acts like absorbers .
hence , it is tempting to introduce a screening length @xmath0 @xcite , measuring how deep the random walkers penetrate the cluster before sticking to it .
however , contrary to the systems studied so far in this paper , which were infinite and homogeneous , a dla cluster is finite and its density from the center decays in average as @xmath395 the determination of @xmath0 is an important matter , being often an essential ingredient in the theoretical attempts to determine @xmath396 .
although mean - field theories are based on contradicting estimates for @xmath0 , typically @xmath397 @xcite or @xmath398 @xcite , they mostly consider that @xmath399 .
however , there is now convincing numerical evidence @xcite showing that @xmath400 at least in @xmath7 , and probably in @xmath28 . to understand this result
, it is worth mentioning the commonly used method to determine @xmath0 @xcite .
a cluster of large size @xmath38 is first grown .
then a large number of independent random walkers are released .
if one of them touches a particle of the cluster , it simply vanishes , and its last position is recorded .
the distribution of the positions of absorption is thus obtained and its standard deviation is considered as a measure of @xmath0 ( before or after averaging over many clusters having the same number of particles ) . although a certain consensus seems to exist in the literature , we do not believe that this variance is a faithful measure of @xmath0 . to see this ,
let us consider the simple example of a perfectly absorbing ellipse in @xmath7 of aspect ratio @xmath401 and main axis @xmath38 .
any method to measure @xmath0 should retrieve the trivial result that @xmath402 in this case .
however , the method presented above would obviously lead to @xmath403 ! it is correct that dla clusters grown in the continuum are statistically isotropic .
however , for a given cluster , the fluctuation of the length of the main branches are typically of size @xmath38 , leading to an automatic numerical evaluation of @xmath37 . in other words , even if the random walkers were only absorbed near the tip of these branches , physically implying a very small value of @xmath0 , the current method of estimating @xmath0 would invariably lead to @xmath404 . as an alternative method ( but which might prove as ineffective as the one above ) , we propose that one should measure the distance of penetration from the _ convex hull _ of the cluster , and to measure the variance of this length _ before _ averaging over many clusters .
at least , this method gives the correct result for the ellipse ( its own convex hull ) , that is , @xmath402 .
finally , inspired by the general conclusions of the previous sections , we would like to give three related arguments in favor of the result @xmath37 , which we think is correct , although it is argued that the available numerical simulations using the method exposed above are not conclusive on this matter . particles in a dla cluster are obviously strongly correlated , arranging themselves in highly ramified structures .
it is thus tempting to apply some of our result to this situation : * there is no finite correlation length in a dla cluster except , rather trivially , for the size of the cluster itself . for long - range correlations
, we have found that @xmath405 . * in a dla cluster , the density correlation function decays as @xmath29 , with @xmath406 .
our study suggests that @xmath407 ( see eq .
( [ rhodla ] ) ) . * since a dla cluster is fractal , the largest voids inside it are of typical linear size @xmath408 . in homogeneous structures , @xmath409 , which suggests again that for dla , @xmath410 .
in this paper , we have justified analytically the heuristic argument presented in the introduction to estimate the screening length in a system of uniformly distributed perfect or imperfect absorbers .
our results in @xmath7 and @xmath28 are in excellent agreement with numerical simulations .
even the numerical prefactors of the two first moments are surprisingly well described by our approach . as a further check
, we found that the theoretical distribution of the distances of absorption is in perfect agreement with numerical simulations . for correlated absorbers ,
our analytical approach is not as rigorous as in the uncorrelated case , although we supplemented our heuristic argument with the exact solution in @xmath6 as well as for two toy models in @xmath31 , which fully confirm our general results .
we find that if the density correlation function decays with an exponent @xmath366 ( @xmath30 in @xmath197 , and @xmath32 in @xmath6 ) up to the correlation length @xmath411 , then the screening length scales as @xmath412 , where @xmath100 is the average void linear size .
these results were confirmed numerically in directed percolation , where the active sites play the role of the absorbers .
finally , we have argued that for dla , the penetration or screening length @xmath0 should be of the same order as the linear size @xmath38 of a cluster . we have also emphasized that the current numerical method of determining @xmath0 , although finding just @xmath37 , could hardly lead to any other result . as a more appropriate method , we propose measuring the distance of penetration from the convex hull of the cluster , and to define @xmath0 as the standard deviation of this length before averaging over many clusters . with this definition ,
the behavior of @xmath0 as a function of the cluster radius @xmath38 is certainly a question worth investigating @xcite . b. ya .
balagurov and v. g. vaks , zh .
fiz . * 65 * , 1939 ( 1973 ) [ sov .
jetp * 38 * , 968 ( 1974 ) ] ; m. donsker and s. varadhan , commun .
pure appl . math .
* 28 * , 525 ( 1975 ) ; p. grassberger and i. procaccia , j. chem . phys . * 77 * , 6281 ( 1982 ) ; t. c. lubensky , phys .
a * 30 * , 2657 ( 1984 ) ; s. renn , nucl .
b * 275 * , 273 ( 1986 ) ; j. w. haus and k. w. kehr , phys . rep . * 150 * , 263 ( 1987 ) ; t. m. nieuwenhuizen , phys .
62 * , 357 ( 1989 ) .
|
we consider the penetration length @xmath0 of random walkers diffusing in a medium of perfect or imperfect absorbers of number density @xmath1 .
we solve this problem on a lattice and in the continuum in all dimensions @xmath2 , by means of a mean - field renormalization group . for a homogeneous system in @xmath3
, we find that @xmath4 , where @xmath5 is the absorber density correlation length . the cases of @xmath6 and @xmath7 are also treated . in the presence of long - range correlations
, we estimate the temporal decay of the density of random walkers not yet absorbed .
these results are illustrated by exactly solvable toy models , and extensive numerical simulations on directed percolation , where the absorbers are the active sites . finally , we discuss the implications of our results for diffusion limited aggregation ( dla ) , and we propose a more effective method to measure @xmath0 in dla clusters .
|
during its first six years in orbit , the _ fermi_large area telescope ( lat ; * ? ? ?
* ) has detected @xmath030 mev gamma - ray emission from more than 40 solar flares , nearly 10 times more than egret @xcite onboard the _ compton gamma - ray observatory _
, grs @xcite onboard the _ solar maximum mission _ ( _ smm _ ) and coronas - f @xcite .
the _ fermi_detections sample both the impulsive @xcite and the long - duration phases @xcite including the longest extended emission ever detected ( @xmath120 hours ) from the sol2012 - 03 - 07 goes x - class flares @xcite .
our understanding of solar flares has also been shaped by decades of hard x - ray ( hxr ) observations , notably by the detection of conjugate footpoint sources by _ smm _
@xcite , coronal sources above soft x - ray loops by _ yohkoh_(e.g .
* ; * ? ? ?
* ) , and double coronal sources suggestive of magnetic reconnection in between by _ rhessi_(e.g .
? * ; * ? ? ?
* ; * ? ? ?
these and many other observations support the standard flare model involving magnetic reconnection and associated particle acceleration in the corona ( for reviews , see , e.g. , * ? ? ?
there are alternatively proposed scenarios , including ( re-)accleration of particles in the chromosphere ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , supported by ( e.g. , * ? ? ?
* ) . of particular interest
are those flares whose footpoints are occulted by the solar limb , allowing coronal emission to be imaged in greater detail than in normal situations dominated by bright footpoints ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . in this letter
, we present _ fermi_and _ rhessi_observations of such a flare whose position was confirmed to be behind the limb by _ stereo_-b .
while gamma - ray emission up to tens of mev resulting from proton interactions has been detected before from occulted solar flares , the significance of this particular event lies in the fact that this is the first detection of @xmath0100 mev gamma - ray emission from a footpoint - occulted flare and presents a unique opportunity to diagnose the mechanisms of high - energy emission and particle acceleration in solar flares .
on 2013 october 11 at 07:01 ut a goes m1.5 class flare occurred with soft x - ray emission lasting 44 min and peaking at 07:25:00 ut .
figure [ fig : lightcurve ] shows the goes , _
stereo_-b , _ rhessi _ , _ fermi_gamma - ray burst monitor ( gbm ; * ? ? ?
* ) and lat lightcurves of this flare .
the lat detected @xmath0100 mev emission for @xmath130 min with the maximum of the flux occurring between 07:20:0007:25:00 ut .
_ rhessi_coverage was from 07:08:0007:16:40 ut , overlapping with _
fermi_for 9 min . images in figure [ fig : stereosdo ] from the _
stereo_-b extreme - ultra violet imager ( euvi ; * ? ? ?
* ) and the _ sdo_atmospheric imaging assembly ( aia ; * ? ? ?
* ) of the photosphere indicate that the active region ( ar ) was @xmath19.9 behind the limb at the time of the flare .
lasco onboard the _ solar and heliospheric observatory _
( _ soho _ ) observed a backside asymmetric - halo cme associated with this flare beginning at 07:24:10 ut with a linear speed of 1200 km s@xmath2 @xcite and a bright front over the northeast . both _ stereo _
spacecrafts detected energetic electrons , protons , and heavier ions including helium , as well as type - ii radio bursts indicating the presence of a coronal heliospheric shock .
_ stereo_-b had an unblocked view of the entire flare and detected a maximum rate of 3.5@xmath310@xmath4 photons s@xmath2 in its 195 channel , corresponding to a goes m4.9 class @xcite if it had not been occulted .
+ we performed an unbinned likelihood analysis of the lat data with the ` gtlike ` program distributed with the _
fermi_`sciencetools ` circular region centered on the sun and within 100from the local zenith ( to reduce contamination from the earth s limb ) . ] for _ rhessi_data , we applied the clean imaging algorithm @xcite using the detectors 3@xmath59 to reconstruct the x - ray images .
we used the fits world coordinate system software package @xcite to co - register the flare location between _ stereo _ and _ sdo_images .
the _ stereo _ light curves are pre - flare background subtracted , full - sun integrated photon rates . we measure the location of the lat @xmath0100 mev gamma - ray emission ( as described in * ? ? ?
* ) and find a best fit position for the emission centroid at heliocentric coordinates of ( @xmath6,@xmath7 ) with a 68% error radius of 250@xmath8 , as shown in figure [ fig : stereosdo](b ) .
_ rhessi _ x - ray sources integrated over 07:11:04@xmath507:16:44 ut are shown as 80%-level , off - limb contours in figure [ fig : stereosdo](d ) .
the temporal evolution of the projected _ rhessi_source heights above the solar limb are shown in figure [ fig : lightcurve](f ) .
the higher - energy emission generally comes from greater heights , consistent with expectations for a loop - top source ( e.g. , * ? ? ?
* ; * ? ? ?
moreover , from _
sdo_/aia movies we find no signature of euv ribbons , even in the late phase during the _ rhessi_night .
together , these observations provide convincing evidence that the footpoints were indeed occulted .
the lat measured 4 photons with energies @xmath01 gev and reconstructed direction less than 1@xmath9 from the center of the solar disk .
all , including a 3 gev photon , arrived after 7:19:00 ut ( outside of the rhessi coverage ) .
we fit the lat gamma - ray spectral data with three models .
the first two , a pure power - law ( pl ) and a power - law with an exponential cut - off ( plexp ) are phenomenological functions that may describe bremsstrahlung emission from accelerated electrons .
the third model uses templates based on a detailed study of the gamma rays produced from pion decay ( updated from * ? ? ?
we rely on the likelihood ratio test ( ts ; * ? ? ?
* ) to estimate the significance of the source ( ts@xmath10 ) as well as to estimate whether the addition of the exponential cut - off is statistically significant .
to this end we define @xmath11ts = ts@xmath12-ts@xmath10 which is equivalent to the corresponding difference of maximum likelihoods computed between the two models . the significance in @xmath13
can be roughly approximated as @xmath14 .
for each interval where plexp provides a significantly better fit than pl ( @xmath1520 ) we also fit the data with a series of pion - decay models to determine the best proton spectral index following the same procedure described in @xcite .
the ts values for plexp and pion - decay fits can not be directly compared @xcite however the plexp approximates the shape of the pion - decay spectrum ( see figure [ fig : sed ] ) thus we expect the pion - decay models to provide a similarly acceptable fit .
we studied the effect of the lat systematic uncertainties ( mainly from the effective area , as considered here ) via the bracketing technique described in @xcite .
the _ rhessi_and gbm nai@xmath16 spectral data are independently fitted with one or two thermal components plus a broken power - law with index fixed to 2 ( to avoid energy divergence ) below the break . table [ tab : latintervals ] summarizes the spectral analysis results .
data from bgo@xmath17 are analyzed using the procedure described in @xcite with an additional 5% systematic error on the background estimation .
figure [ fig : sed ] shows the combined spectra from _ rhessi _ , gbm and lat in four integration intervals .
the discrepancy ( up to a factor of 2.5 ) between the _
rhessi_and gbm flux values is likely due to pile - up in the _ rhessi_detector , and can not be easily corrected .
as is evident from figure [ fig : sed ] , more energy is radiated in hxrs than gamma rays .
+ lcccccc 07:10:0007:15:00 & 16 & 5 & [email protected]@xmath19@xmath20 & & & [email protected]@xmath21 + 07:15:0007:20:00 & 987 & 22 & [email protected]@xmath22 & 145@xmath1827@xmath23 & [email protected]@xmath180.1 & [email protected]@xmath24 + 07:20:0007:25:00 & 1146 & 92 & [email protected]@xmath25 & 162@xmath1826@xmath26 & [email protected]@xmath180.1 & [email protected]@xmath27 + 07:25:0007:30:00 & 435 & 45 & [email protected]@xmath28 & 95@xmath1821@xmath29 & [email protected]@xmath180.1 & [email protected]@xmath30 + 07:30:0007:35:00 & 55 & 4 & [email protected]@xmath31@xmath20 & & & [email protected]@xmath32 + 07:08:00 07:35:00 & 2885 & 233 & [email protected]@xmath33 & 147@xmath1815@xmath34 & [email protected]@xmath35 & [email protected]@xmath36 + 07:16:40 07:35:00 & 2855 & 204 & [email protected]@xmath37 & 155@xmath1816@xmath38 & [email protected]@xmath35 & [email protected]@xmath39 + + + & & & & + & & & & & & + & & & & & & + 07:08:0007:10:58@xmath40 & & [email protected] & [email protected] & [email protected] & & 52@xmath181 + 07:08:0007:10:58 & & [email protected] & [email protected] & [email protected] & & 55@xmath181 + 07:11:1007:16:40@xmath40 & & 16@xmath182 & [email protected] & [email protected] & [email protected] & 253@xmath185 + 07:11:1007:16:40 & & 21@xmath185 & [email protected] & [email protected] & [email protected] & 630@xmath1810 + 07:16:4007:35:00 & & 20 ( fixed ) & [email protected] & [email protected] & [email protected] & 399@xmath188 + 07:08:0007:35:00 & & 20@xmath188 & [email protected] & [email protected] & [email protected] & 388@xmath188 + [ tab : latintervals ]
we have analyzed the data of the 2013 october 11 solar flare from _ fermi _ , _ rhessi _ , _ sdo_and _ stereo_. _ stereo_-b images indicate that the flare occurred in an ar 9.9 behind the limb . _
rhessi_and gbm nai@xmath16 detected hxrs up to 50 kev from the flaring loop - top .
the most unusual aspect of this flare is the lat detection of photons of energies @xmath41100 mev for about 30 minutes with some photons having energies up to 3 gev .
electrons or protons with energies @xmath42 can produce these photons after traversing a column depth of matter @xmath43 and @xmath44 @xmath45 , respectively , which is much larger than the depth @xmath46 @xmath45 penetrated by hxr - producing electrons . for occulted flares
the emitted photons must traverse even larger depths where they may be scattered and absorbed .
we consider three scenarios for the emission site of the gamma rays ; ( i ) deep below the photosphere of the flare site ( ii ) in the corona above the limb , suggestive of trapping of particles , e.g. , by strongly converging magnetic fields and ( iii ) cme - shock accelerated particles traveling back to the sun .
for the first scenario we need continuous acceleration of particles because they penetrate deep into the solar atmosphere and lose energy in a fraction of a second .
most of the radiation they produce also comes from deep within the photosphere so we need to calculate the optical depth , @xmath47 from the emission site to the earth . for @xmath0100 mev photons
the main absorption is via pair production with a cross section @xmath48 , where @xmath49 is the thomson scattering cross section relevant for @xmath50100 kev hxrs and @xmath51@xcite . ] .
the column depth along the line of sight to the observer , @xmath52 , depends on both the position of the flare and the column depth @xmath53 penetrated by the emitting particles of energy @xmath54 .
this depth is determined by the energy loss rates .
high - energy electrons spiraling down a magnetic field line with a pitch angle cosine @xmath55 lose and radiate most of their energy deep in the photosphere . for @xmath56 250 mev ( lorentz factor @xmath57 ) ,
the dominant energy losses are due to coulomb - ionization , whereas for @xmath58 250 mev , the radiative losses are dominated by bremsstrahlung ( over synchrotron and inverse compton ) .
the total loss rate can be approximated as with @xmath60 ( for coulomb logarithm @xmath61 ) and @xmath62 the total density . for extreme relativistic electrons @xmath63 and @xmath64 is the lorentz factor where the two losses are equal . from these , and ignoring the small deviation of @xmath65 from unity , the column depth penetrated by an electron of initial lorentz factor @xmath66 is then @xmath67 for non - relativistic electrons and protons the coulomb collision dominates and for both particles we have @xmath68 the energy dependence of the proton loss rate is similar to that of electrons .
the coulomb losses dominate at low energies but for proton energies @xmath69 gev pion production becomes significant and at @xmath70 gev ( @xmath71 ) it becomes the dominant loss mechanism , and like electron bremsstrahlung , it gives @xmath72 .
making the same approximation as above we get the same equations , ( [ lossrate ] ) and ( [ column ] ) , but now with @xmath73@xmath45 and @xmath74 ( for @xmath75 ) .
an electron of energy @xmath76 radiates photons of energy @xmath77 with @xmath78 .
similarly , assuming that protons of energy @xmath76 produce a @xmath79 with similar energy that decays into two photons of equal energies we can again set @xmath78 ( the exact value of @xmath80 will not change our conclusions drastically ) . for relativistic electrons the radiated photons will be beamed along the pitch angle of the electrons . as shown by @xcite
there will be strong center - to - limb variation of gamma - ray flux , but for flares at a few degrees behind the limb this effect can be ignored . for a flare at the center of the solar disk ( helio - longitude @xmath81 , or angle to the limb @xmath82 ) , the optical depth is @xmath83 and increases toward the limb at a rate that depends on the ambient density profile ; @xmath84 , where @xmath85 , and @xmath86 is the distance from the center of the sun at the depth of emission @xmath53 .
the photons of interest here are produced below the photosphere at column depths @xmath87 @xmath45 and densities @xmath88 @xmath89 , where both quantities increase exponentially with a scale height @xmath90 .
consequently , most of the contribution comes from within a scale height at a radius @xmath91 corresponding to the column depth @xmath92 described above .
if we define @xmath93 and @xmath94 we get @xmath95 for occulted flares @xmath96 . since @xmath97 most of the contribution to this integral comes from very small @xmath98 so we can use the approximation @xmath99 .
this gives @xmath100,\ ] ] where @xmath101 .
thus , we get @xmath102 and @xmath103 for flares at the center and limb of the sun . however , we are interested in behind - the - limb flares with @xmath104 so that @xmath105 . for angles @xmath106 the error functions @xmath107 and @xmath108
. we can use these expressions to calculate the optical depth .
let us consider hxrs emitted by nonrelativistic electrons .
for a 50 kev photon emitted by a @xmath109 and @xmath110 kev electron ( and using the thomson cross section ) we find @xmath111 at the center of the disk , @xmath112 at the limb , and a rapid increase as @xmath113 for a behind - the - limb flare . at @xmath114and @xmath115 km , the optical depth is @xmath116 . for gamma rays
the optical depth is considerably larger because they are emitted deeper in the photosphere . for a flare near the center of the solar disk the optical depth for a 100 mev photon , emitted either by a @xmath117 mev electron or @xmath118 mev proton ,
is about 0.2 and 0.1 , respectively .
but for a flare at the limb these values increase by @xmath119 . for reasonable average pitch angles @xmath120 and even if we include the effects of non - radial field lines this could give @xmath121 .
once the flare source region moves behind the limb , the optical depth increases exponentially as @xmath122 making the detection of any flare for @xmath123impossible .
this is also true for the 1 to 10 mev bgo photons even though the protons producing them do not penetrate as deeply .
considering that the flare occurred at @xmath110behind the limb , we conclude that this scenario is untenable .
the key feature of this behind - the - limb flare is the detection of @xmath0100 mev emission for @xmath130 minutes . to explain this observation
, we also consider the scenario where the photons are produced in the corona by high - energy particles injected promptly into a magnetic trap ( e.g. , * ? ? ?
* ) at @xmath124 cm ( minimum height needed for a source @xmath110@xmath9 behind the limb to be visible above the limb ) above the transition region . for particles to be trapped efficiently we need sufficiently strong field convergence to trap most particles and a low level of turbulence so that the scattering time would be much longer than the energy loss time
otherwise most of the particles will be scattered into the loss cone and radiate deep in the solar atmosphere as in scenario [ sec : belowphotosphere ] .
let us first consider protons with an energy loss rate given by eq .
( [ lossrate ] ) with @xmath125 @xmath45 and @xmath74 .
this gives an energy loss time for a 1 gev proton of @xmath126 so for the observed duration of @xmath127 s we need a density @xmath128 which is what one encounters below the occulted transition region and not at @xmath129 cm above it .
the energy loss rate for electrons in the coronal region is dominated by coulomb collisions at low energies and synchrotron losses for magnetic fields @xmath130 g or inverse compton losses at lower fields .
this rate is again described by eq .
( [ lossrate ] ) but with @xmath131 @xmath45 and @xmath132 .
as shown in @xcite , this gives rise to flat spectra at low energies and a sharp cutoff at @xmath133 with these electrons carrying most of the energy with the longest lifetime of @xmath134 .
for the production of @xmath135 mev photons we need electrons with @xmath136 so that for a lifetime of 30 minutes we need @xmath137 and @xmath138 25 g. while these values for the density ( magnetic field ) are somewhat higher ( lower ) than the ones found at 10@xmath139 cm above the transition region , they can not be fully ruled out .
also , a photon index @xmath140 requires injected electrons with spectral index -1 ( for a thin target case ) , which is much harder than those encountered at lower energies .
thus , this model requires strong convergence , low turbulence and hard spectra . in view of the lat detection of sol2014 - 09 - 01
flare with @xmath141 ( paper in preparation ) that requires a trap at a height @xmath010@xmath142 cm , this model becomes less plausible .
a third possibility is that particles accelerated by a shock associated with this flare originating behind the limb find their way to the photosphere visible to _ fermi_where they produce gamma - rays .
this requires a magnetic connection between the acceleration site and the visible photosphere , e.g. , large overlying loops .
such a connection must have been absent during the impulsive phase and for hxr - producing electrons which are most likely accelerated in smaller loops with both footpoints occulted .
otherwise we would expect a detection of hxrs from the footpoints on the visible side .
furthermore , the lat emission error circle allows for the gamma - ray emission to occur on the visible side of the disk .
this also means that the extended - phase gamma - ray producing particles were not accelerated in small loops and , like longer lasting seps , most likely were accelerated in the shock of the cme . since the magnetic lines draping the cme are most likely connected to the occulted ar , this requires cross - field diffusion that allows migration of particles from the field lines connected to the ar to those connected to the visible disk .
the presence of a strong and short scale turbulence capable of scattering the accelerated particles with a mean free path comparable to their gyro radii will facilitate this migration @xcite .
a longer trapping time of accelerated particles in the downstream region , say e.g. , by converging field lines rooted at the sun , can also help this migration .
this model requires further study . in conclusion ,
the multiwavelength observations of this behind - the - limb flare have provided some interesting theoretical puzzles which can be resolved by more detailed investigation of the scenarios discussed above .
the @xmath143 lat collaboration acknowledges support from a number of agencies and institutes for both development and the operation of the lat as well as scientific data analysis .
these include nasa and doe in the united states , cea / irfu and in2p3/cnrs in france , asi and infn in italy , mext , kek , and jaxa in japan , and the k. a. wallenberg foundation , the swedish research council and the national space board in sweden .
additional support from inaf in italy and cnes in france for science analysis during the operations phase is also gratefully acknowledged .
v.p , w.l and f.r.d.c are supported by nasa grants nnx14ag03 g , nnx13af79 g and nnx12ao70 g .
the authors thank eric grove and ron murphy for helpful suggestions . , g. , mcbreen , s. , connaughton , v. , & briggs , m. 2012 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol . 8443 , society of photo - optical instrumentation engineers ( spie ) conference series , 3
|
we report the first detection of @xmath0100 mev gamma rays associated with a behind - the - limb solar flare , which presents a unique opportunity to probe the underlying physics of high - energy flare emission and particle acceleration . on 2013
october 11 a goes m1.5 class solar flare occurred @xmath1 9.9 behind the solar limb as observed by _
stereo_-b .
_ rhessi_observed hard x - ray emission above the limb , most likely from the flare loop - top , as the footpoints were occulted .
surprisingly , the _ fermi_large area telescope ( lat ) detected @xmath0100 mev gamma - rays for @xmath130 minutes with energies up to 3 gev .
the lat emission centroid is consistent with the _ rhessi_hard x - ray source , but its uncertainty does not constrain the source to be located there .
the gamma - ray spectra can be adequately described by bremsstrahlung radiation from relativistic electrons having a relatively hard power - law spectrum with a high - energy exponential cutoff , or by the decay of pions produced by accelerated protons and ions with an isotropic pitch - angle distribution and a power - law spectrum with a number index of @xmath13.8 .
we show that high optical depths rule out the gamma rays originating from the flare site and a high - corona trap model requires very unusual conditions , so a scenario in which some of the particles accelerated by the cme shock travel to the visible side of the sun to produce the observed gamma rays may be at work .
|
lagrangian lattice field theory has been for the last three decades the most successful non - perturbative technique to compute properties of elementary particles and to solve @xmath1 .
however , some problems have resisted to a solution even by this powerful technique .
for example , it is difficult to estimate wave functions and the spectrum of excited states .
wave functions in conjunction with the energy spectrum contain more physical information than the energy spectrum alone .
although lattice @xmath1 simulations in the lagrangian formulation give good estimates of the hadron masses , one is yet far from a comprehensive understanding of hadrons .
let us take as example a new type of hardrons made of gluons , the so - called glueballs . lattice @xmath1
calculations@xcite predict the mass of the lightest glueball with quantum number @xmath2 , to be @xmath3 . experimentally
, there are at least two candidates : @xmath4 and @xmath5 .
the investigation of the glueball production and decays can certainly provide additional important information for experimental determination of a glueball .
therefore , it is important to be able to compute the glueball wave function .
on the other hand , the hamiltonian formulation seems very suitable to compute the energy spectrum and wave function of the ground state and excited states .
unfortunately , the hamiltonian formulation applied to many - body systems or elementary particle physics has been widely considered to suffer from the absence of a solid many - body solution technique .
there are a few exceptions : in 1-dimension , the density matrix renormalisation group technique works very well @xcite . in ref.@xcite
we have made a proposal how to overcome this problem , i.e. by constructing an effective hamiltonian via monte carlo - the so called monte carlo hamiltonian ( mch ) . in recent years
, monte carlo methods have been widely used to solve problems in quantum physics .
for example , with quantum monte carlo there has been improvement in nuclear shell model calculations @xcite . a proposal to solve the sign problem in monte carlo greens function method , useful for spin models has been made by sorello @xcite .
lee et al .
@xcite have suggested a method to diagonalize hamiltonians , via a random search of basis vectors having a large overlap with low - energy eigenstates . in this article
we will not review the mathematical formulation defining the monte carlo hamiltonian , but rather discuss the underlying physical idea .
also , we will examine some of its characteristic properties , namely the fact that the mch describes physics in a window of energy and temperature .
we ask - which are the parameters which determine the position and size of the window ? how can it be widened ?
also we examine a scaling property - although the mch is calculated at a certain inverse temperature @xmath6 , it holds for all @xmath7 in the respective window .
finally , we discuss avenues of possible future application of the mch - like quantum chaos in many - body systems of quantum field theory , wave functions in elementary particle physics , and scattering in particle physics .
the use of hamiltonian methods in many - body physics has a long history . in general , the hamiltonian is an operator - expressed in terms of fields and its conjugate momenta . a widely adopted strategy to compute physics from such hamiltonian proceeds via construction of a fock - space , on which such hamiltonian operates , and to construct a finite dimensional model subspace ( e.g. tamm - dancoff approximation ) . by computing matrix elements from such subspace ,
one computes a finite dimensional matrix and obtains - via diagonalisation - a spectrum of energies and wave functions .
the problem with such procedure is that a cut - off in fock space particle number and in high momentum is uncontrolled .
a much better idea is the renormalisation group idea by kadanoff and wilson .
they suggest to construct ( via blockspin transformation or intregrating out high momentum components ) a hamiltonian for the physics of critical phenomena , valid in the neighborhood of a fixed point - which needs much less degrees of freedom than the original hamiltonian .
the monte carlo hamiltonian is similar to this in the sense that it aims to decribe physics only in a finite range of parameters ( i.e. a low energy window ) . in lattice field theory ,
one constructs a hamiltonian via the transfer matrix - i.e. going to the limit of transition time @xmath8 .
this leads - for lattice gauge theory - to the well known kogut - susskind hamiltonian , which is expressed in terms of gauge fields ( link variables ) and its conjugate momenta .
in contrast , the mc hamiltonian is obtained from transition matrix elements at finite transition time . as a result ,
the mc hamiltonian is obtained and defined in terms of its matrix elements - and not in terms of operators . as pointed out above
, the crucial question is : in which subspace or model space are those matrix elements to be taken ? contrary to the tamm - dancoff hamiltonian expressed in terms of particle creation and annihilation operators in momentum space , the mc hamiltonian is formulated in a space , which is the analogue of position space in quantum mechanics - the so called bargman space .
it is interesting to note that the problem to conserve manifestly local gauge invariance in lattice gauge - apparently necessary for confinement of quarks - has been solved by wilson , using a position space formulation ( e.g. wilson loops ) and not a momentum space formulation .
it looks like as if the description of gauge theory in nature is simpler in position space .
however , the use of bargman space is by itself not good enough to solve a many - body problem .
an idea - essential for the mc hamiltonian - is to use a model subspace created in a random way ( monte carlo ) , however , guided by a physical principle , which gives a large ponderation to such states which are expected to yield `` large '' ( non - negligeable ) transition matrix elements .
this idea has been taken over from lagrangian lattice field theory , where the solution of path integrals - via monte carlo with importance sampling - allows to compute ( estimate ) physical observables by summing over field configurations @xmath9 which give a `` large '' weight @xmath10 $ ] where @xmath11 is the action .
in other words , we suggest that monte carlo with importance sampling , which allows to solve lattice field theory in the lagrangian formulation , is also the key to solve lattice field theory in the hamiltonian formulation .
this generates what we call an ( importance sampled ) stochastic basis in bargman space , which is absolutely essential for the solution of the hamiltonian many - body problem . .
comparison of results from monte carlo hamiltonian ( filled circles ) with standard lagrangian lattice calculations ( open circles ) . ]
we consider a one - dimensional chain of @xmath12 coupled harmonic oscillators , with anharmonic perturbation .
its euclidean action is given by @xmath13 in the continuum formulation it corresponds to the scalar @xmath0 model , @xmath14 introducing a space - time lattice with lattice spacing @xmath15 and @xmath16 , this action becomes @xmath17 ~ .\end{aligned}\ ] ] the actions given by eq.([action_scalar ] ) and eq.([action_osc ] ) can be identified by posing @xmath18 , @xmath19 , @xmath20 , and @xmath21 . , for average energy @xmath22 . ]
_ energy spectrum_. in ref.@xcite we have tested the method for a chain of harmonic oscillators ( klein gordon model ) where the energy spectrum is known exactly .
we found quite good agreement between the results from mc hamilton and the exact results .
the energy spectrum for the chain of anharmonic oscillators ( scalar model ) obtained from the mc hamiltonian is shown in fig.[[fig1 ] ] .
the model parameters are : @xmath23 , @xmath24 , @xmath25 ( @xmath26 , @xmath27 ) .
approximation parameters are : lattice size @xmath28 , lattice spacing @xmath29 , dimension of stochastic basis @xmath30 .
the statistical errors ( indicated by error bars ) are reasonably small and are more or less of the same size for @xmath31 to @xmath32 .
they increase noticeably for @xmath33 .
also the shape of the curve @xmath34 vs. @xmath35 changes beyond @xmath32 .
this signals the occurence of a low - energy window stretching from @xmath36 to @xmath37 . , for entropy @xmath38 . ]
_ thermodynamical observables_. a solid test of the monte carlo hamiltonian method is a comparison with results from standard lagrangian lattice calculations .
the information of the energy spectrum enters into such thermodynamical functions .
thus thermodynamical functions allow an indirect test of the energy spectrum .
we have chosen to compute the following thermodynamical observables : the partition function @xmath39 , free energy @xmath40 , average energy @xmath41 , specific heat @xmath42 , entropy @xmath11 and pressure @xmath43 .
those are defined by @xmath44 ~ , \nonumber \\
f(\beta ) & = & - \frac{1}{\beta } \log z ~ , \nonumber \\
u(\beta ) & = & { 1 \over z } { \rm tr } \left [ h \exp \left ( -\beta h \right ) \right ] = - { \partial \log z \over \partial \beta } ~ , \nonumber \\
c(\beta ) & = & { \partial u \over
\partial \tau}|_{v } = -k_b \beta^2 { \partial u \over \partial \beta}|_{v } ~ , \nonumber \\
s(\beta ) & = & \frac{1}{\tau}(u - f ) = k_{b } \beta ( u - f ) ~ , \nonumber \\
p(\beta ) & = & - { \partial f \over \partial v } ~ .\end{aligned}\ ] ] here @xmath45 denotes the boltzmann constant .
the temperature @xmath46 is related to @xmath7 via @xmath47 . , for specific heat @xmath48 . ]
fig.[[fig2 ] ] shows the free energy @xmath40 as a function of @xmath7 .
the following behavior is observed .
first , there is good overall agreement in the range @xmath49 .
however , there is an essential difference between the lagrangian approach and the mc hamiltonian approach : in the lagrangian approach , one needs to do a new computation for each value of temperature or inverse temperature @xmath7 .
moreover , the computation is easier in the regime of small @xmath7 , i.e. high temperature .
increasing @xmath7 requires larger lattices , which means that the numerical effort and errors increase . in the mc hamiltonian approach , one compute transition matrix elements for a particular fixed value @xmath50 .
one extracts a spectrum of energy eigenvalues and corresponding wave functions .
suppose we are close to the quantum continuum limit .
then the obtained energy spectrum should become independent of @xmath50 . then plugging those energies in the expressions of thermodynamical functions , eq.([eq : thermodynfunc ] ) , allows to obtain those function for any value of @xmath7 ( asymptotic scaling ) . because in practice one can never reach the continuum limit nor the infinite volume limit , as a consequence one will not be able from a single mc hamiltonian siumulation to obtain thermodynamical functions for any @xmath7 .
rather one can compute those functions only in a bounded domain in @xmath7 .
in other words there is a finite temperature window .
this is one essential difference between both approaches .
there is another difference .
while the lagrangian approach is efficient and precise in the domain of small @xmath7 or high temperature , the mc hamiltonian approach is efficient and precise in the domain of high @xmath7 or low temperature .
this can be seen in fig.[[fig3 ] ] , which displays the average energy .
with increase of @xmath7 , one observes that the lagrangian lattice results fluctuate more than those from the mc hamiltonian .
this phenomenon is more evident in fig.[[fig4 ] ] for the entropy and even more so in fig.[[fig5 ] ] for the specific heat . from the data of @xmath40 , @xmath41 , @xmath11 and @xmath42
, we estimate the temperature window of the monte carlo hamiltonian to range from @xmath51 to @xmath52 .
one may ask : on what does the location and size of the finite temperature window depend ?
it certainly depends on the parameters which control the approach to the continuum limit and the infinite volume limit , i.e. lattice spacing @xmath15 and @xmath16 , on the lattice size @xmath12 and @xmath53 . in lattice
@xmath1 it will depend also on the coupling @xmath54 and @xmath55 .
it will depend on @xmath56 , the size of the stochastic basis .
finally , it will depend also on the precision in computing the transition amplitudes ( statistical error , etc . ) . from the study of the scalar model
, we found that the size of the lattice in temperature direction as well the size of the stochastic basis are most important .
in order to make the mc hamiltonian a valuable tool in condensed matter or elementary particle physics , it is crucial to show that it allows to treat gauge theories and eventually fermions .
work on incorporating gauge theories is presently under way .
if the mc hamiltonian methods works for such theories , there is much interesting physics to be done , where conventional non - perturbative methods , in particular lagrangian lattice field theory , have brought little progress .
let us outline a few examples .
( i ) energies and wave functions of excited states . here , in our opinion , the hamiltonian formulation is most suitable .
many - body wave functions play a role for hadron structure functions in particle physics , electromagnetic form factors in nuclear physics , bose - einstein condensation in atomic physics .
( ii ) s - matrix and scattering amplitudes in high energy physics .
the non - perturbative calculation of scattering amplitudes in high energy physics is an unsolved problem yet . in ref.@xcite krger
has suggested how to compute the s - matrix in a time - dependent manner from spectral information , i.e. energies and wave functions .
the mc hamiltonian may provide such spectral information .
( iii ) quark - gluon phase transition in @xmath1 at finite temperature and finite density .
such phase transition requires to treat @xmath1 in the presence of a non - zero chemical potential . in lagrangian lattice
qcd , the treatment of the gauge group @xmath57 leads to a complex action , which escapes a treatment by monte carlo methods @xcite .
this has been a long standing problem . on the other hand , in the hamiltonian formulation on the lattice
, the presence of the chemical potential does not lead to any complex term .
( iv ) quantum chaos from level density distributions in quantum field theories .
it is a widely used strategy to characterize quantum chaos , e.g. in nuclear or atomic physics , via level density distributions ( wigner for fully chaotic ) and poissonian ( for integrabel non - chaotic ) systems @xcite . to obtain good statistics requires the knowledge of a considerable number of energy levels .
the idea is that such information may be provided by the mc hamiltonian .
we have discussed physical ideas underlying the concept of the monte carlo hamiltonian . for the case of the scalar model we have computed thermodynamical functions and by comparison with standard lagrangian lattice calculations we showed that the mc hamiltonian works .
we outlined possible future applications of the mc hamiltonian in elementary particle physics , nuclear physics and atomic physics .
h.k . and k.j.m.m .
are grateful for support by nserc canada .
is supported by the national science fund for distinguished young scholars , national science foundation of china , the ministry of education , the foundation of the zhongshan university advanced research center and the guangdong provincial natural science foundation of china ( proj .
990212 ) .
caron , h. krger , g. melkonyan , x.q .
luo , k.j.m .
moriarty , quantum monte carlo : recent advances and common problems in condensed matter and field theory , eds .
m. campostrini , m.p .
lombardo , f. pederiva , edizioni ets , pisa ( 2001 ) , p.94 .
|
in order to solve quantum field theory in a non - perturbative way , lagrangian lattice simulations have been very successful .
here we discuss a recently proposed alternative hamiltonian lattice formulation - the monte carlo hamiltonian . in order to show its working in the case of the scalar @xmath0 model
, we have computed thermodynamic functions like free energy , average energy , entropy and specific heat .
we find good agreement between the results from the monte carlo hamiltonian and standard lagrangian lattice computations .
however , the monte carlo hamiltonian results show less fluctuations under variation of temperature .
we address properties of the mc hamiltonian , like a finite temperature window , and scaling properties .
also we discuss possible future applications - like quantum chaos in many - body systems , the non - perturbative computation of wave functions of elementary particles , as well as scattering amplitudes in high energy physics . , ,
|
markarian 421 ( mrk 421 , j1104 + 3812 ) is a nearby bl lacertae object ( @xmath7 ) and was the first extragalactic tev gamma - ray source discovered @xcite .
the tev gamma - ray flux is variable , with flaring behavior observed on time - scales of less than an hour @xcite .
extensive measurements have been performed by several experimental groups based on the imaging cherenkov technique @xcite .
multi - wavelength observations support the synchrotron self
compton ( ssc ) mechanism for the production of tev gamma - rays from this source ( see , e.g. , * ? ? ?
* ; * ? ? ?
tev gamma - rays from extra - galactic sources suffer absorption due to photon - photon interactions with the inter - galactic infrared background radiation @xcite .
according to recent measurements of the infrared background ( see , e.g. , * ? ? ?
* and references therein ) and predictions of the optical depth for tev gamma - rays @xcite , gamma - rays at energies above 10 tev from mrk 421 are expected to be suppressed , since they interact with mid- to far - infrared photons of @xmath0100 @xmath8 m .
mrk 421 became active in 2000 and 2001 , especially at the beginning of 2001 @xcite . during this period
, northern hemisphere observers measured the energy spectrum with good statistics in the region from several hundred gev to @xmath010 tev and reported cutoffs at 36 tev @xcite .
the cutoff energy is consistent with , or slightly smaller than , that measured for mrk 501 during its flaring state in 1997 @xcite . as mrk 501
has a similar redshift ( @xmath9 ) to mrk 421 , this suggests the cutoffs may be due to infrared absorption of tev gamma - rays .
we observed mrk 421 during the 2001 high state with the cangaroo - ii 10 meter telescope , at very large zenith angles of @xmath070@xmath1 .
similar observations have been reported by the durham group for mrk 501 in the high state of 1997 @xcite . for these observations , an effective collecting area @xmath010 times larger than that for observations near the zenith
is obtained , with an accompanying increase in the gamma - ray energy threshold to @xmath010 tev .
the observations were made with the cangaroo - ii 10 meter telescope @xcite , located near woomera , south australia , australia ( 136@xmath147@xmath10e , 31@xmath106@xmath10s ) .
the telescope consists of 114 segmented optical mirrors , each of 80 cm diameter @xcite .
the camera contains 552 half - inch photomultiplier tubes , arranged at [email protected] intervals , and covering a field of view of @xmath03@xmath1 .
mrk 421 was observed for ten nights in early 2001 ; january 24 , 26 , 27 , 30 , 31 and february 1 , and march 14 ( all dates in ut ) , when the source was extremely active . from the cangaroo - ii telescope site ,
mrk 421 culminates at a zenith angle of [email protected] .
approximately two hours observations were made per night .
off source data were taken with the right ascension suitably offset .
an event trigger was registered when 3 individual pixels exceeded a threshold of @xmath02.5 photoelectrons . after rejecting data affected by clouds and those at zenith angles greater than [email protected]
, 14.34 hours on source data and 16.65 hours off source data remain .
a software trigger was applied in order to reduce the effect of pixels randomly triggered by the night sky background .
pixels with pulse - heights of greater than @xmath03.3 photoelectrons , pixel trigger times within 40 nanoseconds of the central value for the event , and three or more adjacent pixels were required .
finally , four or more pixels surviving these cuts were required in each event .
large zenith angle observations are well - suited to searching for gamma - ray signals at higher energies , as a much larger effective area can be achieved compared to observations near the zenith @xcite , though with a higher energy threshold . from monte carlo simulations @xcite , an effective area of @xmath11 at @xmath12 tev was estimated for observations at 70@xmath1 , with the area increasing to @xmath13 for higher energies .
a threshold energy ( where the gamma - ray detection rate is maximized ) of @xmath011 tev was derived for a @xmath14 spectrum .
this is an increase by a factor of @xmath030 in comparison with observations near zenith .
the energy threshold changes by @xmath151 tev if the spectral index is varied by @xmath20.5 .
the selection of the gamma - ray events is based on the parameterization of the elongated shape of the cherenkov light image using the standard parameters : @xmath16 , @xmath17 ( shape ) , @xmath18 ( location ) , @xmath19 ( direction ) , and @xmath20 ( orientation angle ) @xcite . instead of the conventional parameterization cuts
, we adopted the likelihood method @xcite , which has a higher efficiency of gamma - ray discrimination than the conventional parameterization technique .
the likelihood method uses a single parameter , @xmath21 $ ] , where @xmath22 and @xmath23 are the probabilities for the event having been initiated by a gamma - ray from the source or a background event , respectively .
they are the products of individual probabilities for @xmath16 , @xmath17 , and @xmath19 , which are derived from the probability density functions , including the energy dependence .
these functions were obtained using gamma - ray simulations for the signal and the observed off source events for the background .
@xmath24 ranges from 0 to 1 , and the probability of a gamma - ray origin for an event increases as @xmath24 becomes closer to 1 , though the gamma - ray acceptance does not significantly decrease in the range of @xmath25 0.5 .
we adopted a relatively loose cut of @xmath26 0.4 with an additional requirement of @xmath27 . with these cuts and a further cut excluding events with @xmath28 ,
86 % of background events are rejected while 63 % of gamma - ray events are expected to be retained . the resulting event distribution of @xmath20 is shown in figure [ fig : alpha ] ( a ) ( left panel ) .
a clear excess over the background is apparent around the source direction .
the excess is broadly distributed , up to @xmath0 30@xmath1 , due to the deterioration of the pointing resolution , caused by the shrinkage of the gamma - ray shower image .
this spread in @xmath20 distribution is consistent with simulations , as shown in the bottom panel of figure [ fig : alpha ] .
the off source distribution was normalized to that of the on source by the ratio of the number of the events in @xmath29 ( 0.88 ) , which is consistent with the ratio of observation times ( 0.86 ) , within statistical errors .
an excess of 298@xmath252 events , with a significance of 5.7 @xmath3 ( calculated using the method of @xcite ) was obtained in the region of @xmath30 . for the confirmation of the detected signal ,
the conventional parameterization cuts of [email protected]@[email protected] , [email protected]@[email protected] , and [email protected]@[email protected] were applied to the data , and a signal of 286@xmath255 events was obtained with 5.2 @xmath3 significance .
since the observations were undertaken at large zenith angles , @xmath0 70@xmath1 , we carefully examined the data and the simulations in more detail : \1 .
the shrinkage of the shower image , which is problematic for large angle observations , was studied by a comparison between simulations and data using the background events due to cosmic - ray hadrons .
figure [ fig : para ] shows the imaging parameters @xmath17 and @xmath16 , observed at large ( @xmath070@xmath1 ) and small ( @xmath015@xmath1 ) zenith angles , respectively .
the hadron simulations were made using the corsika code ( version 6.004 ) @xcite , considering the cosmic - ray abundance in the tev region @xcite .
the resultant distributions of the simulations agree with the data for both small and large angles .
it is also noted that our high resolution imaging camera , which has a pixel spacing size of 0.115@xmath1 , helped to separate the smaller images of gamma - ray events from those of background events .
the expected @xmath17 and @xmath16 distributions of the gamma - ray events , simulated with the spacing size of [email protected] and [email protected] , are shown in the bottom panels of figure [ fig : para ] . with the larger pixel size , the reconstructed image size increases and
becomes more similar to those of hadrons , with an estimated @xmath050 % decrease in the separation efficiency .
although gamma - ray detection is still possible with the larger spacing , the higher resolution imaging camera is more advantageous for large zenith angle observations .
the @xmath18 distribution of the gamma - ray selected events was compared to those from simulations .
the location of cherenkov images due to gamma - ray cascades in the field of view has a particular distribution , while those due to hadron showers are uniformly distributed . for large zenith angle observations in particular , as the observed distances of gamma - ray shower images decrease
, there is a substantial difference with the background distribution .
figure [ fig : distance ] shows the @xmath18 distribution of the gamma - ray selected events , which is obtained by subtracting the off source distribution from the on source distribution after the likelihood and @xmath20 cuts were applied .
the resulting distribution has a clear peak around [email protected] from the source direction , which agrees reasonably well with simulations and differs from that of the cosmic ray background , which provides additional confirmation of the detection of tev gamma - rays .
the `` standard candle '' at tev energies , the crab nebula , was observed at relatively large zenith angles of @xmath055@xmath1 in november and december 2000 , and the gamma - ray flux was measured for the confirmation of the analysis method and the estimation of the systematic error in the energy scale .
using the same analysis technique as that used for mrk 421 , the differential energy spectrum was derived over the energy range from 2 tev to @xmath020 tev @xcite , which agrees well with other experiments @xcite , within a @xmath015 % error in the energy scale .
these consistencies provide robust supporting evidence for the detection of @xmath4 tev gamma - rays from mrk 421 .
figure [ fig : energyflux ] ( inserted panel ) shows the raw energy spectrum of the observed gamma - ray events from mrk 421 .
the gamma - ray energy was assigned from the pulse - height sum of the individual pixels , using a relation obtained from the simulations .
this method is similar to that described in @xcite , and an energy resolution of @xmath031 % is estimated .
the excess events are distributed in the energy range 745 tev , however one must take care of the spill - over effect from the lower energies due to the finite energy resolution . in order to take this effect into account , simulated gamma - ray spectra , with the spectral indexes and cutoff energies varied , were compared to the data and the observed spectral parameters were determined from the values which minimized the value of @xmath34 . with the assumption of a power - law spectrum ,
the differential flux was fitted by with @xmath34=2.5/2 @xmath36 the cut dependence on @xmath24 and @xmath20 parameters , and the trigger conditions in the simulation , were considered as sources of the systematic uncertainties .
the systematic errors giving rise to uncertainty in the energy scale such as cherenkov photon scattering in the atmosphere are not included here , but are considered in more detail later . the derived spectrum is steeper than those observed at lower tev energies .
the spectral shape was tested with a cutoff spectrum of @xmath37 , as was derived from the measurements by the whipple and hegra - ct groups , with the spectral index being the hardest one observed during the strong flaring period @xcite .
the fitting result did not improve compared to that with the power - law assumption ( @xmath34=5.0/3 @xmath36 ) , as an excess of events above 20 tev is apparent , as shown in fig [ fig : alpha ] ( b ) .
an excess of 103@xmath226 ( 4.0 @xmath3 ) was observed with @xmath38 , while 11 events are expected for the cutoff spectrum , based on an estimation using the event ratio between 1020 tev and over 20 tev .
however , if a cutoff energy of 8 tev is assumed , the consistency with the data becomes better ( 48 events expected for @xmath39 ) .
this cutoff energy is at the high end of the range allowed for mrk 501 ( @xcite , see also @xcite ) .
since these two agns have similar redshifts , the cutoff energies in both spectra are expected to be similar , assuming the attenuation is predominantly due to infrared absorption .
as there is only a 2@xmath3 difference between our observations and this prediction , our result falls in the acceptable range of the absorption hypothesis due to the cosmic infrared background .
figure [ fig : energyflux ] ( main panel ) shows the measured energy flux , assuming the power - law spectrum .
data for the whipple @xcite and hegra - ct groups @xcite , observed during a similar period of the flaring state ( january march 2001 ) are also shown .
the observation periods were not exactly the same and the source varied significantly during this high state , therefore the absolute fluxes are expected to differ at some level .
the absolute flux level determined from the cangaroo - ii data is within the observed range of the flux variation reported by the whipple group @xcite , and the spectral slope around 10 tev is consistent with that of these two groups , supporting the roll - over from the flatter spectrum measured at lower energies . for large zenith angle observations , a large uncertainty in the energy scale , due to the absorption of cherenkov photons in the atmosphere ,
is inevitable .
only rayleigh scattering was considered in the simulation code to avoid over - estimating the gamma - ray energies .
the inclusion of mie scattering and ozone absorption would affect the energy scale by @xmath030 % and @xmath03 % , respectively , based on numerical estimations using the program code of @xcite .
we stress that these effects increase the energy scale .
the use of the `` flat - earth '' approximation for the atmosphere in the simulations requires a @xmath06 % correction which has already been taken into account in the discussion above .
the measurement of spectra at large zenith angles was verified by observations of the crab nebula up to the zenith angles of @xmath055@xmath1 , although calibration using the crab nebula at the same zenith angles as the mrk 421 observations ( @xmath070@xmath1 ) is unfortunately impractical with the current instrumental sensitivity .
the strong gamma - ray emission of mrk 421 ( @xmath0 3 times that of crab nebula ) enabled us to detect the source in only 14 hours . in order to detect the crab nebula at the same significance level , more than 150 hours observations would be required . in summary , owing to the large effective area and the high resolution performance of the cherenkov imaging camera , e@xmath4010 tev gamma - rays from mrk 421 were detected at a high confidence level at zenith angles of @xmath070@xmath1 with 14 hours of observations .
the derived spectrum in the region of 1030 tev is steeper than that around 1 tev , which supports the cutoff spectrum of mrk 421 measured in the 0.210 tev range by other groups .
the excess observed above 20 tev is strongly suggestive of a higher cutoff energy , @xmath08 tev , compared to the lower energy observations .
these observations confirm , with the support of detailed simulations , the viability of the large zenith angle technique .
large zenith angle observations provide a unique method of measuring the spectrum in the important energy range above 10 tev with a relatively short observation time .
the authors thank f. krennrich and d. horns for kindly providing flux data .
this project is supported by a grant - in - aid for scientific research of ministry of education , culture , science , sports and technology of japan and australian research council . the receipt of jsps research fellowships is also acknowledged .
we thank the dsc woomera for their assistance in constructing the telescope .
|
we have observed markarian 421 in january and march 2001 with the cangaroo - ii imaging cherenkov telescope during an extraordinarily high state at tev energies . from 14 hours observations at very large zenith angles , @xmath070@xmath1 , a signal of 298@xmath252 gamma - ray like events ( 5.7 @xmath3 ) was detected at @xmath4 tev , where a higher sensitivity is achieved than those of usual observations near the zenith , owing to a greatly increased collecting area . under the assumption of an intrinsic power - law spectrum , we derived a differential energy spectrum @xmath5 ph./@xmath6/sec / tev , which is steeper than those previously measured around 1 tev , and supports the evidence for a cutoff in the spectrum of markarian 421 .
however , the 4@xmath3 excess at energies greater than 20 tev in our data favors a cutoff energy of @xmath08 tev , at the upper end of the range previously reported from measurements at tev energies .
|
the discovery of the top quark one of the basic components of the standard model
is one of the most important goals for present and future experiments .
indirect evidence for the existence of top is very strong , see for example @xcite , and the chances that it will be detected at the fermilab @xmath3 collider in the next few years are very high .
once top is discovered , the next challenge for experiments is to measure its parameters in particular its mass @xmath4 as precisely as possible .
an unambiguous interpretation of experimental data and the determination of the top quark parameters relies on a clear quantitative understanding of the details of the production process , including the effects of gluon bremsstrahlung at the production stage @xcite . to quantify these effects
, one can study the average amount of energy lost by the top quark the simplest infrared - safe characteristic of gluon bremsstrahlung .
such a quantity is of potential practical importance in top mass determinations and , for energies above threshold , can be calculated quite straightforwardly without undue complications from non - perturbative phenomena or top - width effects @xcite .
the key point here is the large top width , @xmath5 @xcite . since we know from experiment
that the top mass is greater than about @xmath6 @xcite , the decay width is greater than the typical hadronic scale , @xmath7 , and so the top quark decays before non - perturbative effects like the formation of @xmath8-flavoured hadrons become important @xcite .
the bremsstrahlung of gluons off the top quark can have an important impact ( radiation damage " ) on various characterstics of the @xmath0 final state .
taking @xmath9 as an example , it obviously affects the kinematical constraint @xmath10 , over and above the well - known effects of initial radiation .
it also influences the form of the @xmath11 production vertex and measurements of the spectra of secondary leptons from @xmath8-decay @xcite .
since one of the primary goals of a future @xmath1 collider would be to obtain an accurate measurement of the top mass @xmath4 , it is worth considering the impact of gluon radiation on such a measurement .
there are two scenarios which have been considered @xcite
. if the top is not too heavy , it can have sizeable kinetic energy and a mass measurement can be obtained from the invariant mass per hemisphere . in practice
, the distributions in the heavy ( @xmath12 ) and light ( @xmath13 ) hemisphere masses can be fitted to a monte carlo simulation with @xmath4 as a free parameter . however
, a gluon radiated from the @xmath14 system will distort this measurement .
in particular the average difference in the hemisphere masses squared is proportional ( at @xmath15 ) to the average gluon energy , m_h^2 - m_l^2 = e_g .
the more energy is carried away by the gluon , the more asymmetric the hemisphere masses . in practice , as we shall see , @xmath16 can be @xmath17 , which is much larger than the estimated statistical errors on @xmath4 from such analyses .
a monte carlo which does not take full account of gluon radiation at the @xmath0 production stage could yield misleading results on the precision on @xmath4 .
another method which is more suited to a heavier top quark is to reconstruct the mass directly from the three - jet final state @xcite . to avoid combinatorial and other backgrounds , the energy of the three - jet system is typically constrained to be close to the beam energy .
but once again , the emission before the top decay of an energetic gluon can distort this constraint , since naively e_3 jet - e_beam = - e_g .
once again we see the importance of taking the effects of gluon radiation into account in reconstructing the top mass . in this paper we present results for gluon bremsstrahlung in @xmath18 which are of practical interest for physics studies of future linear @xmath1 colliders .
we focus in particular on the gluon energy spectrum which , as we have seen , can have important implications for the top mass measurement .
it is also straightforward , at least in principle , to extend our results to the more complicated case of hadronic @xmath0 production .
since centre - of - mass energies @xmath19 at which multiple gluon radiation should be taken into account @xcite ( @xmath20 ) appear unrealistic for the foreseeable future , we need only consider a first - order perturbative analysis .
this is calculationally quite straightforward , and in fact many parts of the calculation can be taken over from existing qed analyses @xcite ( see also @xcite ) .
the remainder of the paper is organised as follows .
we first set up the theoretical framework which allows us to calculate the gluon radiation distribution for arbitrary production currents .
we derive compact , analytical expressions for the gluon energy spectrum and its average value for vector , axial vector , scalar and pseudoscalar produciton currents .
we then present some numerical calculations which illustrate the size of the effects for typical experimental parameters .
finally , we summarize our results and present our conclusions .
in this section we describe the calculation of the cross section for single primary gluon emission in the process @xmath18 .
our calculation is valid for any collision enegy @xmath21 , and we take both photon and @xmath22 exchanges into account .
we begin by defining some variables : @xmath23 is the total centre - of - mass energy and @xmath4 is the mass of the top quark .
the momenta of the initial and final state particles are labelled by @xmath24 and we define @xmath25 .
the energy fractions shared by the @xmath8 , @xmath26 and gluon in the @xmath1 centre - of - mass frame are z_i & = & 2qp_i q^2 , i=1,2 + z & = & 2qk q^2 , z = 2 - z_1 - z_2 . [ eq1 ] the quark velocity in the @xmath0 centre - of - mass frame is = ( z ) = v = , [ eq2 ] where = m^2q^2 , _ 0 = 41-z . [ eq3 ] the final - state radiation is limited by the maximum kinematically allowed energy 0 z z_max = v^2 .
[ eq4 ] the virtualities of the top quark and antiquark before gluon emission , @xmath27 , can be written in terms of the energy fractions : _ 1 & = & 2 kp_1 = q^2 ( 1-z_2 ) , + _ 2 & = & 2 kp_2 = q^2 ( 1-z_1 ) .
[ eq5 ] it is also convenient to introduce the angular " variable = _ 1 z q^2 = 1-z_2 z .
[ eq6 ] in the calculations which follow , we neglect the electron mass @xmath28 . since we are concerned with energy losses caused by gluon emission at the @xmath0 production stage
, we may treat the @xmath8-quarks as stable objects . only very soft gluons with energy @xmath29
are affected by the top width , and these make a negligible contribution to the average energy loss .
the total non - radiative cross section for @xmath0 production in @xmath1 collisions can be expressed in terms of the vector and axial contributions ( see for example @xcite ) : _
= ^vv(s ) + ^aa(s ) = r_(s ) _
pt(s ) , [ eq7 ] where _ pt ( s ) = 4 ^ 2(s ) 3 s is the point cross section for @xmath30 with the qed coupling evaluated at the scale @xmath31 , and r_(s ) = r^vv(s ) + r^aa(s ) = v ( _ v ^v(s ) + _ a ^a(s ) ) [ eq8 ] with [ eq9a ] _ v & = & ( 3-v^2 ) = 1 + 2 , + _ a & = & v^2 = 1 - 4 .
[ eq9b ] the electroweak factors @xmath32 ( @xmath33 ) are defined as [ eq10 ] ^v & = & - 4 v_e v_t + & & + 3 ( v_e^2 + a_e^2 ) v_t^2
^2 s^2 ( s - m_z^2)^2 + ( s _ z / m_z)^2 , + ^a & = & 3 ( v_e^2 + a_e^2 ) a_t^2 ^2 s^2 ( s - m_z^2)^2 + ( s _ z / m_z)^2 , [ eq11 ] with = g_f m_z^2 4 ( s ) , + v_e = -+ 2 ^2_w , & & a_e = - , + v_t = - ^2_w , & & a_t = . the functions @xmath34 and @xmath35 are shown in fig . 1 .
except for energies close to the @xmath36 peak , which are not relevant for @xmath0 production , the axial current piece is numerically small compared to the vector current piece . at very high energies
the ratio of @xmath37 to @xmath38 tends to a constant value of approximately 0.26 .
furthermore , close to @xmath0 threshold the axial current contribution is further suppressed by two powers of @xmath39 , see eqs .
( [ eq9a],[eq9b ] ) .
we should also note that near threshold , the @xmath0 cross section is strongly modified by coulomb enhancement effects , see refs .
however the extreme threshold region is not of primary interest in the present study , since the perturbation radiation is strongly suppressed there .
we consider in this section the calculation of the differential cross section for the emission of a primary gluon of momentum @xmath40 in @xmath0 production . after integrating over the relative orientation of the quark and gluon momenta with respect to the incoming leptons , the interference between the vector and axial pieces vanishes and the double differential cross section in the quark and antiquark energy fractions ( the dalitz plot distribution )
can be written as = 1 _ = \ { d^2 dz_1 dz_2 } _ v + \ { d^2 dz_1 dz_2 } _ a , [ eq12 ] with \
{ d^2 dz_1 dz_2 } _ c = 1^cc .
[ eq13 ] the subscript @xmath41 denotes the production current , see eqs .
( [ eq7],[eq8 ] ) .
the derivation of the expressions for the radiative matrix element squared , summed over colours and spins and integrated over the azimuthal orientation of the produced particles with respect to the beam direction , drastically simplifies if one uses an analogue of the invariant - integration technique " which was employed in refs .
@xcite for calculations of the qed radiation accompanying two - charged - particle production in @xmath1 annihilation .
the matrix element @xmath42 describing the radiation off the top quark produced via the @xmath43-current exchange can be written as m_c^(1 ) = g_s b_c t^a \ { |u(p_1 )
a^c_v(p_2 ) } j_^e , c , [ eq14 ] with a_^c = _ ^c ( je _ ) + _ _ ^c _ 1 + _ ^c _ _ 2 .
[ eq15 ] here @xmath44 is a su(3 ) colour matrix and @xmath45 is a polarization vector for the gluon .
the matrices @xmath46 ( @xmath47 ) are _
^a = _ _ 5 .
[ eq16 ] the four - vector @xmath48 is the classical current induced by the acceleration of the top - quark colour charges , j_= 2 p_1 _ 1 - 2 p_2 _ 2 , [ eq17 ] and @xmath49 describes the @xmath43-exchange contribution at the electron vertex , j^e , c_= are chosen so that the non - radiative matrix element @xmath50 is m_c^(0 ) = b_c \ { |u(p_1 ) ^c_v(p_2 ) } j_^e , c .
[ eq19 ] following refs .
@xcite , it is convenient to integrate out the angular variables which describe the orientation of the final state quarks and gluons with respect to the incoming leptons .
we first introduce the lorentz - invariant polarization tensor @xmath51 _ ^c = 116 _ , [ eq20 ] with @xmath52 , which describes the radiative decay of a polarized spin - one heavy
object " @xmath43 . since we wish to integrate over the directions of the quark and gluon momenta , the resulting tensor structure can depend only on the four - vector @xmath53 which defines the reference frame in which the quark energies are fixed .
it is convenient to write the polarization tensor @xmath51 in the general form _
^c = ( g _ - n_n _ ) _ c + n_n_^q_c , [ eq21 ] where _ c = _ - ^q_c , ^q_c = n_^c _ n _ .
[ eq22 ] note that because of vector current conservation , the tensor @xmath54 is transverse , i.e. ^q_v = 0 .
[ eq23 ] for the axial current , the second ( longitudinal ) term in eq .
( [ eq21 ] ) vanishes in the limit @xmath55 .
in what follows we will set @xmath56 , and so only the transverse component @xmath57 appears in the final expression for the double - differential distribution , eq .
( [ eq13 ] ) . as a result
, we obtain \
{ d^2 dz_1 dz_2 } _ c = _ c , [ eq24 ] where @xmath58 with @xmath59 the non - radiative polarization operator that one obtains from eqs . ( [ eq21],[eq22 ] ) by substituting the born vertices for @xmath60 in eq .
( [ eq20 ] ) , @xmath61 \ ; .\ ] ] with the numerical normalization chosen for eq .
( [ eq20 ] ) one has 4 _ c^(0 ) = _ c , _
c = 1_c , [ eq25 ] where the born factors @xmath62 are given in eqs .
( [ eq9a],[eq9b ] ) .
explicit calculations give ) and pseudoscalar ( @xmath63 ) exchanges . ]
[ eq26a ] _ v & = & j^2 _ v + ( + ) , + [ eq26b ] _ a & = & j^2 _ a + , with j^2 -g^ j_j_= .
[ eq27 ] in terms of the dimensionless energy fractions defined in eq .
( [ eq1 ] ) , we have j^2 & = & z_1 + z_2 - 1 - 2(1-z_1)(1-z_2 ) - ( 1-z_1)^2 - ( 1-z_2)^2 , + + & = & z^2 ( 1-z_1)(1-z_2 ) - 2 .
[ eq28 ] finally , the dalitz plot distribution is given by substituting the expressions given in eqs .
( [ eq8],[eq9a],[eq9b],[eq24 ] -[eq28 ] ) into eqs .
( [ eq12],[eq13 ] ) . before presenting numerical results
, we make some general comments concerning the global structure of the above results for the @xmath57 . * the first ( classical " ) term in eqs .
( [ eq26a],[eq26b ] ) corresponds to long - distance radiation with polarization vector @xmath64 in the plane defined by @xmath65 .
it leads to a universal contribution to the radiative cross section .
in contrast , the short - distance effects are represented by the second term , which induces a current - dependent , non - universal contribution .
this describes equal production of the gluon states @xmath64 and @xmath66 ( polarization transverse to the @xmath65 plane ) .
* next , consider the case of soft gluon emission , when the radiative cross section can be expanded in powers of the gluon energy fraction @xmath67 . in accordance with the factorization properties of soft radiation @xcite , the first term in eqs .
( [ eq26a],[eq26b ] ) is proportional to the product of the born term @xmath62 and the usual accompanying radiation factor @xmath68 , with the former evaluated at the total centre - of - mass energy @xmath69 of the process .
it is interesting that at small @xmath70 the second ( non - universal ) term in eqs .
( [ eq26a],[eq26b ] ) appears to be @xmath71 compared to the first term , not @xmath72 as one might naively expect . the only correction linear in @xmath70 is incorporated into the classical radiation factor @xmath68 .
the physical origin of this can be understood @xcite by recalling the celebrated low soft bremsstrahlung theorem @xcite , extended to the case of charged fermions by burnett and kroll @xcite .
the application of these classical results is especially transparent for the case of the integrated annihilation cross sections , where the non - radiative process depends only on one energy variable ( the centre - of - mass energy of the charged particles ) .
annihilation , the universal part of the integrated photon - emission cross section is given by the product of the born term taken at the point @xmath73 and the radiation factor describing the emission off the initial electrons @xcite .
this construction remains valid for the more general case of arbitrarily polarized , massive initial - state particles ( see also @xcite ) .
] * in the extreme ultra - relativistic limit , @xmath74 , the @xmath75 and @xmath76 distributions become identical . * the results ( [ eq24]-[eq26b ] ) are not valid in the extreme threshold limit , even within the context of perturbation theory . in this region , the first - order radiation amplitude is strongly modified by the qcd coulomb - like interaction between the @xmath8 and @xmath26 @xcite .
this leads to a distortion of the radiation pattern and to the additional suppression of radiation @xcite . to calculate the inclusive gluon energy distribution it is convenient to rewrite the quantities @xmath77 in terms of the variables @xmath70 and @xmath78 ( see eqs .
( [ eq6],[eq25]-[eq26b ] ) ) .
this gives _ v & = & _ soft + z^2 + ( 12 ) , [ eq29a ] + _ a & = & _ soft + z^2 + ( 12 ) . [ eq29b ] here we have introduced @xmath79 as the classical current contribution which accounts for the @xmath80 and @xmath81 terms , _ soft = 1-z-2 - ^2 , [ eq30 ] and ( 1-(z ) ) ( 1+(z ) ) .
[ eq31 ] explicit integration then gives z \
{ d dz } _ c = z^cc = ( z)v c_f_s , [ eq32 ] with the born factors @xmath62 given in eqs .
( [ eq9a],[eq9b ] ) .
the universal soft `` part of the gluon density reads ( z ) = ( 1+v^2 -2 z)(z ) - 2(1-z ) , [ eq33 ] where ( z ) = .
the hard '' contributions which depend on the the production channel are _
h(z;v ) & = & ( z ) - 1 , + _
h(z;a ) & = & ( z)- 1 .
[ eq34 ] note that the result for @xmath82 coincides with that obtained in ref .
@xcite .
we turn now to the quantity which is the main concern of the present study the average energy lost by the @xmath8-quark by gluon emission at the production stage .
the average gluon energy fraction is _
c = _ 0^v^2 dz z \ { d d z } _ c [ eq35 ] and the mean energy fraction lost by the @xmath8-quark ( antiquark ) is then = = 1 - _ c . it is convenient to write @xmath83 in the form
_ c = c_f \ { + _ c^-1 _ c } , [ eq37 ] where the universal @xmath43-independent piece is = 2v _ 0^v^2 dz ( z ) [ ( 1-z -2)(z ) -(1-z ) ] .
[ eq38 ] the hard " term @xmath84 in eq .
( [ eq37 ] ) is given by _
v & = & i _ - i_1 , + _ a & = & i _ - i_1 , [ eq39 ] with i _ & = & 1v _ 0^v^2dz z^2 ( z ) ( z ) = 1v _ 0^v^2 dz z^2 , + i_1 & = & 1v _
0^v^2dz z^2 ( z ) .
[ eq40 ] performing the @xmath70 integration in eqs .
( [ eq38],[eq40 ] ) , we obtain & = & 18\ { ( 3 + 2v^2 + 3v^4)_v -6(1+v^2 ) } , +
i _ & = & 116\ { ( 53 + v^2 + v^4 + 5v^63 ) _ v -2(53 + 14v^29 + 5v^43 ) } , + i_1 & = & 116\ { ( -5 + 3v^2 + v^4 + v^6)_v + 2(5 - 4v^23 -v^4 ) } , with _
[ eq42 ] the average gluon energy fractions for vector and axial production currents are obtained by combining the results of eqs .
( [ eq35]-[eq42 ] ) .
figure 2 shows the resulting @xmath85 and @xmath86 as a function of the top quark velocity @xmath39 , for @xmath87 .
note that in the relativistic limit , @xmath88 , the average gluon energies diverge logarithmically , _
v _ a c_f , [ vto1 ] reflecting the emergence of the collinear singularity in the massless quark limit . of course when @xmath89 multiple gluon emission becomes important , and the leading logarithms must be resummed to all orders ( see ref .
the behaviour of the average energy in the threshold region , @xmath90 , is discussed in the following section .
figure 3 shows that average gluon energy @xmath91 as a function of the top mass @xmath4 in @xmath1 annihilation at centre - of - mass energy @xmath19 .
the correct proportions of vector and axial current contributions ( fig . 1 ) are included , and the strong coupling is taken to be @xmath87 . for top masses in the experimentally favoured range @xmath92 gev , @xmath16 varies from 14 gev to 4 gev . for @xmath93 , we can take the small-@xmath39 limit of the results of the previous section to obtain [ eq43a ] & & v^4 , + [ eq43b ] i _ & & , + [ eq43c ] i_1 & & .
the high powers of the threshold factor @xmath39 originate as follows : in the case of the classical " universal term @xmath94 , one power of @xmath95 comes from the size of the integration region , @xmath96 ( eq .
( [ eq4 ] ) ) , and another power comes from the dipole suppression of the accompanying radiation , see also refs.@xcite . because of the explicit @xmath97 factor appearing in the hard radiation " spectrum , eq .
( [ eq32 ] ) , the non - universal contribution @xmath98 , and thus acquires an additional @xmath95 suppression relative to the universal @xmath94 piece .
note , however , that the impact of the hard radiation " in the threshold region strongly depends on the production channel . in particular , the born factor @xmath62 in the denominator in eq .
( [ eq37 ] ) can compensate the suppression of the non - classical effects in @xmath99 , see eq .
( [ eq9a],[eq9b ] ) .
this is an interesting example of how the universal nature of soft radiation can sometimes be obscured by process - dependent short - distance effects . in the vector channel , where the ( @xmath100-wave ) born amplitude is not suppressed at threshold
, the soft piece dominates and we find ( see also @xcite ) _ v ~c_f v^4 .
[ eq44 ] however , in the case of the axial - vector ( @xmath101-wave ) production channel , the non - radiative amplitude for the @xmath102 decay is additionally suppressed at threshold . as discussed above ,
non - classical " radiation effects at @xmath15 are then equally important , and induce a significant contribution to @xmath86 . from eqs .
( [ eq37],[eq39],[eq43a]-[eq43c ] ) we find _ a ~c_f v^4 ( 1 + ) , [ eq45 ] with the second term in the @xmath103 brackets coming from the hard radiation " .
the limiting @xmath90 behaviours corresponding to the leading @xmath104 terms in @xmath85 and @xmath86 are shown as dashed lines in fig . 2 .
as we have already explained , the practical consequences of the differences between the average gluon energy in the vector and axial - vector channels for @xmath18 are not large , since the axial contribution is so small in the threshold region , see eqs .
( [ eq8],[eq9a],[eq9b],[eq12 ] ) .
it is , however , worth mentioning that the non - universality of the soft radiation in the threshold region is not restricted to the axial channel .
the same situation arises in the case of @xmath0 production in the scalar ( e.g. higgs exchange ) channel , @xmath105 . in the appendix we present the corresponding results for @xmath106 .
we find that in the limit @xmath90 we have @xmath107 , so that once again the soft radiation contribution is accompanied by a hard radiation term of the same order .
another example which could be of practical interest concerns gluon radiation accompanying two scalar quark ( e.g. stop ) production : .
this is a @xmath101-wave process , dominated by photon exchange . near threshold ,
the average energy loss is again given by eq .
( [ eq45 ] ) , illustrating the non - universality of the soft radiation description for this process also .
all the results derived above have assumed that the top is a _
stable _ particle .
however in practice and particularly near threshold , one must be careful to take account of the effects of the top decay width and of non - perturbative fragmentation .
although the former can affect the distribution of soft gluons ( i.e. gluons with energy @xmath108 ) @xcite , it is not important for the average top energy loss .
we can quantify the possible impact of non - perturbative dynamics by using the string - model results of ref .
@xcite for the average amount of energy lost by the top quark before it decays . near the @xmath0 threshold
this leads to _ non pt = 2 _ t m v , [ eq47 ] with the so - called string tension factor @xmath109 .
since @xmath110 @xcite , these non - perturbative effects decrease as @xmath111 .
taking as a specific example the canonical values @xmath112 and @xmath113 , we obtain _ non pt 3.310 ^ -3 v .
[ eq48 ] comparing eqs .
( [ eq44 ] ) and ( [ eq48 ] ) , one may conclude that the perturbative prediction is only valid for @xmath114 .
it is important that gluon emission in top quark production processes is under control .
otherwise , it may provide a serious source of uncertainty in the determination of the top quark parameters from experimental data .
in this paper we have studied gluon bremsstrahlung off the top quarks in @xmath18 .
we have calculated the exact first - order perturbative expressions for the energy fraction distributions of the @xmath8-quarks .
these are valid in a wide energy range above the @xmath0 threshold . to examine the impact of gluon emission on top quark distributions we have calculated the average energy loss at the production stage .
this infra - red safe quantity characterizes , for example , the difference between the beam energy and the actual energy shared by the produced @xmath8-quarks .
* acknowledgements * we are grateful to victor fadin , david miller , lynne orr and torbjrn sjstrand for useful discussions .
this work was supported in part by the united kingdom science and engineering research council .
in this appendix we present the expressions corresponding to single primary gluon emission in the process @xmath115 , mediated by the exchange of scalar @xmath100 or pseudoscalar @xmath101 particles ( for example , the higgs bosons of the standard model or its supersymmetric extensions ) .
now the invariant polarization operator bears no lorentz indices , and we define ( cf . eq .
( [ eq20 ] ) ) _ c = 18 _ , [ alaeq20 ] where the normalization has been fixed to preserve the relation ( [ eq25 ] ) . here
the scalar functions @xmath60 are analogous to @xmath116 of eq .
( [ eq15 ] ) but with the vertex operators ^s = 1 , ^p = _ 5 , [ alaeq16 ] substituted for @xmath46 . defining the double - differential distribution analogously to eqs .
( [ eq24 ] ) , ( [ eq25 ] ) with _ s = _ a = v^2 , _ p = 1 , we find _ s & = & j^2 v^2 + ( _ 2_1 + _ 1_2 + 2 ) , + _ p & = & j^2 + ( _ 2_1 + _ 1_2 + 2 ) . the gluon energy spectrum is then given by eq .
( [ eq32 ] ) with _
h(z , s ) = _
h(z;p ) = ( z ) , and the mean gluon energy fraction by eq .
( [ eq37 ] ) with _ s = _
p = i _ . near the threshold for the pseudoscalar channel , the soft term dominates and _ p _ v .
in contrast , the transition @xmath117 is @xmath101-wave , and the born term @xmath118 compensates the additional suppression of @xmath119 .
the threshold result for @xmath120 therefore coincides with that for @xmath86 , eqs .
( [ eq43a]-[eq43c ] ) . 99 p. langacker , university of pennsylvania preprint urp-0555-t ( 1993 ) . yu.l .
dokshitzer , v.a .
khoze and s.i .
troyan , university of lund preprint lu - tp-92 - 10 ( 1992 ) .
khoze , l.h . orr and w.j .
stirling , nucl .
* b378 * ( 1992 ) 413 . + v.a .
khoze , j. ohnemus and w.j .
stirling , phys .
* d49 * ( 1994 ) 1237 .
dokshitzer , v.a .
khoze , and s.i .
troyan , in m. derrick , ed .
6th int . conf . on physics in collision
_ , world scientific , 1987 , p. 417 .
bigi , yu.l .
dokshitzer , v.a .
khn and p.m. zerwas , phys . lett .
* 181b * ( 1986 ) 157 .
cdf collaboration : f. abe _ et al .
* d45 * ( 1992 ) 3921 .
+ d0 collaboration : s. abachi _ et al .
_ , fermilab preprint fermilab - pub-94 - 004-e ( 1994 ) .
khn , acta phys .
* 24 * ( 1982 ) 203 .
lipatov and v.a .
khoze , in materials of the x winter school of the leningrad nuclear physics institute , vol .
ii , leningrad ( 1975 ) , p. 409 .
baier _ et al .
. rep . * 78 * ( 1981 ) 294 .
fadin and e.a .
kuraev , sov .
. phys . * 41 * ( 1985 ) 466 .
workshop on @xmath121 collisions at 500 gev : the physics potential , munich - annecy - hamburg 1991 , ed .
p.m. zerwas , desy 92 - 123a , b .
. workshop on physics and experiments with @xmath121 linear colliders , saariselk 1991 , eds .
r. orava , p. eerola and m. nordberg , world scientific , singapore 1992 .
igo - kemenes _
et al . _ ,
@xcite , p. 327 .
dokshitzer , light and heavy quark jets in perturbative qcd , in physics up to 200 , proc .
school of subnuclear physics , erice 1990 , plenum press , new york , 1991 , vol .
baier and v.a .
khoze , preprint nsu , novosibirsk ( 1964 ) , sov .
jetp * 21 * ( 1965 ) 629 .
baier and v.a .
khoze , sov .
jetp * 21 * ( 1965 ) 1145 .
ioffe , phys .
lett . * 78b * ( 1978 ) 277 .
dokshitzer , v.a .
khoze , l.h . orr and w.j .
stirling , nucl .
b403 * ( 1993 ) 65 .
j. jerzak , e. laermann and p.m. zerwas , phys .
* d25 * ( 1982 ) 363 . for a recent review
see : j.h .
khn and p.m. zerwas , advanced series on directions in high energy physics , heavy flavours " , eds .
buras and m. lindner , world scientific , singapore 1992 , p. 434 .
fadin and v.a .
khoze , jetp lett .
* 46 * ( 1987 ) 525 ; sov .
* 48 * ( 1988 ) 309 .
fadin and v.a .
khoze , proc .
24th lnpi winter school , leningrad 1989 , vol .
3 . m.j .
strassler and m.e .
peskin , phys .
* d43 * ( 1991 ) 1500 .
see for example : j. m. jauch and f. rohrlich , _ the theory of photons and electrons : the relativistic quantum field theory of charged particles with spin one - half _ , ( 1976 ) ; a.i .
akhiezer and v.b .
berestetskii , _ quantum electrodynamics _ , interscience ( 1965 ) .
dokshitzer and v.a .
khoze , in preparation .
low , phys .
* 110 * ( 1958 ) 974 .
burnett and n.m .
kroll , phys .
lett . * 20 * ( 1968 ) 86 .
baier and v.a .
khoze , sov .
. phys . * 5 * ( 1967 ) 898 .
nagorskaya and v.a .
khoze , izvestia an armenian ssr , fizika * 3 * ( 1968 ) 371 .
fadin and o.i .
yakovlev , sov .
* 50 * ( 1989 ) 1037 .
t. sjstrand and p.m. zerwas , proc .
workshop on @xmath121 collisions at 500 gev : the physics potential , munich - annecy - hamburg 1991 , ed .
p.m. zerwas , desy 92 - 123a , p. 463 .
* the functions @xmath34 and @xmath35 defined in eqs .
( [ eq10],[eq11 ] ) . * the average gluon energy in @xmath122 as a function of the top quark velocity @xmath39 , for vector and axial vector current production . also shown ( dashed lines )
are the limiting @xmath90 behaviours given in eqs .
( [ eq44],[eq45 ] ) , and the limiting @xmath88 behaviour given in eq .
( [ vto1 ] ) . * the average gluon energy in @xmath11 production in @xmath1 annihilation at @xmath123 , as a function of the top quark mass @xmath4 .
the strong coupling is fxed at @xmath87 .
|
the emission of energetic gluons in @xmath0 production in @xmath1 annihilation can have important experimental consequences , in particular on top quark mass measurements .
we present compact , analytical expressions for the gluon energy distribution and its average value at first order in qcd perturbation theory .
our results are valid for arbitrary masses , collision energies and production currents .
we pay particular attention to top quark production near threshold , and show that in certain cases the soft gluon approximation is insufficient to describe the radiation spectrum .
# 1#1 12_12 dtp/94/14 + lu - tp943 + march 1994 + 1.cm * gluon radiation and energy losses + in top quark production * 1.cm yu.l .
dokshitzer .2 cm _ department of theoretical physics , university of lund + slvegatan 14a , s-22362 lund , sweden _
+ .4 cm v.a .
khoze@xmath2 .2 cm _ department of physics , university of durham + durham dh1 3le , england _
+ .4 cm and .4 cm w.j . stirling .2 cm _ departments of physics and mathematical sciences , university of durham + durham dh1 3le , england _
|
recent precise measurement of the cosmic microwave background ( cmb ) anisotropies by the wilkinson microwave anisotropy probe ( wmap ) @xcite has provided detailed informations about the mechanism of generating the cosmic density fluctuations .
importantly , observed cmb anisotropies are highly consistent with those predicted from scale - invariant purely adiabatic primordial density fluctuations .
such a class of primordial density fluctuations is also consistent with the recent results by the sloan digital sky survey experiments @xcite . among various possibilities ,
inflation is one of the most famous scenarios of generating cosmic density fluctuations consistent with the observations .
another class of scenario of generating the scale - invariant adiabatic density fluctuations , however , exists , which is called the `` curvaton '' scenario @xcite.a similar study of the effect of extra scalar field other than the inflaton , see also @xcite .
( for the recent works on the curvaton scenario , see @xcite . )
if there exists a scalar field @xmath0 ( other than the inflaton ) which acquires primordial fluctuation during the inflation , dominant part of the cosmological density fluctuations may originate from the primordial fluctuation of this scalar field .
( the scalar field @xmath0 is called the curvaton . ) importantly , this scenario works irrespective of the detailed properties of the curvaton field , i.e. , its mass , lifetime , and initial amplitude .
consequently , there are many possible and well - motivated candidates of the curvaton field since , in various scenarios of particle cosmology , many scalar - field condensations which once dominate the universe show up ; if those scalars acquire primordial quantum fluctuations , they may play the role of the curvaton .
thus , the curvaton scenario naturally fits into various scenarios of particle cosmology .
in addition , with the curvaton scenario , we have a chance to test the properties of the particle responsible for the cosmological density fluctuations by collider experiments ; this is the case if the flat direction of a minimal supersymmetric standard model ( mssm ) becomes the curvaton , for example .
( remember that it is difficult to construct a model of inflation within the framework of the standard model or the mssm.in the mssm , however , it may be possible to use the @xmath1-flat direction consisting of the up - type squarks as well as the up - type higgs boson as the inflaton .
for details , see @xcite . )
an important aspect of the curvaton scenario is that the universe is reheated by the decay of the curvaton . in some case , reheating temperature at the curvaton decay becomes relatively low and hence it is non - trivial if a viable scenario of baryogenesis can be found .
in addition , a large amount of entropy is produced at the time of the decay of the curvaton . as a result ,
even if non - vanishing baryon asymmetry is generated before the decay of the curvaton , primordial baryon asymmetry may be too much diluted to be consistent with the present value .
this may be a problem for the case where , for example , the cosmological moduli fields play the role of the curvaton .
one of the possibilities to generate enough baryon asymmetry with large entropy production is to adopt the affleck - dine baryogenesis @xcite .
indeed , in @xcite , it was pointed out that the resultant baryon - to - photon ratio @xmath2 ( with @xmath3 and @xmath4 being the number densities of the baryon and photon , respectively ) can be as large as the presently observed value @xmath5 even with the large late - time entropy production . in the affleck - dine scenario , however , baryonic isocurvature fluctuation may arise since , in this scenario , baryon asymmetry is generated from a coherent motion of a scalar field ( called affleck - dine field ) . in particular ,
if the affleck - dine mechanism is implemented in the curvaton scenario , correlated baryonic isocurvature fluctuation may arise as we will see below .
thus , it is important to study constraints on this case and see if the curvaton scenario with affleck - dine baryogenesis is viable . in this paper
, we consider the curvaton scenario with the affleck - dine baryogenesis .
in particular we study the possible baryonic entropy fluctuation generated in this framework and how the entropy fluctuation can be suppressed .
the organization of the rest of this paper is as follows . in section [ sec : evolutions ] , we first discuss framework and give the thermal history we consider .
then , we introduce equations by which evolutions of the cosmological density fluctuations are governed . analytic solutions to those equations for a simple ( but well - motivated ) case are also discussed . in section [ sec : numerical ] , we numerically solve the evolution equations . in particular , we will see that the baryonic entropy fluctuation provides very stringent constraints on the curvaton scenarios with affleck - dine baryogenesis .
implications of such constraints for some of the curvaton scenarios are discussed in [ sec : implications ] .
section [ sec : summary ] is devoted to the summary of this paper .
let us first introduce the framework . in various models of physics beyond the standard model ,
there are many scalar fields whose potential is approximately parabolic : @xmath6 in the mssm , for example , @xmath7- and @xmath1-flat directions , whose potential is lifted by the supersymmetry - breaking effects , has parabolic potential .
in addition , potential of the cosmological moduli fields are expected to be parabolic when the amplitude of the modulus field is smaller than @xmath8 ( where @xmath9 is the reduced planck scale ) .
another possible candidates of such a scalar field is pseudo - nambu - goldstone bosons @xcite . in this paper , we assume one of such scalar fields plays the role of the curvaton . here ,
we adopt inflation as a solution to the horizon , flatness , and other cosmological problems .
then , the universe starts with the inflationary epoch where the universe is dominated by the potential energy of the inflaton field @xmath10 ; thermal history before the inflation is irrelevant for our discussion . during the inflation , we assume that the amplitude of the curvaton is non - vanishing .
in addition , we consider the case where the effective mass of the curvaton during the inflation is much smaller than the expansion rate during the inflation @xmath11 .
then , the curvaton acquires the primordial fluctuation during the inflation @xmath12 ( here and hereafter , the subscript `` init '' is used for initial values of the quantities at very early epoch when @xmath0 is slowly rolling . ) after the inflation , universe is reheated by the decay of the inflaton field and the universe is dominated by radiation .
we call this epoch the first radiation - dominated epoch ( or rd1 epoch ) since , in our scenario , there are two radiation - dominated epochs . as far as the decay of @xmath0 is negligible , universe consists of two components : radiation generated from the decay products of the inflaton , which we denote @xmath13 , and the curvaton .
after the rd1 epoch , evolution of the energy density of @xmath13 obeys the following equation : @xmath14 where the `` dot '' denotes the derivative with respect to the time @xmath15 .
in addition , ( unperturbed part of ) @xmath0 obeys the following equation : @xmath16 ( as is obvious from the above equations , we have not taken into account of the decay of the curvaton field since the effect of the decay of @xmath0 is negligible for the rd1 and @xmath0-dominated epochs . ) here , @xmath17 ( with @xmath18 being the scale factor ) is the expansion rate of the universe which is given by , when @xmath19 is much larger than the decay rate of @xmath0 , @xmath20 from eq .
( [ dot(rho_r ) ] ) , we can see that @xmath21 . on the contrary ,
the curvaton starts to oscillate when @xmath22 ; after this epoch , energy density of the curvaton decreases as @xmath23 .
thus , the universe can be dominated by the curvaton if the lifetime of the curvaton is long enough .
hereafter , the @xmath0-dominated epoch is called as @xmath0d epoch . after the @xmath0d epoch , the curvaton decays and the universe is reheated again ; we call the second radiation - dominated epoch as rd2 epoch .
after the rd2 epoch , the thermal history is the same as that of the conventional big - bang model ( as far as the zero - mode is concerned ) . the curvaton amplitude @xmath24 at the time of the `` radiation - curvaton equality '' is easily estimated .
( in the following , the subscript `` eq '' is used for quantities at the time of the radiation - curvaton equality . )
the curvaton starts to oscillate when the temperature of the radiation generated by the inflaton decay ( which is denoted as @xmath13 hereafter ) becomes @xmath25 and , after this epoch , the curvaton amplitude is approximately proportional to @xmath26 .
thus , @xmath27 expansion rate at the equality is also easily estimated : @xmath28 assuming that the energy density of the affleck - dine field is always sub - dominant in the rd1 and @xmath0d epochs , which we assume in this paper , the above discussion holds even with the affleck - dine field .
equation of motion of the affleck - dine field is , on the other hand , given byin our convention , the affleck - dine field @xmath29 is a complex scalar field while the curvaton @xmath0 is a real scalar field .
we assume that the effect of the motion of the curvaton in the phase direction is negligible .
( for the effect of the motion of the curvaton in the phase direction , see @xcite . )
@xmath30 where @xmath31 is the potential of the affleck - dine field .
evolution of the affleck - dine field depends on the detailed shape of its potential . in this paper , to be specific , we consider the case where an @xmath7- and @xmath1-flat direction in the mssm plays the role of the affleck - dine field . in this case ,
baryon - number violating interaction in the potential of @xmath29 is assumed to originate from khler interaction .
another possibility is with baryon- ( or lepton- ) number violating superpotential .
then , however , the initial amplitude of the affleck - dine field is usually suppressed .
consequently the resultant baryon asymmetry becomes too small to be consistent with the presently observed value if a large amount of entropy is produced by the decay of the curvaton .
one of the motivation here to consider the affleck - dine baryogenesis is to generate large enough baryon - number asymmetry . in the case of the @xmath7- and @xmath1-flat direction as the affleck - dine field
, it is known that baryon asymmetry can be large enough even with the large entropy production @xcite .
thus , we consider the affleck - dine potential of the form : @xmath32 where @xmath33 is from the supersymmetry - breaking effect and @xmath34 is a constant of @xmath35
. then , potential of the affleck - dine field is dominated by the parabolic term as far as @xmath36 and @xmath29 starts to oscillate when @xmath37 .
after this epoch , @xmath38 is approximately proportional to @xmath26 . denoting the baryon - number of @xmath29 as @xmath39 , relevant part of the baryon - number current
is given by @xmath40 then , we obtain @xmath41 thus , if the affleck - dine field has non - vanishing initial amplitude @xmath42 , baryon - number density is generated when @xmath29 starts to move .
the affleck - dine field starts to move when @xmath37 . parameterizing the initial value of the affleck - dine field as @xmath43
, the baryon - number density at that moment is estimated as @xmath44_{h\sim m_\psi } \sim \left [ b_\psi { \rm i m } \left ( \frac{\partial v_{\rm ad}}{\partial\psi } \psi \right ) h^{-1 } \right]_{h\sim m_\psi } \sim \lambda b_\psi \frac{m_\psi}{m_*^{p-2 } } |\psi_{\rm init}|^p \sin p\theta_{\rm init } , \label{n_b(h = m)}\end{aligned}\ ] ] where coefficients of @xmath35 is neglected .
first , let us consider the case where the present cmb radiation is generated from the decay product of the curvaton . in this case , the resultant baryon - to - photon ratio depends on the epoch when the slow - roll condition of the affleck - dine field breaks down .
( here , we consider the case with @xmath45 ; as we will see below , this relation is required in order to suppress the baryonic entropy fluctuation . ) to discuss the resultant baryon - to - photon ratio , it is convenient to define @xmath46 if @xmath47 , @xmath48 is realized in the rd1 epoch while , for the case of @xmath49 , the affleck - dine field starts to oscillate in the @xmath0d epoch .
if the affleck - dine field starts to move in the rd1 epoch , background temperature at @xmath48 is estimated as @xmath50 . at this moment , with the condition @xmath45 , amplitude of @xmath0 is given by @xmath51 .
then , the ratio @xmath52 , which is a conserved quantity when @xmath53 , is estimated as @xmath54 when the curvaton decays , energy density of @xmath0 is converted to that of radiation and hence the baryon - to - photon ratio is estimated as @xmath55 where @xmath56 is the reheating temperature at the time of the curvaton decay .
thus , in this case , the resultant baryon - to - photon ratio depends on the initial amplitude of the curvaton field .
if @xmath48 is realized in the @xmath0d epoch , on the contrary , the situation changes and the expansion rate at the time when @xmath29 starts to move is determined by the energy density of the curvaton .
consequently , at the time of @xmath48 , amplitude of the curvaton is estimated as @xmath57_{h\sim m_\psi } \sim m _ * \left ( \frac{m_\psi}{m_\phi } \right).\end{aligned}\ ] ] then , the baryon - to - photon ratio is estimated as @xmath58 in this case , the resultant baryon - to - photon ratio does not depend on @xmath59 . if the lifetime of the affleck - dine field is long enough , the present cmb radiation dominantly originate from the decay product of the affleck - dine field . in this case , the expression of the baryon - to - photon ratio changes using the fact that the ratio @xmath60 ( with @xmath61 being the energy density of the affleck - dine field ) is a constant after the affleck - dine field starts to move . the baryon - to - photon ratio is estimated as @xmath62_{h\sim m_\psi } t_{\rm addecay } \sim \frac{\lambda b_\psi |\psi_{\rm init}|^{p-2 } \sin p\theta_{\rm init } } { m_\psi m_*^{p-2 } } t_{\rm addecay } , \label{eta_3}\end{aligned}\ ] ] where @xmath63 is the reheating temperature due to the decay of the affleck - dine field .
then , the baryon - to - photon ratio is independent of @xmath59 irrespective of the time when the affleck - dine field starts to move .
now , we consider the evolution of the density fluctuations . in our scenario , there are two independent sources of the cosmological density fluctuations ; one is the primordial fluctuation of the curvaton given in eq .
( [ dphi_init ] ) and the other is that of the inflaton .
since we are interested in the case where the curvaton contribution dominates , let us consider the evolution of the cosmological density fluctuations generated from the primordial fluctuation of the curvaton .
effects of the primordial fluctuation of the inflaton will be discussed in section [ sec : implications ] .
although we are interested in the case where the curvaton and the affleck - dine field both exist , we first consider the adiabatic mode ( in the @xmath0d epoch and after ) neglecting the effects of the affleck - dine field .
indeed , behaviors of the adiabatic mode can be understood without taking account of the affleck - dine field as far as the energy density of the affleck - dine field is always sub - dominant . in this paper
, we use the newtonian gauge where the line element is described aswe use the convention and notation of @xcite unless otherwise mentioned .
@xmath64 here , @xmath65 is some constant and @xmath66 is the comoving coordinate . in the rd1 and @xmath0d epochs , it is expected that the temperature of the universe is so high that the momentum - exchange of the relativistic particles are efficient enough . in this case
, anisotropic stress vanishes and @xmath67 . with this relation , relevant part of the equations to be solved
are , for the fourier mode with the ( comoving ) momentum @xmath68 , @xmath69 \delta\phi = 4 \dot{\phi } \dot{\psi } - 2 m_\phi^2 \phi \psi , \label{dphidot}\end{aligned}\ ] ] where @xmath70 with @xmath71 being the fluctuation of the energy density of @xmath13 , @xmath72 with @xmath73 being the velocity of @xmath13 , and @xmath74 is the fluctuation of the curvaton amplitude . in order to solve the equations
, we have to specify the initial conditions for the fluctuations ( as well as those for @xmath75 and @xmath0 ) . in order to study the effects of the primordial fluctuation of the curvaton , we concentrate on the mode for which , in the deep rd1 epoch , @xmath0 has non - vanishing fluctuation as given in eq .
( [ dphi_init ] ) while @xmath71 vanishes .
importantly , we are interested in the modes related to currently observed cosmological density fluctuations .
those modes reenter the horizon at the time close to the ( usual ) radiation - matter equality and hence they are superhorizon modes in the rd1 and @xmath0d epochs .
thus , in discussing their behaviors in the rd1 and @xmath0d epochs , all the terms proportional to @xmath76 are irrelevant in the evolution equations . then , solving eqs .
( [ psidot ] ) @xmath77 ( [ dphidot ] ) , fluctuations in the deep rd1 epoch when @xmath78 are given by @xmath79 where the subscript `` rd1 '' is for variables in deep rd1 epoch , and the superscript `` @xmath80 '' is for fluctuations generated from the primordial fluctuation of the curvaton . here , @xmath81 is the expansion rate at @xmath82 ( which is , of course , taken at deep rd1 epoch ) , and @xmath83 notice that the relations given in eqs .
( [ psi_rd1 ] ) @xmath77 ( [ dphi_rd1 ] ) provide initial conditions for our numerical calculations .
evolutions of the fluctuations can be studied by solving eqs .
( [ psidot ] ) @xmath77 ( [ dphidot ] ) .
importantly , in the curvaton scenario , fluctuation in the curvaton sector is converted to the adiabatic density fluctuations after the @xmath0d epoch .
evolutions of the superhorizon fluctuations are the same as those with the baryonic isocurvature fluctuations @xcite ; after the @xmath0d epoch , @xmath84 becomes of the order of @xmath85 , where @xmath86 if the initial amplitude of the curvaton is much smaller than @xmath9 , @xmath0 starts to oscillate when the universe is dominated by radiation . in this case
, @xmath85 becomes the entropy fluctuation between components generated from the decay products of @xmath0 and those generated from @xmath10 ( like @xmath13 ) . since the curvaton can be identified as a non - relativistic component once it starts to oscillate ,
situation is the same as the models with isocurvature fluctuation in non - relativistic component .
then we obtain @xmath87 where the subscripts `` @xmath0d '' and `` rd2 '' are quantities in the @xmath0d and rd2 epochs , respectively .
now , we discuss the entropy fluctuation in the baryonic sector , which is defined as @xmath88 although the precise calculation of the entropy fluctuation will be done in the next section with numerical calculations , it is instructive to discuss the basic behavior of the entropy fluctuation . indeed , in some case , baryonic entropy fluctuation associated with the affleck - dine field can be analytically estimated . here
, we consider the case where the energy density of the affleck - dine field is sub - dominant in the rd1 and @xmath0d epochs .
in such a case , eqs .
( [ eta_1 ] ) and ( [ eta_2 ] ) are applicable , from which we can estimate the entropy fluctuation generated in our scenario . as can be seen in eq .
( [ eta_1 ] ) , if @xmath89 , @xmath2 depends on @xmath59 .
thus , if the curvaton amplitude has primordial fluctuation , it also generates the entropy fluctuation between the baryon and the photon . using eq .
( [ s_bg ] ) , @xmath90 on the contrary , when @xmath91 , @xmath2 is independent of @xmath59 , and we obtain @xmath92 importantly , shape of the cmb angular power spectrum depends on the relative size between the adiabatic and entropy fluctuations .
thus , for our following discussions , it is convenient to define @xmath93 if the curvaton starts to oscillate in the deep rd1 epoch , which is realized when @xmath94 , metric perturbation is also calculated as given in eq .
( [ psi(rd2 ) ] ) and obtain @xmath95 as mentioned before , the cmb angular power spectrum obtained by the wmap is highly consistent with the prediction of the purely adiabatic primordial density fluctuations .
thus , if @xmath96 is too large , resultant cmb angular power spectrum becomes inconsistent with the observations .
indeed , size of the correlated isocurvature fluctuation is constrained to be @xcite @xmath97 with this constraint , it is obvious that the case with @xmath89 is excluded .
consequently , if the affleck - dine mechanism is implemented in the curvaton scenario , one of the following two conditions should be satisfied : ( i ) @xmath98 , or ( ii ) @xmath99 . in order to be more quantitative , in the next section ,
we use numerical method to calculate @xmath100 ( and other quantities ) and derive the constraints . notice that , in the extreme case where the affleck - dine field eventually dominates the universe after the decay of the curvaton , all the components in the universe are generated from the affleck - dine field . in this case , entropy fluctuation vanishes and the cosmic density fluctuations are purely adiabatic .
( see eq .
( [ eta_3 ] ) . )
then , @xmath101 and the resultant cosmic density fluctuations become consistent with the observations . thus ,
hereafter , we concentrate on the cases where the present cmb radiation ( as well as other components in the universe ) are generated from the decay product of @xmath0.in our discussion , for simplicity , it is assumed that the energy density of the affleck - dine field is always sub - dominant .
our results are , however applicable to the case where the energy density of the affleck - dine field becomes comparable to that of the curvaton at some epoch as far as the present cmb radiation ( as well as other components in the universe ) are generated from the decay product of @xmath0 . before closing this section , we comment on the effects of the primordial fluctuation of the affleck - dine field .
if the effective mass of the affleck - dine field is much smaller than the expansion rate during the inflation , it is expected that the affleck - dine field also acquires the primordial fluctuation @xmath102 . since the resultant baryon - to - photon ratio depends on the initial amplitude of @xmath29 , such a primordial fluctuation results in extra entropy fluctuation in the baryonic sector . assuming no correlation between the primordial fluctuations of the curvaton and the affleck - dine field
, @xmath102 generates uncorrelated baryonic entropy fluctuation . using eq .
( [ eta_1 ] ) with @xmath103 , which is required in order to suppress @xmath100 when @xmath104 as will be seen in the next section , we obtain @xmath105 where @xmath106 is the baryonic entropy fluctuation associated with the primordial fluctuation of the affleck - dine field . defining @xmath107
we obtain @xmath108 .
uncorrelated entropy fluctuation is constrained by the wmap results and @xmath109 @xcite .
this constraint can be evaded if the initial amplitude of the affleck - dine field is not much smaller than that of the curvaton .
of course , if @xmath29 acquires effective mass as large as the expansion rate during the inflation , @xmath102 vanishes and there is no ( uncorrelated ) baryonic entropy fluctuation .
hereafter , we consider the case where the effects of the uncorrelated baryonic entropy fluctuation becomes not important , and concentrate on the effects of correlated baryonic entropy fluctuation .
so far , we have seen that the case with @xmath110 is inconsistent with the observations .
one possibility of realizing @xmath111 when @xmath112 is to adopt @xmath113 . in this case ,
( a short period of ) inflation may occur due to the energy density of the curvaton field and eq .
( [ psi(rd2 ) ] ) becomes unapplicable .
in addition , if @xmath114 , analytic estimation of the precise value of @xmath100 is difficult .
thus , in this section , we use numerical method to calculate @xmath115 and @xmath116 associated with the primordial fluctuation of the curvaton .
the first step is to calculate the metric perturbation generated from @xmath117 .
we use the initial conditions given in eqs .
( [ psi_rd1 ] ) @xmath77 ( [ dphi_rd1 ] ) in the deep rd1 epoch , and numerically solve eqs .
( [ psidot ] ) @xmath77 ( [ dphidot ] ) as well as the equations for the zero - mode from the rd1 epoch to @xmath0d epoch .
we checked that @xmath115 becomes constant in the @xmath0d epoch .
metric perturbation in the rd2 epoch is evaluated with the relation @xmath118 .
= 0.5 in fig .
[ fig : psird ] , we plot @xmath119 normalized by @xmath120 as a function of @xmath59 .
we have checked that the ratio @xmath121 is independent of @xmath122 . as we have discussed in the previous section ,
the ratio @xmath123 becomes @xmath124 when @xmath94 .
( see eq .
( [ psi(rd2 ) ] ) . ) in this case , the curvaton becomes ( almost ) non - relativistic matter in the deep rd1 epoch .
once @xmath59 becomes close to @xmath9 , on the contrary , @xmath22 is realized when the energy density of @xmath0 becomes comparable to that of radiation .
then , @xmath119 deviates from the result given in eq .
( [ psi(rd2 ) ] ) and , as seen in fig .
[ fig : psird ] , @xmath125 becomes larger than @xmath126 .
now , we consider the baryonic entropy fluctuation and discuss constraints on the curvaton scenario with the affleck - dine baryogenesis .
for this purpose , we numerically solve eqs.([psidot ] ) @xmath77 ( [ dphidot ] ) simultaneously with the evolution equation of the affleck - dine field @xmath29 . in our analysis ,
entropy fluctuation is evaluated by taking the derivative numerically ; denoting the resultant baryon and curvaton number densities with the initial amplitude @xmath59 as @xmath127 and @xmath128 , respectively , the entropy fluctuation between the baryon and the radiation is calculated as @xmath129 = 0.5 with the entropy fluctuation in the baryonic sector , we can calculate the @xmath100 parameter .
the result is shown in fig .
[ fig : kappab ] on the @xmath59 vs. @xmath122 plane .
the metric perturbation @xmath130 is determined by @xmath59 ( with @xmath85 being fixed ) , and has relatively mild dependence on @xmath59 .
thus , fig .
[ fig : kappab ] primarily shows the behavior of baryonic entropy fluctuation in this model . with this in mind
, the behavior of @xmath100 can be easily understood .
the important feature is that @xmath131 as @xmath59 or @xmath122 becomes large enough .
this is because , as explained in the previous section , baryonic entropy fluctuation is suppressed if the affleck - dine field starts to oscillate after the @xmath0d epoch is realized . in other words , if the expansion rate at the time of the radiation - curvaton equality is larger than @xmath132 , resultant entropy fluctuations become adiabatic . since @xmath133 , @xmath122 or @xmath134
is required to be large enough for suppressing baryonic entropy fluctuations . as a result
, @xmath100 becomes close to zero when the combination @xmath122 or @xmath59 becomes large . with the constraint on the @xmath100 parameter from the wmap ( [ k<0.5 ] )
, we obtain stringent constraint on the initial amplitude and the mass of the curvaton field .
in particular , for a fixed value of the initial amplitude @xmath59 , we obtain an lower bound on @xmath122 .
implications of such a constraint will be discussed in the next section .
in the previous sections , we have studied general constraints without specifying the detailed properties of the curvaron field .
consequently , we have seen that the expansion rate at the time of the radiation - curvaton equality should be larger than the mass of the affleck - dine field ; otherwise , the affleck - dine field starts to move in the rd1 epoch and too large baryonic entropy fluctuaion is generated .
this fact has serious implications to some of the cases .
in particular , it is important to consider the case where the mass of the curvaton ( as well as that of the affleck - dine field ) is from the effect of the supersymmetry breaking . in this case ,
mass of the curvaton is expected to be close to that of the affleck - dine field and we obtain serious constraint on the initial amplitude of the curvaton .
one of the important cases is that the cosmological modulus field plays the role of the curvaton .
the moduli fields in the superstring theory acquire masses from the effects of the supersymmetry breaking and hence their masses are expected to be of the order of the gravitino mass . in addition , their interactions are expected to be suppressed by inverse powers of the gravitational scale @xmath9 and hence , once their amplitudes become smaller than @xmath9 , their potentials are well approximated by the parabolic ones .
thus , if one of such moduli fields has non - vanishing amplitude as well as the primordial fluctuaion , it may play the role of the curvaton . if the mass of the modulus field ( which is close to the gravitino mass ) is of @xmath135 as in the case of the supergravity - induced supersymmetry breaking , its lifetime becomes much longer than @xmath136 .
with such a long lifetime , decay of the curvaton occurs after the big - bang nucleosynthesis ( bbn ) and it spoils the success of the bbn @xcite . if the modulus mass becomes larger than @xmath137 , on the contrary , the cosmological modulus field may decay before the bbn and it may be a viable candidate of the curvaton . in the scenario of the anomaly - mediated supersymmetry
breaking @xcite , for example , this may be the case .
in addition , in the anomaly - mediated models , wino may become the lightest superparticle ; importantly , non - thermally produced winos from the decay of the cosmological modulus field ( i.e. , the curvaton ) can become the cold dark matter @xcite .
in such a model , baryon asymmetry of the universe should be generated with very low reheating temperature and with very large dilution factor ; as a candidate of the scenario of baryogenesis in such a situation , the affleck - dine mechanism has been known to be promising @xcite .
although a sizable hierarchy is possible between @xmath122 and @xmath132 in the anomaly - mediated scenarios , we still obtain remarkable constraint on @xmath59 . even for the case with relatively large hierarchy between @xmath122 and @xmath132 as @xmath138 ( @xmath139 ) ,
for example , @xmath59 should be larger than @xmath140 ( @xmath141 ) ; otherwise , too large baryonic entropy fluctuation is generated as seen in fig .
[ fig : kappab ] . for the case of the anomaly - mediated supersymmetry breaking ,
the mass of the affleck - dine field , which is given by the masses of the mssm particles , is suppressed by the loop - induced factor of order @xmath142 compared to the gravitino mass . even in that case , @xmath143 is required in order to suppress the baryonic entropy fluctuations .
in fact , if the curvaton field has an initial amplitude as large as @xmath8 , it becomes non - trivial if the inflaton contribution to the cosmic density fluctuations is negligible .
importance of the inflaton contribution can be seen by comparing the metric perturbations generated from the fluctuations of the inflaton and the curvaton .
the curvaton contribution is given in eq .
( [ psi(rd2 ) ] ) while the inflaton contribution is given by @xmath144_{k = ah},\end{aligned}\ ] ] where @xmath145 is the inflaton potential , @xmath146 , and the superscript `` ( inf ) '' is for variables generated from the primordial fluctuation of the inflaton field . if @xmath147 , the curvaton contribution dominates .
the relative size depends on the model of the inflation . for example , for the case of the chaotic inflation with inflaton potential of the form @xmath148
, we obtain @xmath149_{k = ah_{\rm inf}},\end{aligned}\ ] ] where we have used eq .
( [ psi(rd2 ) ] ) as an approximation .
( in fact , as shown in fig .
[ fig : psird ] , @xmath150 becomes larger than the value given in eq .
( [ psi(rd2 ) ] ) if the initial amplitude of the curvaton becomes comparable to @xmath9 .
then , the ratio becomes smaller . ) using the fact that the inflaton amplitude is @xmath151 when the cobe scale exits the horizon , @xmath152 becomes comparable to or larger than @xmath130 for @xmath153 even if @xmath103 .
thus , sizable fraction of the cosmological density fluctuation originates from the inflaton fluctuation ; with the chaotic inflation , the inflaton contribution is non - negligible unless there is a large hierarchy between the masses of the curvaton and the affleck - dine field . in other words ,
if we consider the case where @xmath122 is close to @xmath132 , @xmath103 is required and the cosmic density fluctuations may not be dominated by the curvaton contribution . assuming no correlation between the inflaton and curvaton fields , the cmb angular power spectrum has the form @xmath154 where @xmath155 and @xmath156 are the cmb angular power spectra for the cases where the primordial fluctuations of the curvaton and inflaton dominates , respectively .
as we mentioned , @xmath156 may become comparable to ( or even larger than ) @xmath155 .
since the density fluctuations related to the primordial fluctuation of the inflaton field are adiabatic , constraints on @xmath59 ( for fixed value of @xmath157 ) can be relaxed as far as the scale dependence of the primordial density fluctuation is negligible . on the contrary , however ,
if @xmath152 becomes comparable or larger than @xmath130 , significant amount of the cosmic density fluctuations are generated from the primordial fluctuation of the inflaton . in this case , of course , it becomes difficult to relax the observational constraints on the inflaton potential with the curvaton , which has been one of the important motivation of the curvaton scenario .
of course , the chaotic model is not the only possibility of the inflation and , in other case , inflaton contribution to the total cosmic density fluctuations may become minor even though @xmath143 .
model of the inflation with large @xmath11 and small @xmath158 is considered , for example , in @xcite .
in this paper , we have discussed the curvaton scenario with affleck - dine baryogenesis .
even with a large entropy production due to the decay of the curvaton field , affleck - dine mechanism may be able to generate large enough baryon asymmetry and hence , for some of the curvaton scenarios , affleck - dine mechanism is a prominent candidate of the scenario of baryogenesis .
we have seen , however , that baryonic entropy fluctuation is induced if the affleck - dine field starts to move in the rd1 epoch . since large baryonic entropy fluctuation is inconsistent with the observation ( in particular , with the results of the wmap )
, this provides constraints on the scenario ; as we have seen , in order to evade the constraint , mass of the curvaton or the initial amplitude of the curvaton should be large enough ( for fixed value of @xmath132 ) .
we have also discussed implications of such constraints on some scenario of the curvaton .
one of the cases where the affleck - dine scenario is preferred is that the cosmological modulus field plays the role of the curvaton ; in such a case , baryon asymmetry should be generated with large amount of entropy production and with very low reheating temperature . assuming the anomaly - mediated supersymmetry breaking , the ratio @xmath157 is expected to be @xmath159 . with such a ratio
, we have seen that the initial amplitude of the curvaton is constrained to be larger than @xmath8 in order not to generate too large baryonic entropy fluctuation . _
acknowledgments : _ we acknowledge the use of cmbfast @xcite package for our numerical calculations .
the work of t.m . is supported by the grant - in - aid for scientific research from the ministry of education , science , sports , and culture of japan , no.15540247 .
d. n. spergel _ et al .
_ , astrophys .
j. suppl .
* 148 * ( 2003 ) 175 ; a. kogut _ et al . _ ,
astrophys .
j. suppl .
* 148 * ( 2003 ) 161 ; g. hinshaw _ et al . _ ,
astrophys .
j. suppl .
* 148 * ( 2003 ) 135 ; l. verde _ et al . _ ,
astrophys .
j. suppl .
* 148 * ( 2003 ) 195 ; h. v. peiris _ et al . _ ,
astrophys .
j. suppl .
* 148 * ( 2003 ) 213 .
m. tegmark _ et al . _
[ sdss collaboration ] , arxiv : astro - ph/0310723 .
k. enqvist and m. s. sloth , nucl .
b * 626 * ( 2002 ) 395 .
d. h. lyth and d. wands , phys .
b * 524 * ( 2002 ) 5 .
t. moroi and t. takahashi , phys .
b * 522 * ( 2001 ) 215 [ erratum - ibid .
b * 539 * ( 2002 ) 303 ] .
a. d. linde and v. mukhanov , phys .
d * 56 * ( 1997 ) 535 .
n. bartolo and a. r. liddle , phys .
d * 65 * ( 2002 ) 121301 ; t. moroi and t. takahashi , phys .
d * 66 * ( 2002 ) 063501 ; d. h. lyth , c. ungarelli and d. wands , phys .
d * 67 * ( 2003 ) 023503 ; m. s. sloth , nucl .
b * 656 * ( 2003 ) 239 ; r. hofmann , arxiv : hep - ph/0208267 ; k. dimopoulos and d. h. lyth , arxiv : hep - ph/0209180 ; t. moroi and h. murayama , phys .
b * 553 * ( 2003 ) 126 ; k. enqvist , s. kasuya and a. mazumdar , phys .
lett . * 90 * ( 2003 ) 091302 ; f. di marco , f. finelli and r. brandenberger , phys . rev .
d * 67 * ( 2003 ) 063512 ; k. a. malik , d. wands and c. ungarelli , phys .
d * 67 * ( 2003 ) 063516 ; m. postma , phys .
d * 67 * ( 2003 ) 063518 ; c. gordon and a. lewis , arxiv : astro - ph/0212248 ; k. dimopoulos , phys .
d * 68 * ( 2003 ) 123506 ; a. r. liddle and l. a. urena - lopez , phys .
d * 68 * ( 2003 ) 043517 ; j. mcdonald , phys .
d * 68 * ( 2003 ) 043505 ; k. dimopoulos , g. lazarides , d. lyth and r. ruiz de austri , jhep * 0305 * ( 2003 ) 057 ; k. enqvist , a. jokinen , s. kasuya and a. mazumdar , phys .
d * 68 * ( 2003 ) 103507 ; k. dimopoulos , d. h. lyth , a. notari and a. riotto , jhep * 0307 * ( 2003 ) 053 m. endo , m. kawasaki and t. moroi , phys .
b * 569 * ( 2003 ) 73 ; m. postma and a. mazumdar , arxiv : hep - ph/0304246 ; m. postma , arxiv : astro - ph/0305101 ; s. kasuya , m. kawasaki and f. takahashi , phys .
b * 578 * ( 2004 ) 259 ; d. h. lyth and d. wands , phys .
d * 68 * ( 2003 ) 103515 ; d. h. lyth and d. wands , phys .
d * 68 * ( 2003 ) 103516 ; k. dimopoulos , g. lazarides , d. lyth and r. r. de austri , phys .
d * 68 * ( 2003 ) 123515 ; k. hamaguchi , m. kawasaki , t. moroi and f. takahashi , arxiv : hep - ph/0308174 ; d. h. lyth , phys . lett .
b * 579 * ( 2004 ) 239 ; n. bartolo , s. matarrese and a. riotto , phys .
d * 69 * ( 2004 ) 043503 ; j. mcdonald , jcap * 0312 * ( 2003 ) 005 ; j. mcdonald , arxiv : hep - ph/0310126 ; c. gordon and k. a. malik , arxiv : astro - ph/0311102 ; m. bastero - gil , v. di clemente and s. f. king , arxiv : hep - ph/0311237 ; s. gupta , k. a. malik and d. wands , arxiv : astro - ph/0311562 ; l. pilo , a. riotto and a. zaffaroni , arxiv : astro - ph/0401302 ; e. j. chun , k. dimopoulos and d. lyth , arxiv : hep - ph/0402059 ; m. postma , arxiv : astro - ph/0403213 ; d. langlois and f. vernizzi , arxiv : astro - ph/0403258 ; k. enqvist , arxiv : hep - ph/0403273 ; j. mcdonald , arxiv : hep - ph/0404154 . s. kasuya , t. moroi and f. takahashi , arxiv : hep - ph/0312094 . i. affleck and m. dine , nucl .
b * 249 * ( 1985 ) 361 .
t. moroi , m. yamaguchi and t. yanagida , phys .
b * 342 * ( 1995 ) 105 .
k. dimopoulos , g. lazarides , d. lyth and r. ruiz de austri , in ref .
@xcite ; k. dimopoulos , d. h. lyth , a. notari and a. riotto , in ref .
@xcite .
w. t. hu ,
ph . d thesis ( arxiv : astro - ph/9508126 ) .
s. mollerach , phys .
d * 42 * ( 1990 ) 313 .
m. kawasaki and t. moroi , prog . theor .
* 93 * ( 1995 ) 879 ; m. kawasaki and t. moroi , astrophys . j. * 452 * ( 1995 ) 506 ; m. kawasaki , k. kohri and t. moroi , phys .
d * 63 * ( 2001 ) 103502 ; m. kawasaki , k. kohri and t. moroi , arxiv : astro - ph/0402490 .
l. randall and r. sundrum , nucl .
b * 557 * ( 1999 ) 79 ; g. f. giudice , m. a. luty , h. murayama and r. rattazzi , jhep * 9812 * ( 1998 ) 027 ; j. a. bagger , t. moroi and e. poppitz , jhep * 0004 * ( 2000 ) 009 . t. moroi and l. randall , nucl .
b * 570 * ( 2000 ) 455 .
l. pilo , a. riotto and a. zaffaroni , arxiv : hep - th/0401004 .
u. seljak and m. zaldarriaga , astrophys . j. * 469 * ( 1996 ) 437 .
|
we discuss the curvaton scenario with the affleck - dine baryogenesis . in this scenario ,
non - vanishing baryonic entropy fluctuation may be generated even without primordial fluctuation of the affleck - dine field .
too large entropy fluctuation is inconsistent with the observations and hence constraints on the curvaton scenario with the affleck - dine baryogenesis are obtained .
we calculate the baryonic entropy fluctuation ( as well as other cosmological density fluctuations ) in this case and derive constraints .
implications to some of the models of the curvaton are also discussed .
tu-717 + hep - ph/0404253 + april , 2004 + .5 in * * curvaton scenario with affleck - dine baryogenesis .45 in maki ikegami and takeo moroi .45 in _ department of physics , tohoku university , sendai 980 - 8578 , japan _ .4 in
|
bose - einstein condensates of dilute vapors of atoms offer a very promising testing ground for questions associated with superfluidity for a number of reasons .
firstly , these gases are dilute as opposed , for example , to the `` traditional '' superfluid liquid helium . furthermore , the atomic gases can be manipulated in many different ways , including the shape of the confining potential , the strength and the sign of the effective interatomic interaction , the number of different species in multicomponent systems , etc . while the term `` superfluidity '' covers a whole collection of many different phenomena @xcite , we focus in the present study on the metastability of superflow in annular traps @xcite .
persistent flow has been observed recently in a bose - einstein condensate of sodium atoms confined in a toroidal trap @xcite . in this experiment , an initial angular momentum of @xmath0 per particle
was transferred to the atoms and the rotational flow was observed to persist for up to ten seconds , limited only by the trap lifetime and other experimental imperfections .
persistent currents with two units of angular momentum were also observed in the same experiment .
theoretical studies have examined the existence of metastable states in bose - einstein condensed gases trapped in single- @xcite and double - ring - like @xcite confining potentials , as well as other phenomena in double - ring - like traps @xcite .
as shown in ref.@xcite , mixtures of two components that are confined in a strictly one - dimensional single ring support persistent currents , but the interaction strength necessary for metastability increases with the admixture of the second component . for comparable populations of the two components
, the critical value of the coupling becomes infinite .
finally , persistent currents with a value of the circulation higher than one unit were shown to be stable only in single - component systems . in the present study
we investigate the effect of the deviations from purely one - dimensional motion on the metastability of the currents .
we also examine two - component bose - einstein condensates , and find that it is possible to have metastable superflow for sufficiently small admixtures of the second component .
one of the novel results of our study is that in this latter case the deviation from one - dimensional motion gives rise to persistent currents with circulation higher than one unit , as opposed to the purely one - dimensional case .
the paper is organized as follows .
we describe in sec .
ii our model , and in sec .
iii the two methods that we have employed to solve it , i.e. , the mean - field approximation and numerical diagonalization of the hamiltonian . in sec .
iv we present our results .
more specifically , in sec .
iv.a we consider the single - component case and study the existence of metastable states in the transition from quasi - one - dimensional to two - dimensional motion . in sec .
iv.b we examine the effect of the admixture of a second component on the stability of the persistent currents , comparing the obtained results with the ones of strictly one - dimensional motion .
finally , we present our summary and conclusions in sec . v.
let us consider two distinguishable species of bosonic atoms , labelled as @xmath1 and @xmath2 , with populations @xmath3 and @xmath4 respectively ; without loss of generality we assume below that @xmath5 .
we also assume two - dimensional motion , which in an experiment corresponds to the case of very tight confinement in the perpendicular direction . in the plane of motion we model the annular trap by a displaced harmonic potential , @xmath6 which is plotted in fig . [ fig1 ] .
here @xmath7 is the usual radial variable in cylindrical polar coordinates , @xmath8 is the radius of the annulus , @xmath9 is the trap frequency , and @xmath10 is the atom mass , which we assume for simplicity to be equal for both species .
we stress that the results presented below with the potential of eq.([potential ] ) are at least qualitatively similar to the ones that we have obtained using a harmonic - plus - gaussian trapping potential . of eq.([potential ] ) as a function of the cartesian coordinates ( @xmath11 ) , for @xmath12 and @xmath13.,width=340 ] in order to investigate the effects due to the transition from quasi - one - dimensional to two - dimensional motion , we vary the confinement strength @xmath9 in the radial direction . for large @xmath9 ,
the confinement is strong enough to freeze out the motion in this direction , which makes the system quasi - one - dimensional .
as @xmath9 decreases , the confinement becomes weaker and , as a result , the motion becomes two - dimensional . in all the calculations presented below we fix the radius of the annulus to @xmath14 , where @xmath15 is the ususal oscillator length corresponding to some fixed frequency @xmath16 .
the interatomic interactions are modelled via the usual effective contact potential , which is assumed to be repulsive . therefore , the hamiltonian @xmath17 of the system is @xmath18 , where @xmath19 is the single - particle part @xmath20 and @xmath21 is the interaction part , given by @xmath22 here the parameters @xmath23 are proportional to the s - wave scattering lengths for zero - energy elastic atom - atom collisions . for simplicity ,
we take @xmath24 , which is also experimentally relevant in several cases ; furthermore , since the interaction is assumed repulsive , @xmath25 .
we attack this problem using both the mean - field gross - pitaevskii approximation , as well as diagonalization of the many - body hamiltonian . within the diagonalization approach ,
we fix the particle numbers @xmath3 and @xmath4 and the total angular momentum @xmath26 , and use the lanczos method @xcite to obtain the eigenenergies and the corresponding eigenvectors .
this can be done for a range of the values of @xmath27 , yielding the dispersion relation @xmath28 , i.e. , the smallest eigenvalue for some given @xmath27 , allowing us to identify the possible local minima associated with the metastable states that give rise to persistent currents in the system .
the basis states that we choose in this approach are the eigenfunctions of the single - particle problem , which are of the form @xmath29 the quantum number @xmath30 corresponds to the radial excitations , @xmath31 is the quantum number associated with the angular momentum , which results from the rotational symmetry of our problem , and @xmath32 is the usual angular variable in cylindrical polar coordinates . the radial wavefunctions @xmath33 are evaluated numerically . if @xmath34 are the corresponding eigenvalues , we work under the assumption that the typical value of the interaction energy @xmath35 is weak enough , so that @xmath36 .
thus , we set @xmath37 and the only quantum number that remains is @xmath31 , whose highest value is chosen so that @xmath38 . having evaluated the single - particle states , these are then combined in all possible ways to form a basis of the fock states for the many - body problem for some fixed @xmath39 , @xmath40 , and @xmath27 . turning to the mean - field approximation , within this approach
the system is described via the two order parameters of the two components , @xmath41 and @xmath42 . from eqs .
( [ hsp ] ) and ( [ hint ] ) one obtains the two coupled gross - pitaevskii - like equations for these two order parameters @xmath43 in the above equations the two order parameters are normalized as @xmath44 , and @xmath45 .
the parameters @xmath46 are defined as @xmath47 , @xmath48 , and @xmath49 .
finally , @xmath50 and @xmath51 are the chemical potentials of the two components . to solve eqs .
( [ gpe ] ) we make use of a fourth - order split - step fourier method within an imaginary - time propagation approach @xcite . starting with an initial state with some given angular momentum , the imaginary - time propagation drives the system along the dispersion relation @xmath28 until it finds a local minimum .
if @xmath28 has such local minima , the final state may be a metastable one .
otherwise , the initial state decays to the non - rotating ground state of the system .
one should note that the dissipation of the angular momentum @xmath27 is inherent to the diffusive character of this method , and it does not contradict the fact that the hamiltonian commutes with the operator of the angular momentum , since the imaginary - time propagation operator is not unitary , and therefore does not satisfy heisenberg s equation of motion @xcite .
the two methods mentioned above complement each other : the diagonalization gives the full dispersion relation @xmath28 for any fixed angular momentum @xmath26 , but due to numerical limitations in diagonalizing matrices of large size , one has to restrict the number of particles , as well as the values of the interaction strength .
the mean - field gross - pitaevskii approach , on the other hand , does not have these limitations , but does not allow us to evaluate the whole dispersion relation @xmath28 at least in a straightforward way , but rather its local / absolute minima . finally , due to its mean - field character ,
possible correlations between the atoms in certain limiting cases are not captured within this approximation , although they are not expected to be important in the problem that we consider here .
we start with the case of a single component , i.e. , when the total number of atoms @xmath52 is equal to @xmath3 .
we evaluate the minimum interaction strength @xmath53 necessary for the system to support persistent currents as a function of the width of the annulus , or equivalently as a function of the confinement frequency @xmath9 .
as we argued earlier , for large values of @xmath9 , the motion becomes quasi - one - dimensional , while as @xmath9 decreases , the motion becomes two - dimensional .
we solve eqs .
( [ gpe ] ) and also diagonalize the hamiltonian numerically for several values of @xmath9 .
the results of both calculations are shown in fig .
[ fig2 ] , represented by ( black ) circles ( mean - field ) and ( red ) crosses ( diagonalization ) .
as seen from this plot , the results obtained from the two approaches are in good agreement with each other . at this point
it is instructive to see how these data compare with the analytical result of a purely one - dimensional model . to reduce the two - dimensional problem into an effectively one - dimensional one , at least in the limit where the width of the annulus ( set by the oscillator length ) is much smaller than its radius @xmath54
, one may start from the initial expression for the energy and integrate over the radial degrees of freedom @xcite . to simplify this calculation , we assume that the density in the transverse direction is nonzero only for @xmath55 , where @xmath54 is the radius of the annulus and @xmath56 is the oscillator length corresponding to @xmath9 . in this limit , the order parameter may be written in a product form , @xmath57 assuming the above factorization , one may start from the two - dimensional problem and derive the following one - dimensional nonlinear equation @xmath58 where the coefficient of the nonlinear term involves the integral @xmath59 . according to previous studies
@xcite , eq.([prodd ] ) implies that the minimum value of the coupling @xmath60 for metastability of currents with one unit of circulation , i.e. , with @xmath61 , is ( in the limit @xmath62 ) @xmath63 i.e. , it scales as @xmath64 .
clearly the numerical prefactor in the above expression depends on the form of the order parameter in the transverse direction @xmath65 .
if we define @xmath66 , this expression may be written in the more convenient form @xmath67 within this simplified model , the obtained power - law dependence of @xmath68 scaling as @xmath64 is in agreement with the numerical results plotted in fig.2 .
this figure shows that @xmath68 indeed increases linearly as a function of @xmath69 for strong confinement in the transverse direction , when the motion is quasi - one - dimensional . from the numerical data we also see that for @xmath70 , @xmath68 grows faster , as the extra degrees of freedom associated with the motion in the transverse direction start to play a role .
we also notice that for small values of @xmath9 , when the confinement becomes weak , the maximum value of the confining potential at the center of the trap decreases .
for example , for @xmath71 , the density of the system is already substantial at the center of the trap , and the deviations from quasi - one - dimensional motion are very pronounced . for the existence of persistent currents , as a function of @xmath69 in an annular trap with @xmath72 .
the ( black ) circles show the result from the mean - field approximation , and the ( red ) crosses from the diagonalization of the hamiltonian.,width=340 ] we now turn to the question of metastability in a two - component system .
again , we study the minimum interaction strength for the existence of metastable flow when the atoms are confined in a ( two - dimensional ) annular trap . to do this , we take @xmath73 as before , and consider a fixed and relatively weak confinement @xmath13 , varying @xmath74 . this can be done easily within the gross - pitaevskii scheme , where one can choose any value for the relative population @xmath75 , but not within the diagonalization approach , where one has to specify the ( integer ) particle numbers @xmath39 and @xmath40 , and therefore the values of @xmath76 are more restricted .
in addition , in order to achieve small steps in @xmath76 , the number of particles has to be fairly large , which results in hamiltonian matrices of large size , making it hard to diagonalize numerically . as before
, one can obtain a simple analytic result for @xmath68 from the result of a strictly one - dimensional trapping potential @xcite .
following the same arguments as in the previous subsection , then eq.([prodd ] ) implies that , with the assumption of a step function for the density of the gas in the transverse direction , @xmath77 which clearly reduces to eq.([gminn ] ) when @xmath78 . according to this simplified model , @xmath68 diverges at @xmath79 and thus metastability can not exist for @xmath80 . in fig .
[ fig3 ] we show the numerical results that we obtain for the minimum interaction strength necessary for the existence of metastable states as a function of @xmath76 , within the mean - field approximation .
the ( black ) circles correspond to a value of the angular momentum per particle @xmath81 equal to unity .
our calculations indicate a divergence of @xmath68 for sufficiently small values of @xmath76 , pretty much like the one - dimensional case , where this divergence occurs at @xmath82 . on the other hand
, metastability takes place at a stronger interaction strength than in the purely one - dimensional case .
therefore , the deviation from this limit works against the stability of the currents .
this result is consistent with the general statement that metastability of the superflow is not possible in trapping potentials which decrease monotonically with the distance from the center of the trap @xcite . for stability of persistent currents in an annular trap with @xmath72 and @xmath13 , as a function of the relative population @xmath76 , obtained within the mean - field approximation .
the ( black ) circles correspond to @xmath83 , the ( red ) crosses to @xmath84 , and the ( blue ) squares to @xmath85.,width=359 ] what has been mentioned so far has to do with values of the circulation equal to one unit , i.e. , equal to @xmath61 . according to ref.@xcite , in the strictly one - dimensional problem persistent currents with circulation higher than one unit
are stable only in the single - component case , i.e. , for @xmath86 ( or equivalently @xmath87 ) .
interestingly , unlike the one - dimensional case , our mean - field calculations show that persistent currents with a value of the circulation which is higher than one unit are stable for finite admixtures of a second component .
this result is shown in fig.3 in the two higher curves , represented as ( red ) crosses , and ( blue ) squares , corresponding to @xmath88 and @xmath89 , respectively .
the two limiting cases of strong and weak confinement are worth commenting on : as one approaches the one - dimensional limit , i.e. , when @xmath9 increases , the two curves corresponding to higher values of the circulation approach infinity more rapidly , eventually becoming vertical , as in the purely one - dimensional case .
thus , metastability with @xmath88 or @xmath89 is not possible in this limit , in agreement with the results of ref.@xcite .
metastability is not possible , when the confinement becomes very weak , either , i.e. , when the width of the annulus gets large , and the gas has a significantly nonzero density at the center of the trap . in other words ,
the two curves have an infinite slope in the two limiting cases ( of small and large values of @xmath9 ) , but have a finite slope for intermediate values , indicating the stability of persistent currents in this regime .
while the behavior of the system looks similar in both limits , the underlying physical origin of the non - monotonic behavior of the slope of these curves is different in the two extremes . in the limit of small @xmath9 , this is due to the fact that metastability is absent in an almost homogeneous gas @xcite . in the opposite limit of large @xmath9 ,
it is the one - dimensionality of the problem that causes this effect @xcite .
obtained from the diagonalization of the hamiltonian , in a ( two - dimensional ) annular trap , with @xmath90 atoms , @xmath13 , @xmath91 , and @xmath92 .
the ( red ) crosses show the single - component case , with @xmath93 and @xmath87 ( @xmath78 ) , while the ( black ) squares correspond to a two - component system , with @xmath94 and @xmath95 ( @xmath96 ) .
while in the first case there are two local minima in @xmath28 at @xmath97 and @xmath88 , in the second case there is only one minimum , around @xmath97.,width=340 ] the results that we have obtained from the numerical diagonalization of the hamiltonian are shown in fig .
[ fig4 ] and are consistent with those of the mean - field approximation . figure [ fig4 ] shows the dispersion relation , i.e. , the lowest eigenvalue of the hamiltonian for each value of @xmath27 , with @xmath86 ( red crosses ) and @xmath96 ( black squares ) .
more specifically , we have considered @xmath98 atoms , with @xmath93 and @xmath87 in the one case , and @xmath94 and @xmath99 atoms in the other one . in this calculation
the single - particle basis includes all the states with @xmath100 , and we consider the same parameters as before , with @xmath101 , @xmath13 , and @xmath92 .
the dispersion relation @xmath28 in the case of a single component has two local minima , at @xmath83 and @xmath84 , showing that persistent currents are stable in both cases .
this is consistent with the result obtained from the mean - field calculations shown in fig.3 . on the other hand , in the case of two components
, @xmath28 has only one local minimum around @xmath97 .
again , this is consistent with the result of fig.3 , since for the corresponding values of @xmath76 and @xmath102 , only the state with @xmath83 is metastable .
according to the present study , bose - einstein condensed atoms which move in annular traps offer a very interesting system for the study of persistent currents . in principle one could investigate a surface in a three - dimensional phase diagram showing the minimum interaction strength for stable currents as function of both the confinement strength @xmath9 , which determines the deviation from purely - one - dimensional motion , and the population imbalance @xmath76 . given the size and the numerical effort of such a calculation ,
we have instead restricted ourselves to certain parts of this diagram , and in particular we have investigated two different questions .
firstly , we considered the case of one component , and showed that as the width of the annulus becomes larger , and thus there are deviations from the strictly one - dimensional motion , the critical coupling @xmath68 that is necessary to achieve metastability increases .
this result is consistent with the fact that a necessary condition for metastability is that the confining potential and as a result the particle density does not decrease monotonically from the center of the trap ; in the picture of vortex dynamics , there is a force acting on the vortex that is associated with the rotational motion of the superfluid , which is in the opposite direction of the gradient of the single - particle density distribution . secondly , we investigated the case of a mixture of two components , in a fixed annular potential .
we found that @xmath68 which corresponds to persistent currents with one unit of circulation increases with the addition of the second component . also , we found a similar behavior as in the one - dimensional problem , where @xmath68 diverges at @xmath79 .
thus , the main effect of the finite width of the annulus is a relative increase in the critical coupling .
on the other hand , a novel result that is absent in the one - dimensional case is the stability of persistent currents with circulation for the larger component which is higher than one unit .
when the motion is one - dimensional , @xmath68 is infinite , and there are no stable currents . as the width of the annulus increases , these currents become stable for interaction strengths which are sufficiently strong , but finite .
when the annulus becomes even wider , @xmath68 becomes infinite again .
therefore , the effect of the deviation from the one - dimensional motion is rather dramatic in this case . from the above results it is clear that the phase diagram @xmath103 has an interesting structure and that the physics of persistent currents of bose - einstein condensed gases trapped in annular potentials is very rich . with the recent progress in building such trapping potentials in the laboratory
, it will be of much interest to investigate all these effects also experimentally .
this work was financed by the swedish research council .
the collaboration is part of the nordforsk nordic network `` coherent quantum gases - from cold atoms to condensed matter '' .
gmk acknowledges useful discussions with m. magiropoulos and j. smyrnakis .
|
we examine the problem of stability of persistent currents in a mixture of two bose gases trapped in an annular potential .
we evaluate the critical coupling for metastability in the transition from quasi - one to two - dimensional motion .
we also evaluate the critical coupling for metastability in a mixture of two species as function of the population imbalance .
the stability of the currents is shown to be sensitive to the deviation from one - dimensional motion .
|
studies of rr lyrae stars in the milky way globular clusters reveal what is traditionally known as the oosterhoff dichotomy ; the tendency for these clusters to be either oosterhoff i ( oo - i ) or oosterhoff ii ( oo - ii ) objects with a relatively clear zone of avoidance between these two groups .
the left panel of figure 5 in @xcite strikingly illustrates the oosterhoff dichotomy in a plot of the average period of the rr lyrae stars of bailey type ab ( rrab ) versus cluster metallicity for milky way globular clusters .
when one looks at the nearby dwarf galaxies and their globular clusters , one does not see an oosterhoff dichotomy as these objects fall not only into the oo - i / ii groups but also into the gap between those two groups .
in fact the distribution of these extragalactic objects seems to peak in the zone of avoidance @xcite .
the existence of oosterhoff - intermediate objects ( oo - int ) , objects that fall into the zone of avoidance , in the nearby dwarf galaxies poses a significant challenge to the hierarchical merger model for the formation of the milky way halo .
the lack of oo - int clusters in the halo means that the objects that were accreted to form the halo could not have had properties entirely like the present - day nearby dwarf galaxies .
this is the first of a series of papers focusing on variable stars in globular clusters in the large magellanic cloud ( lmc ) .
the goal of this series is to better understand how the rr lyrae stars in oo - int systems compare to those in oo - i / ii systems .
the lmc is an ideal place for this study because of the large number of oo - int clusters that it contains @xcite ; of the twelve lmc globular clusters that contain at least @xmath10 rrab variables , five of these clusters are oo - int ( ngc 1466 , ngc 1853 , ngc 2019 , ngc 2210 , and ngc 2257 ) . in this paper
we identify , classify , and discuss the variable stars found in the lmc globular cluster ngc 1466 .
subsequent papers will focus on the variable stars in other lmc globular clusters .
ngc 1466 is an old globular cluster that is located relatively far from the center of the lmc .
it has a metal abundance of [ fe / h ] @xmath11 , see discussion in section [ sec : abprop ] , and is not very reddened , e(b - v ) = @xmath12 @xcite . @xcite used color - magnitude diagrams obtained with the hubble space telescope to determine that the age of ngc 1466 is within @xmath6 gyr of the ages of the milky way globular clusters m3 ( ngc 5272 ) and m92 ( ngc 6341 ) .
ngc 1466 features a well - populated horizontal branch that extends through the instability strip ( see figure 5 in walker 1992 ) ; thus it is expected to contain a significant number of rr lyrae stars .
rr lyrae stars were first found in ngc 1466 by @xcite , and additional rr lyrae were discovered by @xcite .
the most recent study of variables in ngc 1466 was conducted by @xcite , who found @xmath13 rr lyrae stars ; @xmath14 rrab and @xmath15 rrc stars . due to the density of unresolved stars in the cluster core , walker did not perform photometry of stars within a radius of @xmath16 arcsec from the cluster center , suggesting that there are probably additional rr lyrae stars that could not be detected in his study .
the core radius of ngc 1466 is @xmath17 arcsec , as measured by @xcite using images from the hubble space telescope , which is entirely within the region of the cluster where walker was unable to perform photometry .
advances in instrument resolution and image - subtraction techniques now make it easier for us to search for variable stars in more crowded regions . in this paper
we present the results of a new search for variable stars in ngc 1466 and use the rr lyrae in ngc 1466 to confirm the distance modulus and oosterhoff classification of the cluster .
we also present fourier fits to the light curves and derive physical properties for the stars from these fits .
a total of @xmath18 @xmath19 and @xmath20 @xmath21 images of ngc 1466 were obtained using the soi imager on the soar @xmath22-m telescope in february and december of 2008 .
an additional @xmath23 @xmath19 and @xmath24 @xmath21 images were obtained using andicam on the smarts @xmath25-m telescope operated by the smarts consortium from september 2006 to january 2007 .
this represents a larger data set than the 35 bv pairs obtained by @xcite .
exposure times for the soar observations were between @xmath3s and @xmath26s for @xmath19 and between @xmath27s and @xmath26s for @xmath21 , but were usually @xmath28s and @xmath29s for the @xmath19 and @xmath21 observations , respectively .
smarts exposures were @xmath30s for each filter .
the images from both telescopes were bias subtracted and flat - field corrected using iraf . as noted in @xcite , ngc 1466 is located near two very bright stars which create problems with scattered light across some of the images , especially in many of the images obtained with the smarts telescope .
we used the procedure described in detail by @xcite to remove this scattered light ; this was the same method employed by @xcite .
peter stetson s daophot ii / allstar packages @xcite were used to locate , measure , and subtract stars from all images ; this produced images that only contained the sky background and scattered light .
the subtracted images were then smoothed using a @xmath18x@xmath18 pixel median filter . the average sky level in the smoothed images was determined and then the smoothed images were subtracted from the original images , removing the background light .
the average sky values were then added back into the new images , creating a final set of images with a constant sky background .
stetson s daophot ii / allstar packages were run on the new set of images in to order obtain instrumental magnitudes for each star .
observations of the landolt standard fields pg0231 , ru149 , sa95 , and sa98 @xcite were used to determine an initial transformation from instrumental magnitudes to the standard system .
@xcite showed walker s photometry to be in good agreement with hst / wfpc2 observations of ngc 1466 , after these observations were transformed from the on - board hst system to the standard johnson - cousins system .
we compared the results of our initial transformation to the standard system using walker s photometry of isolated non - variable stars across a range of brightnesses within the cluster .
we made small corrections to the zero points and color terms from our initial transformation equations in order to achieve better agreement with walker s photometry .
this produced an average difference between our data and walker s of @xmath31 and @xmath32 .
this is a similar level of scatter to that obtained by @xcite in their comparison with walker .
the profile fitting photometry used in daophot ii / allstar does not work well in very crowded regions , such as the center of ngc 1466 . in order to locate variable stars in this region , the soar data on the cluster center was also searched using the isisv2.2 image subtraction program @xcite .
the original images were used for the isis processing as the scattered light is less of a problem when using image subtraction techniques .
the light curves produced by isis are in relative fluxes .
there is a method to transform the relative fluxes obtained by isis into magnitudes in the standard system by using daophot to obtain magnitudes of the stars in the reference frame created by isis @xcite .
four of our variable stars were found only by isis ( v@xmath33 , v@xmath34 , v@xmath35 , v@xmath27 ) .
these stars are located deep in the cluster center and are too crowded to obtain accurate profile fitting photometry , even in the reference image .
these four light curves are shown in relative fluxes ; for the other variables we present their light curves obtained from daophot ii / allstar .
figure [ cmd ] shows our cmd for ngc 1466 .
the cluster features a well - defined horizontal branch ( hb ) that extends across the instability strip .
the majority of the hb stars are blue , with few hb stars located redward of the instability strip . as was noted by @xcite
, the rr lyrae stars appear to be less luminous than the blue hb stars near the blue edge of the instability strip .
as can be seen in the figure , there are some non - variable stars in the instability strip region .
these are due to either poor photometry due to blending in the cluster center or are line - of - sight field stars .
we used the @xmath19-band time series , which had more phase points than the @xmath21-band one , to identify candidate variable stars using peter stetson s allframe / trial package @xcite .
period searches on the candidate variables were carried out using supersmoother @xcite , period04 @xcite , and a discrete fourier transform @xcite as implemented in the peranso software suite .
this produced light curves which were then examined by eye to confirm the reality of each variable star candidate .
light curves of potential rr lyrae stars were then fit to template light curves @xcite and checked by eye in order to confirm variable classification and period .
resulting primary periods are typically good to @xmath36 or @xmath37 days while errors for the secondary periods of rrd stars are slightly worse but on the same order of magnitude .
a similar search of the light curves produced with isis was done to find additional variables .
lccccccccccc v1 & 03:44:44.7 & @xmath3871:40:44.9 & rrab - bl & 0.55035 & 0.64 & 0.77 & 19.380 & 19.793 & 0.421 & wa415 & 2454094.6963 + v4 & 03:44:20.3 & @xmath3871:39:57.1 & rrab - bl & 0.57239 & 0.87 & 1.16 & 19.406 & 19.804 & 0.425 & wa67 & 2454094.7100 + v5 & 03:44:26.6 & @xmath3871:39:48.9 & rrab & 0.57863 & 0.67 & 0.82 & 19.391 & 19.781 & 0.403 & wa108 & 2454510.5801 + v6 & 03:44:30.8 & @xmath3871:39:54.8 & rrab - bl & 0.52805 & 1.19 & 1.49 & 19.290 & 19.672 & 0.410 & wa182 & 2454502.2325 + v8 & 03:44:26.3 & @xmath3871:40:06.5 & rrab - bl & 0.51970 & 0.89 & 1.2 & 19.273 & 19.599 & 0.358 & wa106 & 2454094.7502 + v9 & 03:44:26.7 & @xmath3871:40:17.2 & rrab & 0.52260 & 1.24 & 1.52 & 19.414 & 19.743 & 0.362 &
wa109 & 2454086.9323 + v12 & 03:44:28.9 & @xmath3871:40:15.0 & rrab & & & & & & & wa146 & + v13 & 03:44:33.8 & @xmath3871:40:48.1 & rrc & 0.33936 & 0.55 & 0.69 & 19.298 & 19.564 & 0.276 & wa262 & 2454094.5656 + v14 & 03:44:36.2 & @xmath3871:40:48.7 & rrab - bl & 0.56414 & 0.93 & 1.29 & 19.402 & 19.803 & 0.421 & wa320 & 2454501.3551 + v15 & 03:44:35.4 & @xmath3871:40:44.2 & rrab & 0.56708 & 0.81 & 1.05 & 19.209 & 19.675 & 0.487 & wa303 & 2454094.9557 + v16 & 03:44:34.9 & @xmath3871:40:45.1 & rrab & 0.69427 & 0.53 & 0.69 & 19.360 & 19.817 & 0.466 & wa290 & 2454092.6208 + v18 & 03:44:26.7 & @xmath3871:40:14.5 & ? & 0.48837 & 0.15 & 0.29 & 18.903 & 20.168 & 1.268 & wa110 & 2454510.7987 + v19 & 03:44:47.1 & @xmath3871:40:39.8 & rrab & 0.61620 & 0.61 & 0.79 & 19.360 & 19.785 & 0.435 & wa426 & 2454097.6522 + v20 & 03:44:30.7 & @xmath3871:41:22.5 & rrab & 0.59084 & 1.05 & 1.33 & 19.280 & 19.648 & 0.396 & wa183 & 2454094.6908 + v21 & 03:44:51.9 & @xmath3871:39:53.4 & rrab & 0.52150 & 1.22 & 1.54 & 19.315 & 19.632 & 0.359 & wa448 & 2454048.0181 + v22 & 03:44:35.5 & @xmath3871:41:05.7 & rrc & 0.33081 & 0.51 & 0.65 & 19.200 & 19.543 & 0.353 & wa301 & 2454094.7036 + v23 & 03:44:38.6 & @xmath3871:41:24.7 & rrab & 0.49342 & 1.25 & 1.51 & 19.291 & 19.681 & 0.421 & wa361 & 2454094.4714 + v26 & 03:44:34.2 & @xmath3871:40:51.4 & rrab & 0.51364 & 1.03 & 1.38 & 19.339 & 19.702 & 0.400 & wa274 & 2454094.7460 + v28 & 03:44:21.3 & @xmath3871:40:38.1 & rrc & 0.32039 & 0.57 & 0.7 & 19.338 & 19.652 & 0.324 & wa71 & 2454094.7532 + v30 & 03:44:31.3 & @xmath3871:39:52.0 & rrab & 0.69445 & 0.66 & 0.91 & 19.273 & 19.677 & 0.421 & wa194 & 2454095.2544 + v31 & 03:44:32.1 & @xmath3871:39:50.5 & ? & 0.34272 & & & & & & & 2454510.6763 + v32 & 03:44:29.4 & @xmath3871:40:13.7 & ac ? & 0.49661 & 0.42 & 0.70 & 18.118 & 18.614 & 0.505 & & 2454094.7437 + v33 & 03:44:29.0 & @xmath3871:40:43.9 & rrc & 0.27841 & 0.51 & 0.57 & 19.355 & 19.600 & 0.247 & wa148 & 2454094.8536 + v34 & 03:44:28.0 & @xmath3871:39:60.0 & ? & 0.61594 & 0.31 & 0.45 & 18.194 & 18.896 & 0.709 & wa129 & 2454040.9638 + v35 & 03:44:29.7 & @xmath3871:40:42.5 & rrab - bl & 0.58172 & 0.6 & 0.78 & 19.248 & 19.668 & 0.432 & wa164 & 2454094.6090 + v38 & 03:44:49.0 & @xmath3871:40:48.7 & rrc & 0.34858 & 0.44 & 0.54 & 19.354 & 19.682 & 0.335 & wa437 & 2454092.5444 + v39 & 03:44:30.6 & @xmath3871:40:07.2 & rrab & 0.56248 & 1.12 & 1.37 & 19.156 & 19.538 & 0.407 & wa179 & 2454510.8795 + v40 & 03:44:31.0 & @xmath3871:40:01.1 & rrab & 0.53606 & & & & & & wa185 & 2454092.8046 + v41 & 03:44:39.4 & @xmath3871:40:21.3 & rr ? & & & 0.59 & & 19.598 & & & + v42 & 03:44:30.4 & @xmath3871:41:00.6 & rrab & 0.63482 & 0.51 & 0.66 & 19.384 & 19.865 & 0.489 & wa177 & 2454094.6637 + v44 & 03:44:22.0 & @xmath3871:38:13.1 & rrab & & & & & & & wa75,s27 & + v45 & 03:44:23.1 & @xmath3871:41:16.0 & rrc & 0.31043 & 0.38 & 0.49 & 19.233 & 19.503 & 0.275 & wa82,s27 & 2454094.4913 + v46 & 03:44:27.4 & @xmath3871:39:14.0 & rrab & 0.68909 & 0.38 & 0.49 & 19.352 & 19.813 & 0.466 & wa121 & 2454094.3662 + v48 & 03:44:29.4 & @xmath3871:40:00.0 & rrc & 0.37697 & 0.44 & 0.54 & 19.288 & 19.580 & 0.316 & wa153,s3 & 2454092.6531 + v49 & 03:44:30.2 & @xmath3871:40:09.5 & rrab & 0.68635 & 0.53 & 0.7 & 19.255 & 19.773 & 0.528 & & 2454510.2159 + v50 & 03:44:31.1 & @xmath3871:40:28.7 & rrab - bl & 0.58194 & 1.05 & 1.24 & 19.444 & 19.778 & 0.356 & & 2454092.5885 + v51 & 03:44:31.6 & @xmath3871:40:12.7 & ? & 0.35892 & & & & & & & 2454410.5428 + v52 & 03:44:31.6 & @xmath3871:40:34.6 & rrc & & & & & & & & + v53 & 03:44:31.9 & @xmath3871:40:15.8 & lp & & & & & & & & + v54 & 03:44:32.1 & @xmath3871:40:12.9 & ? & 0.55608 & & & & & & & 2454510.3605 + v55 & 03:44:32.5 & @xmath3871:40:13.3 & rrab - bl & 0.60951 & & & & & & & 2454510.6458 + v56 & 03:44:33.4 & @xmath3871:40:01.8 & rrab & 0.62238 & 0.73 & 0.96 & 19.189 & 19.701 & 0.527 & & 2454510.5354 + v57 & 03:44:33.5 & @xmath3871:39:51.7 & rrc & 0.35692 & 0.42 & 0.5 & 19.295 & 19.685 & 0.395 & wa249,s29 & 2454094.6595 + v58 & 03:44:33.5 & @xmath3871:39:59.0 & rrab & 0.56752 & 1.05 & 1.28 & 19.368 & 19.905 & 0.570 & & 2454510.4638 + v59 & 03:44:33.7 & @xmath3871:40:20.2 & ? & & 0.46 & 0.54 & 18.194 & 18.651 & 0.460 & & 2454511.0736 + v60 & 03:44:33.9 & @xmath3871:40:11.4 & ac?,rrab ? & 0.52210 & & & & & & & 2454510.1325 + v61 & 03:44:33.9 & @xmath3871:40:01.7 & rrab - bl & 0.66927 & 0.86 & 1.13 & 19.510 & 19.985 & 0.497 & & 2454510.6914 + v62 & 03:44:35.5 & @xmath3871:40:35.8 & ? &
0.50414 & 0.71 & 1.08 & 18.758 & 19.266 & 0.531 & & 2454094.2817 + v63 & 03:44:35.6 & @xmath3871:39:54.0 & rrab & 0.58760 & 0.82 & 0.99 & 19.415 & 19.890 & 0.490 & wa306,s31 & 2454510.6730 + v64 & 03:44:38.4 & @xmath3871:40:39.5 & rrab & 0.68767 & 0.49 & 0.59 & 19.307 & 19.809 & 0.508 & wa356,s30 & 2454092.8083 + v65 & 03:44:39.0 & @xmath3871:40:03.7 & lp & & & & & & & & + v66 & 03:44:40.0 & @xmath3871:41:10.3 & rrc & 0.33633 & 0.52 & 0.67 & 19.277 & 19.608 & 0.342 & wa382,s17 & 2454094.7809 + v67 & 03:44:45.1 & @xmath3871:42:13.1 & ? & & & & & & & & + v68 & 03:44:56.5 & @xmath3871:38:54.3 & rrc & 0.34914 & 0.47 & 0.65 & 19.410 & 19.749 & 0.348 & wa458 & 2454040.8262 + [ vartable ] a total of @xmath1 variables were found , including @xmath39 rr lyraes , @xmath6 additional candidate rr lyrae , @xmath7 long - period variables , a candidate anomalous cepheid , and @xmath8 variables of unknown classification . we were able for the first time to identify double - mode ( rrd ) variables among the rr lyrae population of ngc 1466 . of the confirmed rr lyrae stars , @xmath3 were rrab stars , @xmath4 were rrc stars , and @xmath5 were rrd s .
the rrd variable stars are discussed separately in section [ sec : rrd ] .
figures [ abcurves]-[othercurves ] show the light curves for the rrab , rrc , and the other variable stars .
table [ vartable ] lists the identified variable stars , except for the rrd stars , as well as their classification , period , @xmath19 and @xmath21 amplitudes , intensity - weighted @xmath19 and @xmath21 mean magnitudes , and magnitude - weighted mean @xmath40 color .
the intensity - weighted mean magnitudes and the magnitude - weighted mean color were obtained through the fitting of the light curves with template light curves @xcite .
( for the relation between these average quantities and the color of the equivalent static star , the reader is referred to bono et al .
notes on some of the individual stars are in the following subsections .
table [ phottable ] contains the photometric data for the variable stars .
lccccc v01 & @xmath19 & @xmath41 & @xmath42 & @xmath43 & @xmath44 + v01 & @xmath19 & @xmath45 & @xmath46 & @xmath47 & @xmath48 + v01 & @xmath19 & @xmath49 & @xmath50 & @xmath51 & @xmath52 + v01 & @xmath19 & @xmath53 & @xmath54 & @xmath55 & @xmath56 + v01 & @xmath19 & @xmath57 & @xmath58 & @xmath59 & @xmath60 + [ phottable ] we use a naming system that is an extension of the one used in @xcite .
walker identified the variable stars found by wesselink by their number in that paper , preceded by a `` we '' .
however , walker did not use an extension of wesselink s numbering scheme for the new variables that he discovered .
instead , he denoted the new variables by their number in his photometry table 3 .
it is , however , convenient to have an identification system that refers only to the known variable stars in ngc 1466 .
therefore we have extended wesselink s numbering system , adding the additional variables that are not included in @xcite to the end of wesselink s orignal list of variables .
all variables are named in the form of vxx , with stars up through v@xmath13 being ones that were originally found by wesselink . a cross identification of the original walker numbers and our variable star numbers is included in table [ vartable ] .
figures [ findout ] and [ findin ] show the locations of all of the variable stars listed in the table .
we found @xmath61 of walker s ( 1992 ) @xmath13 rr lyrae stars along with @xmath62 new rrab stars , @xmath6 new rrc , and @xmath7 new rrd stars .
six of the previously identified rr lyrae stars in ngc 1466 are reclassified as rrd stars .
we did not find walker s variable @xmath63 , which appears to be a blend of two stars in our images , neither of which display variability .
the additional rr lyrae stars are all located in the central region of the cluster , which was not searched by walker due to crowding .
the rrab stars have intensity weighted mean magnitudes of @xmath64 and @xmath65 , while the mean magnitudes for the rrc stars are @xmath66 and @xmath67 .
walker found @xmath68 and @xmath69 for the rrab stars while the rrcs had @xmath70 and @xmath71 , which are in good agreement with our results . in general our results agreed with those of walker for the stars that appeared in both of our lists , with the exceptions noted below .
walker @xmath72 , v@xmath73 , is listed as an unclassified variable for reasons that are stated in the next subsection .
we were unable to determine periods for walker s ( 1992 ) variables @xmath74 ( v@xmath75 ) and @xmath76 ( v@xmath77 ) due to not having enough data points for accurate period determination .
these two rrab stars were not detected in all images , but the shape of the partial light curves are entirely consistent with their classification as rrab stars .
v@xmath61 was classified only as a candidate rr lyrae star due to aliasing issues which prevented us from determining a period or classification as rrab or rrc .
@xcite was unable to measure this star due to its proximity to a brighter star .
no significant differences between our periods and those calculated by @xcite were found .
v@xmath78 was the only star that did not turn out to be a misclassified rrd to display a difference in period greater than @xmath79 days .
walker found a period of @xmath80 days while we found a period of @xmath81 days .
we ran a period search on walker s light curve for this star and found that there is a secondary signal for a period around @xmath82 days , suggesting that this disagreement in periods is due to aliasing in walker s light curve .
the limited accuracy of the periods determined by both walker and ourselves prevents the identification of any evolutionary rr lyrae period changes .
we did not obtain enough observations to definitively determine if any of the rr lyrae stars have the blazhko effect . however , based on the scatter in their light curves , v@xmath6 , v@xmath22 , v@xmath62 , v@xmath5 , v@xmath83 , v@xmath84 , and v@xmath85 are potential blazhko stars . v@xmath86 and
v@xmath87 also show scatter in their light curves that could be due to the blazhko effect , but these two stars are located in the crowded region near the cluster center so the scatter could also be due to contamination .
v@xmath18 features a light curve that has a fair amount of scatter as well as a shape that is unusually symmetric for an rrab star . despite its unusual shape ,
the light curve was best fit with a period of @xmath88 days and thus is tentatively classified as an rrab . the increased number of observations ( over wesselink 1971 and walker 1992 ) allowed for the identification of eight rrd stars in ngc 1466 .
table [ rrdtable ] lists the identified rrd stars , their fundamental mode and first overtone periods and amplitudes , and their period ratios .
fits to the two periods were calculated using the period04 program @xcite .
the secondary periods are typically good to @xmath89 or @xmath90 days while the primarly periods are typically slightly more accurate .
lccccccccccccccc v2 & 03:44:27.4 & @xmath3871:40:59.7 & rrd & 0.4751 & 0.3534 & 0.7451 & 0.12 & 0.42 & 0.15&0.53 & 19.375 & 19.758 & 0.389 & wa124 & 2454094.4975 + v3 & 03:44:20.2 & @xmath3871:40:00.2 & rrd & 0.4717 & 0.3505 & 0.7431 & 0.13 & 0.41 & 0.15 & 0.53 & 19.368 & 19.779 & 0.416 & wa66 & 2454021.7034 + v7 & 03:44:28.4 & @xmath3871:39:13.3 & rrd&0.4859 & 0.3615 & 0.7440 & 0.35 & 0.43 & 0.42 & 0.52 & 19.326 & 19.642 & 0.310 & wa135 & 2454094.4294 + v10 & 03:44:27.4 & @xmath3871:40:19.0 & rrd&0.4730 & 0.3518 & 0.7437 & 0.26 & 0.48 & 0.30 & 0.52 & 19.425 & 19.797 & 0.370 & & 2454094.5513 + v24 & 03:44:19.9 & @xmath3871:41:25.5 & rrd & 0.4938 & 0.3675 & 0.7442 & 0.12 & 0.43 & 0.15 & 0.60 & 19.292 & 19.642 & 0.357 & wa65 & 2454094.4549 + v25 & 03:44:36.4 & @xmath3871:39:53.9 & rrd&0.4715 & 0.3508 & 0.7440 & 0.20 & 0.44 & 0.40 & 0.65 & 19.360 & 19.705 & 0.350 & wa323 & 2454029.6612 + v29 & 03:44:15.5 & @xmath3871:40:11.9 & rrd&0.4824 & 0.3589 & 0.7440 & 0.45 & 0.58 & 0.36 & 0.38 & 19.428 & 19.750 & 0.320 & wa53 & 2454084.3778 + v47 & 03:44:28.9 & @xmath3871:40:27.8 & rrd&0.4733 & 0.3519 & 0.7435 & 0.42 & 0.58 & 0.48 & 0.60 & 19.410 & 19.760 & 0.350 & & 2454094.6365 + [ rrdtable ] six of the rr lyrae identified by @xcite turned out to be rrd stars .
five were originally identified as rrc stars and one as an rrab .
v@xmath7 , v@xmath91 , v@xmath92 , and v@xmath14 all have first overtone periods that agree with the rrc period that was found by walker .
v@xmath93 was previously classified as an rrc with a period of @xmath94 days ; however , we obtained a first overtone period of @xmath95 days for it .
walker classified v@xmath96 as an rrab star with a period of @xmath97 days , which does not correspond with either the fundamental or first overtone period that we found . of the two new rrd stars ,
v@xmath98 was thought to be a potential variable by @xcite .
@xcite did not classify this star as a variable due to its proximity to a blue hb star , which made photometry difficult .
figure [ peterson ] shows the ratio of the first overtone period ( @xmath99 ) to the fundamental mode period ( @xmath99 ) vs fundamental period ( petersen diagram ) for the rrd stars in ngc 1466 ; the rrd stars in the lmc field @xcite are also plotted .
the ngc 1466 rrd stars fall in the same area of the petersen diagram as the majority of the lmc field stars .
this is the same region where rrd stars in milky way oosterhoff - i clusters tend to fall ( see , e.g. , fig . 1 in @xcite , and fig . 13 in @xcite ) .
two long - period variables , v@xmath100 and v@xmath101 , were found in our study of ngc 1466 .
we were unable to determine periods for either of these stars ; in fact , we do not have enough light curve information to say for sure that their variability is periodic .
both stars are located near the tip of the giant branch on the cmd , as can be seen in figure [ cmd ] , and have several possible periods in the range of @xmath27 to @xmath102 days .
v@xmath103 appears to be a potential anomalous cepheid ( ac ) , having an average @xmath19 luminosity that is approximately @xmath104 magnitudes brighter in @xmath19 than the rr lyrae stars .
the star has a potential period of either @xmath105 days or @xmath106 days .
the @xmath106 day period would be consistent with the period - luminosity diagram for acs from @xcite . based on equations
@xmath7 through @xmath10 in @xcite and the distance modulus that we derive for ngc 1466 , v@xmath103 would most likely be a first overtone pulsator .
if the @xmath105 day period is correct , the star would be too bright to be an ac , based on the period - luminosity diagram , and would most likely be a foreground rr lyrae .
the shape of v@xmath103 s light curve is similar to an rrab star , which would be inconsistent with a @xmath105 day period .
based on this we feel that the @xmath106 day period is correct and that this star is either an ac or , less likely , a foreground rrab .
nine variables of unknown classification were found .
@xmath22 of these stars were only found by isis as they were either located in the crowded cluster center ( v@xmath33 , v@xmath34 , and v@xmath35 ) or near the bright star in the southeast corner of the images ( v@xmath107 ) .
thus their light curves are in differential fluxes , which does not allow us to obtain positive classifications for them . while periods and light curve shape can be used to determine potential classifications , without having a magnitude we can not distinguish between a cluster rr lyrae , a foreground rr lyrae , or an ac , for example .
notes on the other stars with unknown classification follow
. - v@xmath73 ( walker @xmath72 ) is slightly brighter than the horizontal branch but is very red , @xmath108 .
it has a period of @xmath109 days and an amplitude of @xmath110 mag in @xmath19 .
@xcite noted that in his images this star had an elongated shape that suggested a likely unresolved companion .
our images also show this star to be elongated .
the hst images of @xcite were checked in order to investigate the possibility of a close companion , however v@xmath73 did not fall within the hst field
. - v@xmath111 ( walker @xmath112 ) is about one magnitude brighter than the rr lyrae level and has a @xmath113 .
@xcite had labeled this star as a variable but @xcite considered it a constant rhb star .
the period , @xmath114 days , and light curve shape of this star suggest that it could be an rrab star whose color and brightness are affected by contamination .
- star v@xmath115 is located in the cluster center and has an isis light curve that suggests several possible periods .
its daophot light curves suggest that it has a similar color to the rr lyrae stars but is about one magnitude brighter ; this is possibly due to contamination . our images and the hst images of @xcite confirm the existence of neighbor stars that could be effecting the photometry of v@xmath115 through blending .
however , v@xmath115 displays a @xmath21 to @xmath19 amplitude ratio that is similar to the other rr lyraes in the cluster ( fig .
[ bvamp ] ) .
v@xmath115 could also potentially be an ac or a type ii cepheid .
- v@xmath27 is located in the crowded cluster center and has a clean isis light curve that suggests a period of @xmath116 days .
its light curve from daophot suggests a luminosity approximately two magnitudes brighter than the rr lyrae stars but the light curve is messy , possibly due to blending .
- star v@xmath1 is located in the cluster center and has a light curve that suggests that it is a variable with a period near @xmath117 days .
the period and the light curve shape suggest that v@xmath1 is possibly an rr lyrae but the light curve is messy and it is brighter than the other rr lyrae stars , likely due to contamination by nearby stars .
the contamination explanation is supported by the fact that v@xmath1 has a @xmath21 amplitude that is slightly too large large for its @xmath19 amplitude , compared to the rr lyrae stars in ngc 1466 .
the hst images of @xcite confirm a close neighbor for v@xmath1 which does not appear as a separate star on our images .
the rrab light curves were fit with a fourier series of the form @xmath118 where @xmath119 is the magnitude as a function of time and @xmath120 , where @xmath121 is the pulsation period .
the rrc light curves were fit in a similar fashion but a cosine series was used instead of the sine series . fourier decomposition was not carried out on the rrd stars as after we deconvolved their light curves there were not enough data points for accurate fourier analysis .
the fourier fitting was done using a fourier fitting code originally written by geza kovcs with an analysis wrapper written by nathan de lee .
error analysis for the fits was done based on @xcite .
a number of fits were performed for each light curve , with a varying number of terms , and the resulting fits were plotted and then checked by eye to determine if it was a successful fit .
not all light curves were fit successfully , and the optimum number of terms to fit each individual light curve varied .
@xmath16 rrab and @xmath93 rrc light curves were successfully fit .
tables [ abcoeff ] and [ ccoeff ] give the amplitude ratios , @xmath122 , and phase differences , @xmath123 , for the low - order terms ( @xmath124 ) , and the number terms used in the fourier fit , for the rrab and rrc stars , respectively . for the rrab stars , we also list the jurcsik - kovcs @xmath125 value @xcite .
this value helps to separate rrab stars with `` regular '' light curves from those with `` anomalous '' light curves , with lower @xmath125 values representing more `` regular '' light curves .
lccccccccc v4 & 0.346 & 0.498 & 0.301 & 0.155 & 2.16 & [email protected] & 1.12 & 30.46 & 8 + v5 & 0.243 & 0.380 & 0.361 & 0.206 & 2.31 & [email protected] & 1.69 & 49.89 & 8 + v6 & 0.422 & 0.322 & 0.303 & 0.174 & 2.68 & [email protected] & 1.83 & 9.62 & 6 + v8 & 0.349 & 0.402 & 0.324 & 0.194 & 2.47 & [email protected] & 1.39 & 40.96 & 7 + v9 & 0.463 & 0.416 & 0.244 & 0.099 & 2.30 & [email protected] & 1.10 & 7.63 & 7 + v14 & 0.355 & 0.440 & 0.351 & 0.205 & 2.25 & [email protected] & 1.34 & 14.33 & 9 + v19 & 0.252 & 0.432 & 0.177 & 0.117 & 1.94 & [email protected] & 1.17 & 6.53 & 7 + v20 & 0.363 & 0.502 & 0.356 & 0.271 & 2.26 & [email protected] & 1.26 & 45.97 & 6 + v23 & 0.424 & 0.456 & 0.361 & 0.237 & 2.25 & [email protected] & 1.04 & 47.39 & 10 + v26 & 0.367 & 0.369 & 0.249 & 0.139 & 2.26 & [email protected] & 0.78 & 34.99 & 7 + v30 & 0.232 & 0.457 & 0.305 & 0.153 & 2.36 & [email protected] & 1.93 & 23.20 & 10 + v42 & 0.204 & 0.411 & 0.251 & 0.098 & 2.46 & [email protected] & 2.20 & 10.83 & 10 + v56 & 0.276 & 0.400 & 0.282 & 0.135 & 2.50 & [email protected] & 1.67 & 5.31 & 8 + [ abcoeff ] lcccccccc v22 & 0.272 & 0.179 & 0.085 & 0.067 & 4.67 & [email protected] & 0.89 & 6 + v28 & 0.276 & 0.252 & 0.103 & 0.126 & 4.73 & [email protected] & 1.55 & 6 + v33 & 0.246 & 0.255 & 0.039 & 0.011 & 4.38 & [email protected] & 4.95 & 8 + v38 & 0.228 & 0.114 & 0.064 & 0.028 & 4.58 & [email protected] & 1.74 & 6 + v45 & 0.199 & 0.090 & 0.042 & 0.055 & 4.41 & [email protected] & 1.53 & 6 + v57 & 0.222 & 0.139 & 0.080 & 0.045 & 4.71 & [email protected] & 2.01 & 7 + v66 & 0.259 & 0.181 & 0.090 & 0.033 & 4.30 & [email protected] & 3.24 & 9 + [ ccoeff ] @xcite , @xcite , and kovcs & walker ( 1999 , 2001 ) provide formulae , determined empirically , for calculating the metallicity , absolute magnitude , and effective temperature of rrab stars using the fourier coefficients from their @xmath19 light curves .
[ fe / h ] , @xmath127 , @xmath128 , and @xmath129 are determined using the following equations , which come from equations ( 1),(2),(5 ) , and ( 11 ) in @xcite , respectively : @xmath130 } = -5.038 - 5.394\,p+1.345\,\phi_{31 } , \ ] ] @xmath131 @xmath132 and @xmath133}.\ ] ] the values for [ fe / h ] derived through this method are in the scale of @xcite , which can be transformed to the more common @xcite scale by the relation @xmath134_{j95 } } = 1.431{\rm [ fe / h]_{zw84}}+0.880 $ ] @xcite .
the dereddened @xmath40 and @xmath135 color indices were calculated using equations ( 6 ) and ( 9 ) from @xcite : @xmath136 and @xmath137 these color indices are also used to calculate effective temperatures via equations ( 11 ) and ( 12 ) in @xcite , @xmath138\end{gathered}\ ] ] and @xmath139 , \end{gathered}\ ] ] where @xmath140 is obtained from equation ( 12 ) in @xcite , @xmath141 assuming a mass of @xmath142 . the resulting physical properties for the rrab stars are given in table [ abphysical ] .
the mean metallicity of the rrab stars is @xmath134_{j95}}=-1.40\pm0.09 $ ] which in the zinn & west scale is @xmath134_{zw84}}=-1.59\pm0.06.$ ] previously published metallicities for ngc 1466 showed a wide range of values .
@xcite reported a metallicity of @xmath134}=-2.17 $ ] based on spectroscopic observations of the calcium triplet in two giant stars in the cluster whose individual metallicities were @xmath134}=-1.85 $ ] and @xmath143 .
@xcite argued that the more metal - poor of these two stars was not a cluster member based on photometric results .
walker instead reported that the cluster metallicity was @xmath134}=-1.85 $ ] , based on the remaining star from olszewski et al . and comparisons of the rr lyrae stars and the cluster cmd to that of other clusters .
@xcite used stellar popular synthesis models to find metallicities of @xmath134}=-1.30 $ ] and @xmath144 , depending on whether they used the full spectrum or just the cn spectrum
. a more metal poor result of @xmath134}=-2.25 $ ] was adopted by @xcite for their work with cluster surface brightness profiles .
a far more metal - rich value of @xmath134}=-1.64 $ ] was found by @xcite using the equivalent widths of metal lines in integrated spectra .
our value of @xmath134}=-1.59 $ ] is most consistent with the santos & piatti result .
@xcite demonstrated that the fourier parameters of the @xmath19-band light curves of rrc stars could be used to calculate various physical properties for these stars . unlike the empirically determined relationships between physical properties and fourier parameters for rrab stars , the relationships for rrc stars were determined using light curves that were created with hydrodynamic pulsation models .
the mass , luminosity , temperature , and helium parameter of rrc stars can be calculated using the equations : @xmath145 @xmath146 @xmath147 and @xmath148 which are equations ( 2),(3),(6 ) , and ( 7 ) in @xcite , respectively .
it should be noted that the helium abundance parameter , @xmath149 , is not equal to the helium abundance , @xmath150 .
equation ( 10 ) in @xcite , @xmath151 allows us to calculate the absolute magnitude for the rrc stars .
the metallicity in the @xcite scale can be calculated using equation ( 3 ) from @xcite : @xmath152_{zw84}}=52.466\,p^{2}-30.075\,p+0.131\,\phi_{31}^{2}\\+0.982\,\phi_{31}-4.198\,\phi_{31}p+2.424.\end{gathered}\ ] ] it should be noted that while these equations are commonly used in the literature , the equations for mass , luminosity , and temperature violate the pulsation equation @xcite .
while the values calculated for these properties can not all be physically correct , they can be used for the purpose of making comparisons with the physical properties calculated the same way for rrc stars in other clusters .
the calculated physical properties for the rrc stars are given in table [ cphysical ] .
lcccccccc v4 & -1.717 & 0.748 & 1.164 & 3.805 & 0.336 & 3.810 & 0.490 & 3.811 + v5 & -1.278 & 0.824 & 1.153 & 3.805 & 0.365 & 3.804 & 0.526 & 3.802 + v6 & -0.764 & 0.828 & 1.008 & 3.819 & 0.313 & 3.825 & 0.461 & 3.818 + v8 & -1.085 & 0.846 & 1.054 & 3.815 & 0.330 & 3.817 & 0.482 & 3.813 + v9 & -1.774 & 0.736 & 1.125 & 3.810 & 0.295 & 3.824 & 0.438 & 3.825 + v14 & -1.577 & 0.762 & 1.149 & 3.806 & 0.338 & 3.811 & 0.493 & 3.811 + v19 & -1.725 & 0.749 & 1.209 & 3.800 & 0.354 & 3.804 & 0.513 & 3.805 + v20 & -1.511 & 0.737 & 1.139 & 3.807 & 0.341 & 3.810 & 0.496 & 3.810 + v23 & -1.434 & 0.810 & 1.064 & 3.815 & 0.314 & 3.821 & 0.462 & 3.819 + v26 & -1.359 & 0.823 & 1.057 & 3.816 & 0.317 & 3.820 & 0.466 & 3.818 + v30 & -1.759 & 0.679 & 1.292 & 3.791 & 0.378 & 3.795 & 0.544 & 3.797 + v42 & -1.056 & 0.805 & 1.175 & 3.802 & 0.374 & 3.802 & 0.538 & 3.799 + v56 & -1.155 & 0.775 & 1.136 & 3.806 & 0.357 & 3.807 & 0.517 & 3.804 + mean & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + [ abphysical ] lcccccc v22 & -1.970 & 0.657 & 0.724 & 1.762 & 3.859 & 0.254 + v28 & -1.669 & 0.591 & 0.611 & 1.713 & 3.864 & 0.272 + v33 & -1.420 & 0.788 & 0.633 & 1.675 & 3.871 & 0.281 + v38 & -1.419 & 0.696 & 0.495 & 1.693 & 3.864 & 0.284 + v45 & -1.764 & 0.720 & 0.672 & 1.724 & 3.864 & 0.266 + v57 & -1.614 & 0.666 & 0.526 & 1.715 & 3.862 & 0.276 + v66 & -1.425 & 0.711 & 0.515 & 1.690 & 3.865 & 0.283 + mean & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + [ cphysical ] it has been noted in the literature that rrc masses calculated using the @xcite equations can at time be too small,@xmath153 , approaching the mass of a degenerate helium core at the helium flash @xcite .
table [ cphysical ] shows that this occurs for three of our stars .
this is likely due to a problem with the equations used but still produces values that are suitable for comparison purposes .
the mean metallicity of the rrc stars is @xmath134_{zw84}}=-1.61\pm0.08 $ ] , which is in good agreement with the value obtained for the rrab stars .
from the fourier fits we obtain a mean @xmath19 absolute magnitude of @xmath154 for the rrc stars and @xmath155 for the rrab stars . the zero points used for these calibrations
are different and there is some debate as to which one , if either , is correct
. instead we use the calibration of the absolute magnitude - metallicity relation for rr lyrae stars by @xcite .
the average of our fourier derived metallicities for the rrab and rrc stars is @xmath134_{zw84}}=-1.60\pm0.05 $ ] , which gives us an absolute magnitude of @xmath156 .
the average apparent magnitude of the rrab and rrc stars are @xmath64 and @xmath66 , respectively .
since the average magnitude for the rrab and rrc stars are essentially the same , we combine them for the purpose of calculating the distance , resulting in an average magnitude of @xmath157 . using the reddening value of @xmath158 from @xcite and a standard extinction law with @xmath159
, we obtain a reddening - corrected distance modulus of @xmath160 .
this distance modulus is approximately equal to the distance modulus of @xmath161 that @xcite derived for the lmc .
the average periods for the rr lyrae stars in ngc 1466 are @xmath162 days and @xmath163 days .
we found @xmath3 rrab , @xmath4 rrc , and @xmath5 rrd stars , giving the cluster a @xmath164 .
all three of these values are consistent with an oosterhoff - intermediate ( oo - int ) classification for ngc 1466 .
the minimum period for rrab stars has also been shown to be a good indicator of oosterhoff class .
ngc 1466 has a minimum period for rrab stars of @xmath165 days , which is consistent with it being an oo - int object ( catelan et al .
2011 , in preparation ) . -band amplitude ( top panel ) , for the rr lyrae stars in ngc 1466 . solid red lines and dashed green lines indicate the typical position for rr lyrae stars in oosterhoff i and oosterhoff ii clusters , respectively @xcite .
rrab stars for which we obtained a value of @xmath166 are indicated with filled circles , those with a @xmath167 are indicated with open circles , and the rrab stars for which we were unable to obtain a good fourier fit are indicated by crosses .
the bottom panel is the same as the top panel but using the @xmath21-band amplitudes.,title="fig:",scaledwidth=45.0% ] -band amplitude ( top panel ) , for the rr lyrae stars in ngc 1466 .
solid red lines and dashed green lines indicate the typical position for rr lyrae stars in oosterhoff i and oosterhoff ii clusters , respectively @xcite .
rrab stars for which we obtained a value of @xmath166 are indicated with filled circles , those with a @xmath167 are indicated with open circles , and the rrab stars for which we were unable to obtain a good fourier fit are indicated by crosses .
the bottom panel is the same as the top panel but using the @xmath21-band amplitudes.,title="fig:",scaledwidth=45.0% ] another indicator of oosterhoff status is the position of the rr lyrae stars in the bailey period - amplitude diagram .
figure [ vperamp ] shows the diagrams for ngc 1466 using both the @xmath19 and @xmath21-band amplitudes .
the typical location of the rrab and rrc stars in oosterhoff i ( oo - i ) and ii ( oo - ii ) clusters are indicated by solid red lines and dashed green lines , respectively @xcite . in both diagrams the rrab stars display a wide scatter .
many are located between the oo - i and o - ii loci , which is consistent with an oo - int object .
however , there is a significant number of rrab stars that are clustered around the oo - i line , which is more typical for an oo - i object .
the rrc stars fall almost exclusively between the reference lines , as is expected for an oo - int object .
we found an hb type of @xmath168 for ngc 1466 .
this was determined using the stars in an annulus with an inner radius of @xmath98 arcsec and an outer radius of @xmath74 arcsec ; this represented a region for which we felt our sample of the hb stars was complete .
this hb type , combined with the metallicity of @xmath134}\approx-1.60 $ ] that we obtained from the rr lyrae stars , places the cluster outside of the `` forbidden region '' where oo - int objects tend to fall in the hb type - metallicity diagram ( fig . 7 in catelan 2009 ) .
if one used a lower metallicity similar to what was favored by @xcite or @xcite , ngc 1466 would be located much closer to the forbidden region .
a lower metallicity would also be consistent with the theoretical predictions by @xcite for oo - int behavior , whereas the higher metallicity would indicate an hb morphology that is too red with respect to the predictions of the models ( see fig .
8 in catelan 2009 ) .
figure [ hist ] shows the distribution of both the raw and the fundamentalized periods for the rr lyrae stars in ngc 1466 .
a sharply peaked fundamentalized period distribution is not present , in contrast to what was found in @xcite in a number of ooi and ooii clusters .
the distribution of the raw periods for the rrab stars displays two peaks , a broad one around @xmath169 days and a sharper peak at @xmath170 days .
this is similar to the distribution seen in ngc 1835 @xcite although the shorter period peak for ngc 1466 is broader and its longer period peak occurs at a slightly longer period .
soszyski suggested that this two - peaked distribution may be an indicator of a sum of oo - i and oo - ii characteristics , but our bailey diagram shows no evidence of an oo - ii population .
tables 7 and 8 in @xcite show the mean physical parameters for rrc and rrab stars , respectively , in a selection of previously studied globular clusters .
the parameters that we calculated for ngc 1466 fall between those for oo - i and oo - ii clusters , consistent with ngc 1466 having an oo - int classification .
a more thorough comparison of the physical parameters for the rr lyrae stars in ngc 1466 and the other globular clusters in our study will be the subject of a future paper in this series .
a photometric study of the lmc globular cluster ngc 1466 was conducted in order to locate variable stars , with a specific interest in the cluster s rr lyrae population . @xmath39 rr lyrae stars and @xmath6 candidate rr lyrae were found .
we discovered @xmath8 rr lyrae stars which were not found by @xcite , previously the most extensive search for rr lyrae in ngc 1466 .
of our @xmath39 rr lyrae stars , @xmath5 are rrd stars , making this the first time double - mode rr lyrae stars were found in this cluster . a candidate anomalous cepheid variable was found in ngc 1466 .
acs are rare in globular clusters , having previously been found only in ngc 1786 , ngc 5466 , and @xmath171 cen @xcite .
interestingly , these three systems are ooii , though this of course may be just a coincidence , particularly since it is believed that acs in globular clusters should be the progeny of blue straggler stars ( e.g. , sills et al . 2009 and references therein ) , which in principle should not be related to the oosterhoff dichotomy .
we performed fourier analysis on the rr lyrae light curves and used the resulting fourier parameters to calculate physical properties for the stars which will be compared to the properties of rr lyrae stars in other clusters in a future paper in this series .
we obtained a reddening - corrected distance modulus of @xmath172 from the rrab and rrc stars .
the average periods for the rrab and rrc stars and the ratio of number of rrc to total number of rr lyrae all suggest an oosterhoff - intermediate classification for ngc 1466 .
the position of the rrab stars in the bailey diagram provides for a less clear oosterhoff determination .
we wish to thank the referee , j.m .
nemec , for his comments and suggestions which helped improve this paper .
support for m.c . is provided by proyecto fondecyt regular # 1110326 ; by proyecto basal pfb-06/2007 ; and by fondap centro de astrofsica 15010003 .
m.c . and j.b .
are supported by the ministry for the economy , development , and tourism s programa inicativa cientfica milenio through grant p07 - 021-f , awarded to the milky way millennium nucleus .
support for h.a.s . and
is provided by nsf grants ast 0607249 and ast 0707756 .
alard , c. 2000 , a&as , 144 , 363 bono , g. , caputo , f. , & stellingwerf , r. f. 1994 , , 423 , 294 bono , g. , caputo , f. , & stellingwerf , r. f. 1995 , , 99 , 263 cacciari , c. , corwin , t. m. , & carney , b. w. 2005 , , 129 , 267 catelan , m. 2004a , , 600 , 409 catelan , m. 2004b , in variable stars in the local group , asp conf .
ser . , 310 , ed .
kurtz , & k.r .
pollard ( san francisco : asp ) , 113 catelan , m. 2009 , , 320,261 catelan , m. , & corts , c. 2008 , , 676 , l135 clement , c.m .
et al . 2001 , , 122 , 2587 clementini , g. , corwin , t. m. , carney , b. w. , & sumerel , a. n. 2004 , , 127 , 938 contreras , r. , et al .
2010 , , 140 , 1766 corwin , t.m . , catelan , m. , smith , h.a . , borissova , j. , ferraro , f.r .
, & raburn , w.s .
2003 , , 125 , 2543 deb , s. , & singh , h.p .
2010 , , 402 , 691 ferraz - mello , s. 1981 , , 86 , 619 johnson , j.a . ,
bolte , m. , stetson , p.b .
, hesser , j.e . , & somerville , r.s .
1999 , , 527 , 1999 jurcsik , j. 1995 , aca , 45 , 653 jurcsik , j. 1998 , , 333 , 571 jurcsik , j. , & kovcs , g. 1996 , , 312 , 111 jurcisk , j. , & kovcs , g. 1999 , newa rev . , 43 , 463 kovcs , g. 1998 , , 69 , 49 kovcs , g. , & walker , a.r . 1999 , , 512 , 271 kovcs , g. , & walker , a.r . 2001 , , 371 , 579 landolt , a.u . 1992 , , 104 , 340 layden , a.c . 1998 , , 115 , 193 lenz , p. , & breger , m. , 2005 , commun .
asteroseismol . , 146 , 53 mackey , a.d . , & gilmore , g.f .
2003 , , 338 , 85 mochejska , b. j.,kaluzny , j. , stanek , k. z. , sasselov , d. d. , & szentgyorgyi , a. h. 2001 , , 121 , 2032 ; erratum : , 121 , 3284 morgan , s.m . , wahl , j.n . , & wieckhorst , r.m .
2007 , , 374 , 1421 nemec , j.m . ,
nemec , a.f.l , & lutz , t.e .
1994 , , 108 , 222 nemec , j.m .
, walker , a.r . , jeon , y.b .
2009 , , 138 , 1310 olsen , k.a.g . , et al .
1998 , , 300 , 665 olszewski , e.w .
, schommer , r.a . ,
suntzeff , n.b . , & harris , h.c .
1991 , , 101 , 515 peterson , j.o .
1986 , , 170 , 59 popielski , b. l. , dziembowski , w. a. , & cassisi , s. 2000 , aca , 50 , 491 pritzl , b.j . , armandroff , t.e . ,
jacoby , g.h .
, & da costa , g.s .
2002 , , 124 , 1464 pritzl , b.j . ,
armandroff , t.e . ,
jacoby , g.h .
, & da costa , g.s .
2005 , , 129 , 2232 reimann , j.d .
1994 , phd thesis ( university of california ) santos jr .
, j.f.c . , & piatti , a.e .
2004 , , 428 , 79 sills , a. , karakas , a. , & lattanzio , j. 2009 , , 692 , 1411 simon , n.r . , & clement , c.m .
1993 , , 410 , 526 soszyski , i. , et al .
2003 , aca , 53 , 93 stetson , p.b .
1987 , , 99 , 191 stetson , p.b .
1992 , in asp conf .
25 , astronomical data analysis software and systems i , ed .
worrall , c. biemesderfer , & j. barnes ( san francisco : asp ) , 297 stetson , p.b .
1994 , , 106 , 250 stetson , p.b . , & harris , w.e .
1988 , , 96 , 909 thackeray , a.d . , &
wesselink , a.j .
1953 , nature , 171 , 693 walker , a.r .
1992 , , 104 , 1395 wesselink , a.j . 1971 , , 152 , 159 wolf , m.j . ,
drory , n. , gebhardt , k. , & hill , g.j .
2007 , , 655 , 179 zinn , r. , & west , m.j . 1984 , , 55 , 45 zorotovic , m. , et al .
2010 , , 140 , 912
|
this is the first in a series of papers studying the variable stars in large magellanic cloud globular clusters .
the primary goal of this series is to better understand how the rr lyrae stars in oosterhoff - intermediate systems compare to those in oosterhoff i / ii systems . in this paper
we present the results of our new time - series @xmath0 photometric study of ngc 1466 .
a total of @xmath1 variables were identified in the cluster , of which @xmath2 are new discoveries .
the variables include @xmath3 rrab stars , @xmath4 rrc s , @xmath5 rrd s , @xmath6 candidate rr lyrae , @xmath7 long - period variables , @xmath6 potential anomalous cepheid , and @xmath8 variables of undetermined classification .
we present photometric parameters for these variables . for the rr lyrae stars
physical properties derived from fourier analysis of their light curves are presented .
the rr lyrae stars were used to determine a reddening - corrected distance modulus of @xmath9 .
we discuss several different indicators of oosterhoff type and find ngc 1466 to be an oosterhoff - intermediate object .
|
semiflexible polymers such as filamentous proteins resemble elastic rods on a molecular scale , while exhibiting significant thermal fluctuations on the scale of micrometers or even less .
this has made them useful as model systems allowing for direct visualization via optical microscopy .
but , semiflexible polymers are not just large versions of their more well - studied flexible cousins such as polystyrene .
filamentous proteins , in particular , have been shown to exhibit _ qualitatively _ different behavior in their networks and solutions .
a fundamental reason for this is the fact that the thermal _ persistence length _ , a measure of filament stiffness as the length at which thermal bending fluctuations become apparent , can become large compared with other important length scales such as the spacing between polymers in solutions , or the distance between chemical crosslinks in a network .
one of the most studied semiflexible polymers in recent years has been f - actin , a filamentous protein that plays a key structural role in cells @xcite .
these occur in combination with a wide range of specific proteins for crosslinking , bundling and force generation in cells .
these composites , together with other filamentous proteins such as microtubules , constitute the so - called _ cytoskeleton _ that gives cells both mechanical integrity and structure . this biopolymer gel is but one example of a large class of polymeric materials that can store elastic energy in a combination of bending and extensional deformations of the constituent elements .
such systems can be called semiflexible gels or networks .
one of the important lessons from recent experimental and theoretical studies is that the shear modulus of cross - linked semiflexible networks bears a much more complex relationship to the mechanical properties of the constituent filaments and to the microstructure of the gel than is the case for flexible polymer gels @xcite .
recently , it has been shown that semiflexible gels exhibit a striking cross - over @xcite between a response to external shear stress that is characterized by a spatially heterogeneous strain ( a _ non - affine _ regime @xcite ) and a uniform strain response ( an _ affine _ regime @xcite ) .
this crossover is governed primarily by cross - link density and molecular weight ( filament length ) .
the bulk shear modulus of the network simultaneously increases by about six orders of magnitude at this nonaffine - to - affine cross - over .
the underlying mechanism responsible for this abrupt cross - over appears to be the introduction of a new , mesoscopic length scale in the problem that is related to both the bending stiffness of the constituent polymers and the mean spacing between consecutive cross - links along the chain @xcite .
one can associate this mesoscopic length with the length below which the deformation of the network departs from the standard affinity .
the nonaffinity length @xmath0 , introduced in refs.@xcite , can be qualitatively understood as the typical length over which one finds nonaffine deformation in the network . in this previous work we presented a scaling analysis that relates this mesoscopic length scale to the network density and the stretching and bending moduli of the constituent filaments .
the macroscopic shear response of the network is then controlled by a competition between @xmath0 ( the nonaffinity length ) and the filament length , @xmath1 . on the one hand , when the filament length is long , nonaffine corrections to the deformation field , which are localized to regions within @xmath0 of the filament ends ,
do not significantly affect the mechanical properties of the network ; the shear modulus of the macroscopic system is well - described by calculations based on affine deformation reflecting the fact that nonaffine deformations of the filament ends are subdominant corrections in this limit . moreover ,
the elastic energy is stored primarily in the ( homogeneous ) extension and compression of filaments . on the other hand , when the filaments are of a length comparable to , or shorter than the nonaffinity length , _ i.e. _ @xmath2 then the nonaffine deformations of the ends play a large , even dominant role in determining the mechanical properties of the network .
the network is found to be generally more compliant , and the elastic energy under applied shear stress is stored in a spatially heterogeneous manner in the bending of filaments .
the existence of these distinct regimes as a function of filament length reflects a fundamental difference of these semiflexible polymer networks with respect to their flexible counterparts : polymers can maintain their mechanical integrity and state of stress / strain across network nodes or crosslinks .
these results naturally lead one to pose a number of basic questions regarding the elastic properties of semiflexible networks . while these random networks must on basic theoretical grounds appear as continuum , isotropic materials at the longest length scales , these considerations do not predict the length at which the continuum approximation applies . the previous identification of the nonaffinity length , @xmath0 as the only mesoscopic length associated with the nonaffine - to - affine cross - over in uniformly sheared semiflexible networks suggests that this length should more generally control the cross - over to continuum behavior @xcite .
after all the affine deformation of the network under uniform stress at scales large compared to @xmath0 shows that in one case at least the nonaffinity length controls the cross - over to continuum behavior .
one of the principal results of the present work is the demonstration that @xmath0 more generally controls this cross - over to continuum mechanics in semiflexible gel systems .
prior work has focussed exclusively on simple shear and uniaxial extension . in order to better examine the universality of the previous results ,
we study the opposite limit of a highly localized external force in the form of a point force monopole or dipole . if one can show that the elastic ( displacement ) green s function of the system similarly depends on only one additional parameter @xmath0 then it would appear that this quantity completes the coarse - grained elastic description of the system on all length scales down to the mean distance between cross - links .
it may be , however , that the deformation field of these semiflexible networks is much more complex and the simplification introduced by @xmath0 in the description of the network s response to uniform shear strain can not be generalized to deformations resulting from more general forcing conditions . while our results below , indeed , show that @xmath0 does largely control the crossover to ( the _ far field _ ) continuum elasticity , the observed elastic green s function is sensitive to the local structure of the network on length scales below @xmath0 .
below we discuss how we quantify the structure of the green s function and its approach to the form required by continuum elasticity .
the observed elastic green s function , however , depends not only on the @xmath0dependent lam coefficients of the material , but also on local properties of the displacement field immediately surrounding the the point force . in effect
one may imagine that , upon the application of the point force , the network acts as a type of composite material : within a distance @xmath0 of the point force it deforms in a way not well described by continuum theories , while outside of that zone it does appear to act like an elastic continuum .
the complete green s function depends , of course , on the material properties of both media .
unfortunately , only one of those media ( the outer zone ) is well characterized by the simple continuum theory of an isotropic elastic solid , so the complete green s function remains complex and depends on the detailed network structure within the inner zone surrounding the applied force .
there is another set of questions that may be addressed via the study of the network s response to point forces and force dipoles .
such forces not only probe the material properties of the network in a manner complimentary to the uniform strains explored earlier , they also have direct physical implications for microrheology in f - actin networks and for the dynamics of the cytoskeleton in response to the activity of nanoscale molecular motors , _ e.g. _ myosin .
fluctuation - based microrheology , an application of the fluctuation - dissipation theorem to the study of rheology via the statistical analysis of the thermally fluctuating position of sub - micron tracer particles embedded in the medium , requires one to understand the elastic green s function of the medium .
thus understanding the response of semiflexible networks to localized forces has direct experimental implications and consequences for force production in the cytoskeleton . in the biological context , the semiflexible network making up the cytoskeleton is generally found in association with molecular motors that , to a good approximation , generate transient localized force dipoles in the material . to both
understand force generation in the cell as well as the material properties of these cytoskeletal networks driven into nonequilibrium steady states by these molecular motors , one must determine the displacement field associated with such motor - induced forces .
notwithstanding our biological motivation for this work , our findings also bear on the broader problem of elastic modes in amorphous materials .
it has been shown that the vibrational modes of deep quenched lennard
jones systems approach a continuum description only on scales exceeding some mesoscopic length @xmath3 @xcite ; for the protocols considered , a value @xmath4 particle dimensions was robustly found .
this was physically identified with a length scale for non affinity , suggesting a direct correspondence with our @xmath0 ( although our @xmath0 can be controlled by varying the mechanical properties of the constituents ) .
a comparable length was also found to control the self averaging of the green s function to the form expected by continuum elasticity @xcite .
these findings for radially interacting particles are broadly in keeping with our own investigations for semiflexible polymer networks .
we also mention here that the relationship between continuum elasticity and the green s function has also been discussed for mildly disordered spring networks @xcite .
the remainder of this paper is organized as follows : in section [ s : model ] we develop our model of semiflexible , permanently cross - linked gels , summarize the numerical simulations used to study it , and discuss the expected structure of the displacement field when averaged over numerous realizations of the network . in section [ s : results ] we report our results for both point forces in [ ss : point - force ] and force dipoles in [ ss : dipole ] .
we then discuss our studies of the bulk elastic properties of these networks in [ ss : bulk ] before concluding in section [ s : discussion ] .
a highly successful continuum continuum model of individual semiflexible polymers is the _ worm like chain_. this treats the linear filaments as elastic rods of fixed contour length and negligible thickness , so that the dominant contribution to the elastic energy comes from bending modes and the hamiltonian linearized to small deviations from a straight configuration that is given by @xmath5 where @xmath6 is the transverse displacement of the filament relative to an arbitrary straight axis , @xmath7 is its arc length , and the elastic modulus @xmath8 gives the bending energy per unit length @xmath9 .
the longitudinal response of the wormlike chain model is calculated from the increase in free energy due to an extensional stress applied along the mean filament axis @xcite . however , the numerical algorithm used in our simulations is based on the minimization of the hamiltonian of the system , and hence is fundamentally athermal . the reason for this choice is essentially one of efficiency : assuming there are no subtle convergence issues , minimization is expected to be faster than stochastic modelling and , it is anticipated , give better statistics for a given cpu time .
this does , however , mean that the entropic mechanism governing the longitudinal response is absent , and an explicit energetic term is required .
an unconstrained filament at @xmath10 forms a straight configuration , and thus elongation of its end to
end distance must be accompanied by a change in the absolute contour length .
it is therefore natural to incorporate longitudinal modes by adding a second elastic term to the hamiltonian for the extension or shortening of the filament backbone , @xmath11 where @xmath12 gives the strain or relative change in local contour length , and @xmath13 is the young s modulus of the filament ( essentially a spring constant normalized to 1/[length ] ) .
of course , such modes also exist in thermal systems , but may be dominated by the entropic spring terms @xcite , except possibly for very short filament segments or very densely cross - linked gels .
the connection between thermal / entropic and athermal longitudinal compliance is discussed in greater detail in ref .
@xcite .
the two elastic coefficients @xmath8 and @xmath13 together define a length scale @xmath14 , which will shall refer to as the intrinsic _ bending length _ by observing that an isolated filament constrained to have different tangents at its end points will deform with this characteristic length . to avoid potential confusion , however , we note that this is _ not _ the typical length scale for bending deformations of a semiflexible filament . rather , the bending energy of filaments tends to make the longest unconstrained wavelength bending mode the dominant one .
thus , for instance , in a crosslinked gel , filaments are expected to be bent primarily on a length comparable to the distance between crosslinks .
nonetheless @xmath15 is a useful measure of filament rigidity , in that large @xmath15 corresponds to rigid filaments , and small @xmath15 to flexible ones .
although @xmath8 and @xmath13 have been introduced as fundamental coefficients , if the filament is regarded as a continuous elastic body with uniform cross section at zero temperature , then they can both be expressed in terms of the characteristic filament radius @xmath16 and intrinsic bulk modulus @xmath17 as @xmath18 and @xmath19 .
thus @xmath20 , and thinner filaments are more flexible than thick ones ( as measured by @xmath15 ) , as intuitively expected .
given the possibility of entropic effects in @xmath13 , however , we shall treat these as independent parameters of the theory @xcite the gel is constructed by depositing filaments of monodisperse length @xmath1 and zero thickness onto a two dimensional substrate . the center of mass position vector and orientation of the filaments
are uniformly distributed over the maximum allowed range , so the system is macroscopically isotropic and homogeneous . whenever two filaments overlap they are cross - linked at that point .
deposition continues until the required mass density , as measured by the mean distance between crosslinks @xmath21 , has been reached .
the network thus constructed can be described by three lengths : @xmath22 , and @xmath21 and one modulus scale : @xmath13 . in two - dimensions
@xmath21 characterizes both the mass density and cross - link density in spatially random , isotropic networks .
the length @xmath15 characterizes the mechanical properties of the constituent filaments via the ratio of their bending to stretching compliance .
the overall modulus scale @xmath13 will be absorbed into the point forces applied to the network .
previously @xcite , we identified an additional length @xmath23 , where @xmath24 . this _
non affinity length _ characterized the crossover from non affine to affine network response .
specifically , for filament lengths @xmath1 much larger than @xmath0 ( _ i.e. _ , high molecular weight ) , the bulk network properties could be understood quantitatively in terms of affine strains , while significant non affine effects were observed for @xmath25 .
although this length arose naturally from considerations of bulk network properties @xcite , its dependence on @xmath21 and @xmath15 can be understood in terms of a balance of stretching / compression and bending energies of a single filament , treated in a self consistent way within a network @xcite .
it is important to note that this ( material ) length is intermediate , between the ( geometric ) network length @xmath21 and the macroscopic scale .
in fact , in dilute networks , for which @xmath26 , we shall argue below that the network can be thought of on scales @xmath27 as a quasi continuum : continuous , as opposed to discrete , but not necessarily described by macroscopic continuum elasticity .
we shall find that this non affinity length will play a key role in our analysis of the displacement field on this intermediate / quasi continuum scale . in order to apply the point forces ,
a crosslink is chosen at random and identified as the origin of the system ; it is this crosslink that will later be perturbed .
a fixed circular boundary at radius @xmath28 from the origin is imposed , and any filaments or filament segments that extend beyond the boundary are simply removed or truncated , respectively .
filaments ending on the rigid boundary are fixed there by another freely rotating bond as are found at all cross - links in the system .
the allowed free rotation at the boundary means that the boundary supplies arbitrary constraint forces on the network but can not support any localized torques .
details of the simulation method have been presented elsewhere @xcite .
here we briefly summarize the procedure , with particular attention on those aspects that are central to the problems studied in this paper .
the system hamiltonian @xmath29 is constructed from discrete versions of ( [ e : h_wlc ] ) and ( [ e : h_hooke ] ) applied to the geometry generated by the random deposition procedure described above .
the degrees of freedom @xmath30 , or ` nodes , ' are the position vectors of both crosslinks and midpoints between crosslinks , the latter so as to incorporate the first bending mode along the filament segments .
additional nodes could be included at the cost of additional run time , but are expected to have a small effect , since the longest unconstrained wavelengths tend to dominate bending deformations .
crosslinks are treated as constraints on the relative position of each connected filament segment , but not on their relative rotation .
physically this corresponds to an inextensible but freely rotating linkage .
as previously noted , constrained bending at crosslinks has a small effect except at high network concentrations ( specifically , when @xmath21 becomes comparable to @xmath16 ) @xcite .
nodes on the boundary are immobile .
note that , as in our earlier work @xcite , the network is assumed to be initially unstressed on both macroscopic and microscopic length scales .
there are two ways in which the system may be perturbed .
the first , which we call _ monopole _ forcing , is to apply an arbitrarily small external force @xmath31 to the crosslink at the origin .
the network is then allowed to relax to a new configuration consistent with mechanical force balance at every node .
the second , which we call _ dipole _ forcing , is to introduce a geometrical defect into the system by moving the central crosslink along the contour length of one of the filaments to which it belongs , but not the other .
physically , this corresponds , _
e.g. _ , to a motor introducing relative motion of one filament with respect to another filament . in the simulations ,
this effect is incorporated by infinitesimally ` shifting ' the image of the central node with respect to other nodes on one filament .
once the perturbation has been specified , the displacements of the nodes in the new mechanical equilibrium are calculated by minimizing the system hamiltonian @xmath29 using the conjugate gradient method @xcite .
this generates a displacement field for the particular geometry under consideration , as shown in fig .
[ f : eg_indiv_run ] .
note that @xmath29 is linearized about small nodal displacements @xmath32 from their original positions @xmath33 , so linear response is assured .
the bulk response to non linear strains has been recently studied by onck _
et al . _
@xcite .
( grey line segments ) perturbed by an external force , denoted by the large red arrow , applied to the crosslink at the origin of a circular system of radius @xmath34 .
the arrow lengths are logarithmically calibrated to the magnitude of the displacement of at each crosslink . in this example , @xmath35 , @xmath36 and the force is perpendicular to one of the filaments that form the central crosslink ; forces can also be applied parallel to a filament .
, width=302 ] as shown in fig .
[ f : eg_indiv_run ] , the displacement field for a particular network is quite complex and generically shows anti - correlations between displacements and the local mass distribution .
although these fluctuations reflect inherent and possibly interesting physical properties of the gels , a more basic and immediately applicable quantity to measure is the _
mean _ displacement field , found by averaging many individual runs with differing geometries but identical system parameters @xmath28 , @xmath1 , @xmath21 and @xmath0 ( or equivalently @xmath28 , @xmath1 , @xmath21 and @xmath15 ) .
two examples for differing @xmath0 ( the radius of the green circle centered on the point of force application ) are given in fig .
[ f : eg_averaged ] .
these plots demonstrate one of the primary results of this paper , namely that the crossover between continuum ( or quasi
continuum ) response at large lengths , to a more exotic displacement field at shorter lengths , happens at a length of order @xmath37 with a prefactor close to unity .
networks with the same parameters as in fig .
( [ f : eg_indiv_run ] ) ( including one filament fixed perpendicular to the external force ) except for @xmath0 , which is _ ( a ) _ @xmath38 and _ ( b )
_ @xmath39 . for easy visualization
, a green circle of radius @xmath0 has been inserted into the background of each plot .
vectors near the center of each system have not been plotted for clarity .
, title="fig:",width=302 ] networks with the same parameters as in fig .
( [ f : eg_indiv_run ] ) ( including one filament fixed perpendicular to the external force ) except for @xmath0 , which is _ ( a )
_ @xmath38 and _ ( b ) _ @xmath39 . for easy visualization
, a green circle of radius @xmath0 has been inserted into the background of each plot .
vectors near the center of each system have not been plotted for clarity .
, title="fig:",width=302 ] in order to precisely describe the structure of the deformation field , we consider its most general possible form . for monopole forcing , the displacement field @xmath40 at position vector @xmath41 relative to the origin
can be projected onto 3 other vectors , namely the direction of the external force @xmath42 , the unit position vector @xmath43 and the axis @xmath44 of one of the filaments to which the crosslink is attached , @xmath45 where @xmath46 is the magnitude of the external force , and @xmath47 is the shear modulus as predicted for affine deformation .
this depends on @xmath1 and @xmath21 but _ not _
@xmath0 , and is included here to factor out the density dependence of the response .
as defined , the @xmath48 ( and the @xmath49 below ) are dimensionless quantities in two dimensions .
each @xmath48 can be further decomposed into angular modes in @xmath50 , where @xmath51 , @xmath52 terms in @xmath53 vanish since the ensemble averaged response must be invariant under @xmath54 , and @xmath55 terms with @xmath56 odd also vanish due to @xmath57 invariance .
this latter symmetry may appear to violate the known polarity of typical semiflexible biopolymers such as f - actin and microtubules @xcite ; however , we are interested here in the _ mechanical _ properties of the filaments , which , within the approximations of our model , are indeed invariant under @xmath57 . for dipole
forcing , the above procedure is followed with the additional simplification that @xmath58 is proportional to @xmath59 , since the displacement ( and hence force ) dipole induced by the motion of a motor will always be parallel to one filament axis .
the decomposition is therefore somewhat simpler , @xmath60 the scalar @xmath46 is the magnitude of the force dipole ; since it is actually a displacement that is imposed , @xmath46 is unknown as will be treated as a fitting parameter .
the angular decomposition is identical to before , @xmath61 later sections will refer to the continuum solution for each of the two forms of forcing .
these are given in the appendix .
for the monopole case , only two continuum modes are non
zero , namely @xmath62 and @xmath63 .
we shall refer to these components of the green s function as _
continuum modes _ in order to distinguish from those components ( _ non continuum modes _ ) that must vanish in the continuum .
we do this even though for our finite systems the _ non continuum modes _ do not vanish .
the dipole modes are slightly more subtle : after averaging of many dipole fields generated by the simulation , the resulting field is _
quadrupolar_. this is an immediate consequence of the means of forcing the system .
recall that in an elementary step , a motor moves parallel to one filament axis .
this has the effect of compressing the filament in front of the dipole , while stretching the trailing segment .
thus two filament segments are perturbed , each of which can be treated as a force dipole which , for a particular network geometry , will be of different magnitudes and hence the resulting field is dipole .
however , the net bias of this dipole is symmetrically distributed around zero , and thus vanishes after averaging , leaving a quadrupole field as shown in fig . [ f : eg_quadrupole ] . as derived in the appendix , the non
zero modes for this field are @xmath64 , @xmath65 , @xmath66 and @xmath67 .
individual fields induced by imposing a displacement dipole at the origin .
the orientation of the dipole is given by the red arrow .
as explained in the text , the mean dipole moment vanishes after averaging and the resulting field is quadrupole . a green circle at a radius @xmath68 from the origin is also shown .
, width=302 ]
the mechanical response to a localized perturbation depends on the distance from the point of perturbation .
we divide the discussion of these results into the following parts .
first , we examine the response to a point force using the decomposition of the displacement field outlined in previous section .
we contrast the decay of the non - continuum modes of the displacement field with the behavior of the continuum modes of the displacement field and discuss the finite - size effects in the simulations . at large length scales , we find that the deformation field approaches a quasi continuum form , in which all but a small number of ensemble average modes decay rapidly toward zero .
the remaining modes of the deformation field are the same as those predicted by continuum elasticity . in other words ,
the strain field about a point force reflects the expected tensorial character and rotational symmetries based on continuum elasticity theory .
we find that @xmath0 again plays a central role in controlling the cross - over from the near - field to the quasi - continuum .
we then turn to the spatial structure of the elastic energy density field around the point force paying particular attention to the partitioning of that energy density between stretching and bending modes of the filaments .
we observe that the ratio of these energy contributions achieves the bulk value over much shorter distances from the point force than does the structure of the displacement field acquire its far - field , or bulk structure .
we then extend our analysis to consider force dipoles in the medium .
lastly , we extend our previous analysis of the bulk elasticity of the filament network by examining both the young s modulus and the poisson ratio of the network . in this section
we focus on the short length scale behavior of the monopole response , for which the non continuum modes are non zero .
we wish to distinguish two distinct forms of this non continuum behavior : _
( i ) _ higher angular modes @xmath69 and @xmath70 with @xmath71 are non zero , and _
( ii ) _ the @xmath72 modes do not vanish , _
i.e. _ the response depends on the orientation of the filament to which the force is applied .
this latter observation gives a clear indication of how the response can ` see ' the microscopic structure of the gel on short length scales .
an example demonstrating the appearance of non continuum modes at short lengths is given in fig .
[ f : gf2_5p2_r1l ] , which shows the @xmath73 mode for systems with different crosslink densities @xmath74 but with the filament flexibility chosen to give the same @xmath75 in each case .
this plot shows the decay of the @xmath76 amplitude of the component of the displacement field in the direction of the applied force ( at the origin ) . in all three systems
one observes a rapid decay of this angular harmonic that must vanish for a continuum isotropic system .
moreover , the characteristic length scale for this decay appears to be of order @xmath0 ( @xmath77 ) , although we examine this point more quantitatively below . for all network densities and values of @xmath0
studied it is clear that the magnitude of @xmath73 vanishes rapidly with distance from the point force .
no data are shown for larger values of @xmath74 since at high network densities the numerical convergence of the strain field is so slow as to prevent attaining meaningful statistics .
( which is dimensionless ) versus distance from point of force application @xmath78 for systems with the same @xmath36 and crosslink densities @xmath74 as given in the key ( @xmath1 is the filament length ) .
the system radius @xmath34 in all cases . ,
width=321 ] this picture of non
continuum modes decaying to near zero at a length comparable to @xmath0 holds also for the @xmath79 modes of @xmath70 and for all the @xmath71 modes of @xmath69 . for reasons of space we do not show these data here . instead we present data in fig .
[ f : gn0_5p2_r1l ] that demonstrate the decay of @xmath80 with @xmath78 for three networks densities such that in each case @xmath81 .
this non continuum mode of the displacement field measures the circularly averaged amplitude of the component of displacement in the direction @xmath82 , _
i.e. _ along the axis of the rod to which the force has been applied .
clearly , such behavior has no counterpart in an isotropic continuum elastic model .
it is interesting to note that this amplitude also appears to decay exponentially with a decay length of order @xmath0 . versus @xmath83 for @xmath36 , @xmath34 and the @xmath74 given in the key .
, width=321 ] in order to test quantitatively whether @xmath0 indeed controls the decay of those components of the displacement field that have no counterparts in the continuum theory , we examine , as an example , @xmath84 vs. @xmath78 more closely in fig .
[ f : log - lin - gf2 ] . here
we plot this for a range of values of networks density _ and _ of @xmath0 . if , as suggested above , the decay is exponential with characteristic length @xmath0 , then plotting these data log - linear with radial distances scaled by @xmath0 should cause all these curves to exhibit the same slope .
we have shifted the data sets vertically in order to facilitate visual comparison .
( we note that , although a known force is applied to the origin of our sample , since we expect the unknown local constitutive relations in our system to depend on density and other parameters , the amplitudes can not be directly compared . here
, we wish only to establish the nature of the decay of the non continuum modes . )
we have also introduced the solid lines corresponding to @xmath85 merely as guides to the eye .
these are not fits , although fits to these data for various @xmath86 and @xmath0 demonstrate decay lengths of @xmath0 within 10% . versus the scaled distance from perturbation @xmath87 for systems with various values of @xmath0 as given in the key .
the system radius @xmath34 in all cases .
data within a boundary layer of width @xmath0 from @xmath88 is not shown ( see text for details).,width=321 ] thus , we observe a near field regime in which the decay of the @xmath84 appears to be generically more rapid than @xmath89 followed by the quasi - continuum regime where the exponential decay with decay length @xmath0 is observed .
this demonstrates that the nonaffinity length @xmath0 , indeed , controls the approach toward the expected continuum behavior ( _ i.e. _ , vanishing of the non continuum modes ) .
similar results ( not presented here for reasons of space ) are also found for the other non continuum modes .
specifically , the decay lengths are also found to be @xmath0 within 10% .
of course , the fixed zero - displacement boundary condition at @xmath88 requires all components of the displacement field to vanish there .
the essential distinction between the continuum and non - continuum modes lies in the manner in which their amplitudes decay upon approach to the rigid boundary . the non - continuum modes examined above decay exponentially with a decay length proportional to @xmath0 .
we demonstrate in the next section that the continuum mode amplitudes , in contrast , do not appear to decay in the same way .
in particular , their approach to zero at the rigid boundary is _ not _ controlled by @xmath0 . in this section
we address two related points : ( i ) the fundamental change in the structure of the displacement field as one moves away from the immediate vicinity of the applied force , and ( ii ) the role of finite size effects in our numerical simulation in two - dimensions .
we can not address the former point without confronting the latter one for the following reason .
due to the presence of the rigid boundary , all modes of the displacement field decay to zero at the boundary .
we show in this section , however , that the decay of the non - continuum modes is controlled by internal , mesoscopic length scales in the network , while the decay of the continuum modes is determined by the macroscopic geometry of the system including the presence of the rigid boundary .
since the distinction between the continuum and non - continuum modes depends essentially on the presence of the boundary and specifically on the separation of the system size @xmath28 from the internal length scales controlling the decay of the non - continuum modes ( @xmath90 ) , we consider this distinction along with a more general discussion of finite size effects in our simulation . versus @xmath83 for @xmath36 , @xmath34 and differing @xmath74 as shown in the legend .
, width=321 ] to have a well - defined elastic response in two - dimensions ( meaning that the displacement field vanish at large distances from the applied point force ) one needs to impose a rigid boundary . in order to eliminate the differential influence of the boundary on the various angular modes of the displacement field we chose this boundary to be a circle of radius @xmath28 centered on the point of force application
this rigid boundary forces all components of the displacement field to vanish exactly at @xmath88 .
such a rigid boundary can be expected to introduce a boundary layer near the edges of our system , which we observe and discuss below .
figure [ f : gr0-r = l ] shows the decay of the @xmath91 mode amplitude for systems of size @xmath34 and @xmath75 .
we note that the decay of this _ continuum _ mode is qualitatively distinct from the decay of the non - continuum modes shown in figs .
[ f : gf2_5p2_r1l ] , [ f : gn0_5p2_r1l ] , and [ f : log - lin - gf2 ] .
we do not observe an exponential decay of this continuum mode amplitude .
nevertheless , observing an exponential decay having a long decay length , or , more reasonably , the product of an algebraic function and such a weak exponential decay would be difficult to resolve in this plot . to study these issues further
, we plot in fig.[f : gr0_syssize ] @xmath92 for various system sizes from @xmath93 to @xmath94 .
we observe both an apparent convergence for the larger systems to a common curve when distances are measured relative to the size of the system , as well as a systematic down - turn near the boundary .
the first of these observations suggests that the sample geometry controls the dependence of this continuum mode , as opposed to intrinsic lengths like @xmath0 .
the down - turn in these data within a distance of order @xmath0 ( here , @xmath95 ) of the boundary is to be expected from the observations above concerning the role of the length scale @xmath0 .
one can view this crossover length as the distance along a filament over which a force applied to the filament dissipates / expands into a ( quasi-)continuum stress / displacement field .
thus , the effects of the boundaries on the individual filaments contacting the boundary are expected to propagate at least this distance into the system .
because of this , we have also removed all data within a distance @xmath0 of the boundaries in fig . [
f : log - lin - gf2 ] above , which exhibit a similar down - turn near the boundary . in fig .
[ f : lambda_collapse ] we examine the radial dependence of @xmath92 for three different values of @xmath96 for a large system where @xmath97 . in this case
, we see a coincidence of the data for different @xmath0 , indicating that the long - range behavior of @xmath92 is not controlled by @xmath0 .
although we were unable to obtain sufficient statistics on the non continuum ( and sub dominant ) modes for larger systems , the combination of the qualitatively different behavior from the continuum modes , together with the consistent dependence on the crossover length @xmath0 lead us to conclude that the principal results of the prior section are not strongly influenced by finite - size effects .
thus , @xmath0 controls the disappearance of the non - continuum modes of the system .
versus @xmath83 for @xmath36 , @xmath98 and the system radii @xmath28 given in the key .
the continuum response for the bulk elastic moduli corresponding to these parameter values ( as given by ( [ e : ctm_gr0 ] ) ) is shown as a solid line .
, width=321 ] versus @xmath83 for @xmath99 on log linear axes , with @xmath96 and @xmath74 as given in the key .
each curve has been shifted vertically by a @xmath0dependent scale factor to attempt data collapse .
since @xmath0 varies by over a factor of 4 , the rough collapse is enough to rule out any significant @xmath0dependence on the decay .
error bars are not shown for clarity.,width=321 ] we also see a large discrepancy between the expected continuum solution and the observed displacement of the simulated filament network as shown in fig . [
f : gr0_syssize ] .
this difference can not be simply attributed to finite - size effects ; as can be seen in fig .
[ f : gr0_syssize ] , there is no observable convergence of the larger system data for the continuum modes to the predictions of continuum elasticity for a material having the appropriate lam coefficients .
nevertheless , we clearly observe the more rapid decay of the non - continuum modes on a length - scale controlled by @xmath0 .
one is left with the following puzzle : the tensorial structure and the rotational symmetry of each component of the displacement field approaches the form required by the continuum theory , but the continuum model appears to never quantitatively agree with the numerical data .
we may speculate about the underlying cause of this discrepancy .
it is clear that that the deformation of the material in the immediate vicinity of the point force ( this `` near zone '' extends out to a distance @xmath90 from the point force ) and in a boundary region at the rigid wall is not well described by any continuum theory .
we suggest that under the application of the point force at the origin , the network effectively partitions itself into three different elastic materials . in the near field region
@xmath100 around the point force the deformation response to the point force is quite complex .
similar complexities appear to exist near the rigid wall , where @xmath101 .
these regions apply a complex set of tractions on the circles @xmath102 and @xmath103 bounding the intermediate region that deforms in a manner consistent with some continuum elastic material .
because these tractions are not themselves determined by a simple , continuum model , the resultant deformation of the intermediate region is also not simply derivable from an analysis of the elastic green s function for a continuum .
one might imagine that for very large systems having a consequently larger intermediate region , the complexities of the tractions in the transition zones become less significant .
because of the rigid boundary at @xmath88 and the low - dimensionality of the system , this may not be the case .
the continuum green s function contains logarithmic terms and , due to the rigid boundary , growing polynomial terms as well .
thus , we do not expect convergence to the continuum green s function even for significantly larger systems .
one may ask whether three dimensional systems will show similarly poor convergence to the continuum solutions .
further research here is needed .
we suggest that we do indeed observe the approach of the structure of the elastic green s function to that predicted by continuum theory .
we term the region surrounding the point force where the deformation field has the expected form the quasi - continuum .
based on our numerical data , we do not expect to find a region in which the deformation field agrees quantitatively with the predictions of continuum elasticity using the lam coefficients appropriate to the medium as determined by uniform stress measurements .
this suggests that semiflexible networks admit a highly complex point force response that can not be fully captured by continuum elasticity even in the far - field .
the full implications of this complexity have not been explored .
another instructive measure of the monopole response is the density of elastic energy @xmath104 at a given distance from the point perturbation , which measures gradients of the displacement field and therefore complements the analysis of @xmath105 given above .
in addition , since the simulation contains explicit terms for the transverse and longitudinal filament deformation modes , it is straightforward to measure the partitioning of the energy between these modes .
based on previous work @xcite , under homogeneous shear strain the partitioning of elastic energy between the bending and stretching modes of the filaments is determined entirely by the affine - to - nonaffine crossover .
the ratio of @xmath106 was found to control this energy partitioning at a macroscopic or average level .
it remains to be seen how this partitioning of the elastic energy occurs in vicinity of a point force . the freedom to choose the angle between the point force and the direction of the filament to which that force is applied allows one to determine locally the partitioning of the elastic energy between bending and stretching modes of the filaments .
forces directed along the filament axis generate primarily stretching deformations in the immediate vicinity of the origin ( where the force is applied ) .
forces directed perpendicular to the filament axis , however , locally create a large bending deformation .
as seen in the previous homogenously imposed strain deformation calculations , for any given value of @xmath106 the network responds very differently to the bending or stretching deformations .
thus it is not surprising that the decay of the energy density from the point of force application to the boundary is strongly dependent on whether the force is applied parallel or perpendicular to one of the filaments at the crosslink .
[ f : toted_fixedlam_par ] shows @xmath104 for parallel forces , where @xmath0 is fixed but the filament density @xmath74 varies .
there is an approximate data collapse onto a single curve as shown .
the most notable discrepancy occurs near the boundary , where the low density data remains higher than those for larger @xmath74 values .
this is most likely due to the boundary layer already discussed above for the displacement modes .
( in arbitrary units ) versus distance @xmath78 from an externally applied force that is directed along the filament to which it is applied .
@xmath74 is varied as given in the key , and @xmath34 and @xmath36 in all cases .
, width=321 ] and the data has been collapsed in the small @xmath107 regime . the solid line is proportional to @xmath108 , showing @xmath21 controls the initial decay of elastic energy . in the lower panel
the same energy densities are plotted against distance scaled by @xmath0 .
this shows the energy density decays with that characteristic length scale at longer ranges .
error bars are not shown for reasons of clarity .
, width=321 ] in contrast , @xmath104 for forces perpendicular to a filament exhibits much richer behavior , as shown in fig .
[ f : toted_fixedlam_perp ] , which shows the same data plotted against @xmath107 and @xmath87 . for large distances ,
the data for different @xmath74 appear to differ by only a scale factor .
since there is one arbitrary multiplicative factor for each curve ( as only the magnitude of the applied force is fixed , not its displacement and hence work done on the network ) , these portions of the curves can be made to collapse after scaling , as in the parallel force case , suggesting that beyond some near - field regime , the decay of elastic energy density is once again exponential and governed by @xmath0
see fig .
[ f : toted_fixedlam_perp ] lower panel .
however , in this near - field regime for perpendicular forces there is clearly no possibility of such a collapse .
instead , it appears that the more rapid decay of elastic energy density in this near - field regime is governed by the length @xmath21 , as can be seen in the upper panel of fig .
[ f : toted_fixedlam_perp ] where the ( rescaled ) energy density data are plotted vs. @xmath107 and shown to decay as @xmath109 for distances @xmath110 ( ignoring finite size effects ) .
based on the two different regimes of data collapse shown in fig .
[ f : toted_fixedlam_perp ] , we conclude that there is a near - field region in which the decay of elastic energy is governed by a microscopic length scale the mean distance between cross - links and a longer - range regime in which the spatial decay of elastic energy density is controlled by the mesoscopic length @xmath0 . to better study this cross - over
, we examine the partitioning of elastic energy between bending and stretching modes of the filaments .
recall that in the affine limit , the elastic energy is stored primarily in the stretching and compression of the filaments . in the nonaffine regime , on the other hand , the elastic energy is stored almost entirely in bending modes of the filaments . plotted in fig .
[ f : fraced_fixedlam_perp ] is the proportion of elastic energy due to longitudinal filament deformation for the perpendicular force case .
three data sets are displayed having different values of @xmath111 and @xmath0 see figure caption .
the distance from the point of force application has been scaled by the geometric mean of @xmath111 and @xmath1 leading to the observed coincidence of the crossover between regimes in the data .
close to the point of the application of the force , the displacement response is clearly dominated by bending modes of the filaments .
this is to be expected since the perpendicularly directed force directly injects bending energy into the system at the origin
the point of force application .
the energy partitioning is inconsistent with that which is expected based on previous homogeneous shear measurements . from that work ,
one expects the network to determine the energy partitioning based only on the ratio : @xmath106 .
we observe the fraction of stretching energy to rapidly increase towards the expected value based on @xmath106 @xcite and we may characterize each curve has having a ` knee ' separating the region of varying energy ratio from a nearly constant intermediate regime where the partitioning of elastic energies corresponds well with the previously identified fraction of stretch / compression energy under macroscopic strain . this correspondance is demonstrated in fig .
[ f : fraced_fixedlc_perp ] .
the horizontal lines show the fraction of stretching energy in periodic systems subjected to macroscopic shear , and clearly coincide with the plateau reached beyond the knee . at the largest distances
one notices the vanishing of bending energy as the fraction of stretching energy approaches unity .
this last effect is due to the fact that each filament has a freely rotating bond at the outer , rigid wall .
as this wall can not support torques , the bending energy vanishes in a boundary layer whose width is determined by the mean distance between cross - links .
we return to this point below . , for three systems having differing values of @xmath111 ( see key ) but the same value of @xmath0 . , width=321 ] ( in fact , the exact same geometries ) , @xmath34 but different @xmath0 .
the different values of the stretching energy fraction in the intermediate regime reflect the differing values of the ratio @xmath106 for the networks . for comparison ,
the macroscopic energy fraction under bulk shear is shown as horizontal dashed lines , in the same order as the data points ( _ i.e. _ @xmath38 , 0.191 and 0.412 from top to bottom ) .
, width=321 ] the observed coincidence of the knees of all three curves under the rescaling of distances by @xmath112 suggests that this length sets the scale over which injected bending energy is redistributed into the combination of bending and stretching appropriate for long - length scale deformations .
based on these observations we note that in all networks the strain field acquires the structure of the point force response based on continuum elasticity over a ( typically mesoscopic ) length scale of @xmath0 . when the networks is subjected to large , local bending deformations , however , it readjusts the partitioning of bending to stretching energy over a generally much shorter length scale , @xmath111 .
thus the system is able to repartition the local elastic energy storage to the value appropriate for its @xmath106 ratio over smaller length scales than does the system recover the expected long length scale structure of its continuum elastic response .
we have already shown that in the intermediate field regime the fraction of stretching elastic energy in the system approaches its far - field value as determined by the ratio : @xmath106 .
nevertheless , we in fact find that at the edges of our sample the network energy becomes stored solely in stretching modes regardless of the value of @xmath106 .
as mentioned above we attribute this final redistribution of the elastic energy density between bending and stretching modes to a boundary effect imposed by the freely - rotating nature of coupling of the network filaments to the rigid boundary . to further test
that this final redistribution is indeed a boundary effect , we consider a few larger system sizes as shown in fig .
[ f : fraced_syssize_perp ] . in that figure
the force remains perpendicular to the direction of the filament to which it is applied while the system size is varied from @xmath113 to @xmath114 for a network of fixed @xmath0 .
when the data are plotted against the radial distance from the point force scaled by system size , we find an excellent collapse in the intermediate regime and in the putative boundary layer .
for the smallest system size considered , @xmath34 , we note poorer data collapse in the near field region suggesting that one must study systems that are are least larger than a single filament length to access the bulk behavior of the network with quantitative accuracy . clearly , all three curves taken together are consistent with the notion of an elastic boundary layer that is produced by the freely - rotating boundary at the wall and that extends distance approximately equal to @xmath0 into the sample .
term of the hamiltonian for perpendicular forces , as a function of distance from the origin @xmath78 for system sizes @xmath34 , @xmath115 and @xmath94 . in all cases
@xmath98 and @xmath75 .
, width=321 ] we now consider the mechanical response of the network to localized force dipoles at the origin .
understanding this response function is central to elucidating the effect of molecular motor activity in the cytoskeleton .
many of the general features observed in the response of the system to point force monopoles are also in evidence .
for example in fig .
[ f : dip1 ] we see the rough collapse in the far field of the mode amplitude @xmath116 to the continuum solution for three different networks having the same value of @xmath0 . note that the amplitude has been scaled by @xmath117 since the magnitude of the imposed force dipole for each realization of the network will depend on the distance to the next constraint , _ i.e. _ cross - link , which is @xmath21 . since the displacement field when averaged over network realizations is quadrupolar , being the difference in two dipole , that length must enter squared . the analogous displacement field amplitude , @xmath116 from the continuum solution ( solid line in fig .
[ f : dip1 ] ) was calculated using unit forces so this amplitude is known only up to an overall scale factor ; that scale factor was adjusted to best fit the data .
a similar comparison can be made for the other continuum modes of the displacement field ( see the appendix ) .
[ f : dip2 ] , for example , shows @xmath118 with the same arbitrary prefactor .
the agreement to the continuum theory scaled as discussed above is quite poor . as is also seen in the point - force response of the strain field , higher angular modes , which vanish in the continuum theory such as @xmath119
are significantly non zero in the data .
the amplitude of the non - continuum modes in the dipolar response is even more dramatic than in the point - force response of the network examined earlier . the simulation data for the response of the network to force dipoles can be broadly summarized as follows : the amplitude of the continuum modes of the dipolar displacement field from the naive continuum theory do not agree with simulation data . in light of the discussion regarding such disagreements between the monopole data and the continuum calculation , this is perhaps not surprising .
more interestingly , the observed displacement field has significant amplitudes of higher order angular modes . in short the universality of the response of semiflexible networks to localized point forces
does not appear to extend to their response to localized force dipoles .
we speculate that the principal difference between these two cases stems from the fact that the network s response to the force dipole probes the more detailed microscopic structure of the network in the immediate vicinity of the point of the force dipole application .
the amplitude of the lowest order force multipole communicated from the near field region to the far field where we expect a continuum based theory to apply is not constrained by elementary force balance .
neither are the amplitudes of any higher order force multipoles generated within the near field region .
thus , upon reaching the inner edge of the far field region the force dipole imposed at the origin has generated a highly complex set of tractions on the rest of the material whose structure depends on local details of the connectivity of the network near the origin .
in contrast for the case of the force monopole , the dominant term in those tractions is the fixed total monopolar force acting on the intermediate region .
the higher order force multipoles created in the inner region decay rapidly with distance from the origin leaving one with highly reproducible results for the response of the network to applied forces .
the amplitude of the force dipole communicated from the near field region to the intermediate region , on the other hand , is not similarly constrained .
it appears that one generically generates large amplitude higher order force multipoles in addition to whatever force dipole is communicated to the intermediate region making convergence to a simple dipolar form slow and difficult to observe in our finite samples . for dipole forcing for fixed @xmath36
compared to the continuum solution ( smooth line ) , which has one free fitting parameter , namely the overall magnitude of the dipole forcing .
both data and curve are negative , so the magnitudes have been plotted to allow use of a logarithmic axis .
each data curve has been scaled by @xmath120 to ensure the same mean dipole magnitude ( see text ) .
the system radius was @xmath34 .
, width=321 ] for same systems as in fig .
[ f : dip1 ] .
the only fitting parameter for the continuum solution is the same as used previously for @xmath121 , so there are no remaining free fitting parameters in this plot . much of the data for @xmath122 is negative and hence not visible on these axes . , width=321 ]
lastly , we present new results on homogeneous deformations of semiflexible networks . it has already been shown that the macroscopic elastic moduli of this class of model networks depend in a crucial way on the ratio of @xmath0 to the filament length @xmath1 @xcite . for @xmath123 ,
the deformation is approximately affine , whereas non affine deformation modes dominate when @xmath124 .
previously this was demonstrated only for the shear modulus ; here we can now confirm that the young s modulus @xmath125 behaves in an identical manner .
[ f : y ] shows @xmath125 measured from uniaxial extension of a rectangular cell , scaled by the prediction for an affine strain , plotted against @xmath96 for the range of @xmath74 considered in this paper .
there is a clear data collapse , as for the shear modulus .
furthermore the deviation from the affine prediction is small for @xmath123 , but becomes increasingly pronounced as @xmath96 increases .
this confirms that @xmath96 controls the macroscopic elastic response of these systems .
[ f : nu ] gives the poisson ratio @xmath126 for the same systems as in fig .
[ f : y ] .
it is striking to observe that , within error bars , @xmath126 is consistent with the value @xmath127 , which is the value expected for an affine deformation @xcite .
however , it is apparent from this figure that the measured values are consistent with @xmath127 for _ all _ data points , even those well into the non affine regime .
the mechanism behind this striking robustness currently evades us ( is has nothing to do with incompressibility , which fixes @xmath128 in two dimensions ) . note that @xmath126 at the rigidity percolation @xmath129 ( at which the elastic moduli vanish ) is @xmath130 @xcite ,
which is clearly inconsistent with the data in fig .
[ f : nu ] and confirms our earlier claims that the non affine regime is distinct from the scaling regime of the transition . on log
the affine prediction , which depends only on @xmath74 , can be found in @xcite .
the symbols are larger than the error bars . , width=321 ] for the same systems as in fig .
[ f : y ] .
the solid line corresponds to @xmath127 .
, width=321 ]
in this paper we have presented the results of numerical studies on the response of semiflexible networks to both point forces and to homogeneously imposed strain .
the data presented on the young s modulus taken in combination with previous work on the static shear modulus shows that the mechanical response of the system can be understood in terms of lam coefficients that depend on the ratio of the filament length to the nonaffinity length : @xmath106 .
the poisson ratio of the material , however , appears to be remarkably insensitive to this ratio . we can offer no explanation for this insensitivity at this time .
the data presented on the point force response form a necessary compliment to previous work on the development of a long length scale elastic theory of such materials . based on these data
, it appears that the storage of elastic energy and the structure of the strain field is rather complex in the immediate vicinity of the applied point force .
we characterize these quantities by considering three qualitatively different regimes as a function of radial distance from the applied point force . immediately surrounding the point force in the near - field regime
we find that the partitioning of elastic energy into bending and stretching is determined primarily by the angle between the applied force and the direction of the filament to which that force is applied .
the disorder - averaged strain field is rather complex having higher angular harmonics that predicted by continuum elasticity theory . in the intermediate field regime
farther from the point force the partitioning of elastic energy is determined solely by the ratio @xmath106 as found in the homogeneous shear measurements .
the higher angular harmonics present in the strain field appear to decay exponentially with a decay length proportional to @xmath0 with a constant of proportionality near unity .
these results taken in combination with our previous work suggests that one may think of @xmath0 as setting the minimum length scale for the applicability of continuum modeling quite generally . in the quasi - continuum regime ( @xmath131 )
, we find a strain field consistent with structure of that predicted by continuum model .
the strain itself , however , can not be simply computed from a knowledge of the effective lam coefficients .
we believe the source of this discrepancy is the fact that the intermediate region , being poorly described by the continuum theory , applies a much more complex set of tractions to the material in the far - field where the continuum theory must apply .
if these boundary conditions were known , we suspect that one could in fact calculate the resulting displacement field using the continuum theory .
this belief is supported by the fact that in the far - field regime the displacement field for different networks having the same @xmath0 collapse onto the same curve as one would expect for any effective continuum theory in which the lam coefficients depend on @xmath0 . to summarize these results
, we suggest that there is an emerging description of the mechanics semiflexible networks .
there mechanical behavior over length scales longer than @xmath0 appears to be described by a modified elastic theory in which the effective lam constants depend on the ratio @xmath106 . for point forces and
presumably any force applied over regions having a characteristic length scale that is small compared to @xmath0 , the local response of the network is quite complex and the material appears to be anomalously compliant ; at longer length scales , however , the structure of the deformation field appears to be consistent with the predictions of continuum elasticity .
we believe that it should be possible to construct an elastic continuum theory of these networks that is applicable in both the intermediate and far - field and that is based on a modified gradient expansion of the strain field incorporating explicitly the mesoscopic length @xmath0 .
the behavior of the strain field on scales much smaller than @xmath0 appears to depend on other , more microscopic length scales .
understanding the response of semiflexible networks to localized forces is a necessary for both microrheological investigations of semiflexible networks and for understanding the effect of molecular motors in the cytoskeleton .
clearly , further numerical investigations are required as well as a theoretical examination of the development of elastic continuum models applicable to the intermediate and far - field regimes . moreover ,
additional investigations are required to examine the analogous questions in three dimensional semiflexible networks . while we expect the basic physics outlined above , including the existence of a mesoscopic length @xmath0 , to persist in three dimensions , the expected @xmath132 decay of the elastic green s function should alter the results .
moreover the more rapid decay of the displacement field and energy density in the continuum three dimensional system may further simplify the structure of the analogous green s functions for the semiflexible network .
ajl and fcm acknowledge the hospitality of the isaac newton institute for mathematical sciences where part of this work was done .
dah was jointly funded by the european community marie curie and the japanese society for the promotion of science programs .
ajl was supported in part by nsf - dmr0354113 .
here we derive the displacement field @xmath133 predicted by continuum elasticity for the cases of both monopole and dipole forcing . for the monopole case ,
a point force @xmath134 is applied to the origin of an elastic sheet with lam coefficients @xmath135 and @xmath0 , or equivalently @xmath135 and the poisson ratio @xmath136 . for an isotropic elastic body ,
the stress obeys @xmath137 and the resulting equation for force balance is @xcite @xmath138 solving this in polar coordinates @xmath139 with the boundary condition @xmath140 at a radius @xmath28 from the origin eventually leads to @xmath141 \nonumber\\ & & \left.\mbox{}-\left(r / r\right)^{2 } \left[2\hat{r}_{i}\hat{r}_{j}\hat{f}_{j}-c_{2}\hat{f}_{i}\right ] \right\}\,,\end{aligned}\ ] ] where @xmath142 and @xmath143 . then
the only two non zero modes according to the definition of the @xmath48 given in ( [ e : g_modes ] ) are @xmath144 a naive calculation of the corresponding dipole solution would simply superpose the above monopole solution for two point forces , @xmath145 and @xmath146 with @xmath147 , and then take the limit @xmath148 .
however , this ignores the boundary at radius @xmath28 , which should be kept fixed but is shifted a distance @xmath149 by the above procedure . an exact calculation would require the monopole solution for force applied near to ( but not at ) the center of a circular system , which , since it no longer obeys radial symmetry , is likely to be highly complex . here
we ignore such issues and simply use the above monopole solution , in the expectation that it will closely approximate the exact case except possibly near the boundary .
the displacement field induced by the force dipole is then @xmath150 \nonumber\\ & & \mbox { } + 8\hat{r}_{i}({{\bf\hat}{r}\cdot{\bf\hat}{f}})^{2 } \left[c_{2}(r / r)^{-2}-c_{1}\right ] \nonumber\\ & & \mbox { } + 2\hat{f}_{i}({{\bf\hat}{r}\cdot{\bf\hat}{f } } ) \left[2(c_{1}-1)-2c_{2}(r / r)^{-2}\right . \nonumber\\ & & \mbox{}\hspace{2 cm } \left.-(c_{2}-1)(r / r)^{2}\right ] \bigg\}\,.\end{aligned}\ ] ] as explained in sec .
[ ss : decomposition ] , the dipole solution above applies for a single realization on large length scales , but after averaging over many networks on short lengths ( as in the simulations ) the dipole moment vanishes , leaving a quadrupole displacement field .
the required quadrupole is one consisting of two parallel dipoles of equal and opposite magnitude , aligned along their axes , _
@xmath151 \nonumber\\ & & \mbox { } + 24\hat{r}_{i}({{\bf\hat}{r}\cdot{\bf\hat}{f } } ) \left[c_{2}(r / r)^{-2}-c_{1}\right ] \nonumber\\ & & \mbox { } + 8\hat{f}_{i}({{\bf\hat}{r}\cdot{\bf\hat}{f}})^{2 } \left[1 - 2c_{1}+3c_{2}(r / r)^{-2}\right ] \nonumber\\ & & \mbox { } + 16\hat{r}_{i}({{\bf\hat}{r}\cdot{\bf\hat}{f}})^{3 } \left[2c_{1}-3c_{2}(r / r)^{-2}\right ] \bigg\}\,.\end{aligned}\ ] ] this gives the displacement field in response to known forces of magnitude @xmath46 .
since it is rather the displacement that is controlled , @xmath46 is a free parameter . finally , the non
zero modes are @xmath152
|
previous research on semiflexible polymers including cytoskeletal networks in cells has suggested the existence of distinct regimes of elastic response , in which the strain field is either uniform ( affine ) or non uniform ( non affine ) under external stress .
associated with these regimes , it has been further suggested that a new fundamental length scale emerges , which characterizes the scale for the crossover from non affine to affine deformations .
here , we extend these studies by probing the response to localized forces and force dipoles . we show that the previously identified nonaffinity length [ d.a .
head _
et al . _
pre * 68 * , 061907 ( 2003 ) . ] controls the mesoscopic response to point forces and the crossover to continuum elastic behavior at large distances .
|
in the past decade , there has been a number of studies on dissipative processes associated with the inflaton decay during its evolution .
these studies have shed light into the possible effects of the dissipative processes .
for instance , it was realized that dissipation effectively slows down the rolling of the inflaton scalar field @xmath3 toward the true vacuum .
these processes are capable of supporting the scenario of inflation @xcite .
recently , inspired by several new developments , the problem of inflation with thermal dissipation has attracted many re - investigations .
the first progress is from the study of the non - equilibrium statistics of quantum fields , which has found that , under certain conditions , it seems to be reasonable to introduce a dissipative term ( such as a friction - like term ) into the equation of motion of the scalar field @xmath3 to describe the effect of heat contact between the @xmath3 field and a thermal bath .
these studies shown that the thermal dissipation and fluctuation will most likely appear _ during _ the inflation if the inflaton is coupled to light fields@xcite
. however , to realize sufficient e - folds of inflation with thermal dissipation , this theory needs to introduce tens of thousands of scalar and fermion fields interacting with the inflaton in an _ ad hoc _ manner@xcite .
namely , it is still far from a realistic model .
nevertheless , this study indicates that the condition necessary for the standard " reheating evolution a coupling of inflaton with light fields
is actually also the condition under which the effects of thermal dissipation during inflation should be considered . secondly , in the case of a thermal bath with a temperature higher than the hawking temperature , the thermal fluctuations of the scalar field plays an important and even dominant role in producing the primordial perturbations of the universe .
based on these results , the warm inflation scenario has been proposed . in this model , the inflation epoch can smoothly evolve to a radiation - dominated epoch , without the need of a reheating stage@xcite .
dynamical analysis of systems of inflaton with thermal dissipation@xcite gives further support to this model .
it is found that the warm inflation solution is very common .
a rate of dissipation as small as @xmath4 , @xmath5 being the hubble parameter during inflation , can lead to a smooth exit from inflation to radiation .
warm inflation also provides explanation to the super - hubble suppression .
the standard inflationary cosmology , which is characterized by an isentropic de sitter expansion , predicts that the particle horizon should be much larger than the present - day hubble radius @xmath6 .
however , a spectral analysis of the cobe - dmr 4-year sky maps seems to show a lack of power in the spectrum of the primordial density perturbations on scales equal to or larger than the hubble radius @xmath6@xcite . a possible explanation of this super - hubble suppression is given by hybrid models , where the primordial density perturbations are not purely adiabatic , but mixed with an isocurvature component .
the warm inflation is one of the mechanisms which can naturally produce both adiabatic and isocurvature initial perturbations@xcite . in this paper
, we study the power spectrum of mass density perturbations caused by inflation with thermal dissipation .
one purpose of developing the model of warm inflation is to explain the amplitudes of the initial perturbations .
usually , the amplitude of initial perturbations from quantum fluctuation of the inflaton depends on some unknown parameters of the inflation potential .
however , for the warm inflation model , the amplitude of the initial perturbations is found to be mainly determined by the energy scale of inflation , @xmath0 . if @xmath0 is taken to be about @xmath7 gev , the possible amplitudes of the initial perturbations are found to be in a range consistent with the observations of the temperature fluctuations of the cosmic microwave background ( cmb ) @xcite .
that is , the thermally originated initial perturbations apparently do not directly depend on the details of the inflation potential , but only on some thermodynamical variables , such as the energy scale @xmath0 .
this result is not unexpected , because like many thermodynamical systems , the thermal properties including density fluctuations should be determined by the thermodynamical conditions , regardless of other details .
obviously , it would be interesting to find more
thermodynamical " features which contain only observable quantities and thermodynamical parameters , as these predictions would be more useful for confronting models with observations .
guided by these considerations , we will extend the above - mentioned qualitative estimation of the order of the density perturbations to a quantitative calculation of the power spectrum of the density perturbations .
we show that the power spectrum of the warm inflation does not depend on unknown parameters of the inflaton potential and the dissipation , but only on the energy scale @xmath0 .
the spectrum is found to be of power law , and the index of the power law can be larger or less than 1 .
more interestingly , we find that for a given @xmath0 , the amplitude and the index of the power law are not independent from each other .
in other words , the amplitude of the power spectrum is completely determined by the power index and the number @xmath0 . comparing this result with the observed temperature fluctuations of the cmb
, we find that both amplitude and index of the power spectrum can be fairly well fitted if @xmath1 gev .
this paper is organized as follow : in sec .
ii we discuss the evolution of the radiation component for inflationary models with dissipation prescribed by a field - dependent friction term .
in particular , we scrutinize the physical conditions on which the thermal fluctuations dominate the primordial density perturbations .
section iii carries out the calculations of the power spectrum of the density perturbations of the warm inflations . and finally , in sec .
iv we give the conclusions and discuss further observational tests .
let us consider a flat universe consisting of a scalar inflaton field @xmath3 and a thermal bath .
its dynamics is described by the following equations@xcite .
the equations of the expanding universe are @xmath8,\ ] ] @xmath9,\ ] ] where @xmath10 is the hubble parameter , and @xmath11 the planck mass .
@xmath12 is the effective potential for field @xmath3 , and @xmath13 is the energy density of the thermal bath . actually the scalar field @xmath3 is not uniform due to fluctuations .
therefore , the field @xmath3 in eqs.(2.1 ) and ( 2.2 ) should be considered as an average over the fluctuations .
the equation of motion for scalar field @xmath3 in a de - sitter universe is @xmath14 where the friction term @xmath15 describes the interaction between the @xmath3 field and a heat bath . obviously , for a uniformed field , or averaged @xmath3 , the term @xmath16 of eq .
( 2.3 ) can be ignored .
statistical mechanics of quantum open systems has shown that the interaction of quantum fields with thermal or quantum bath can be described by a general fluctuation - dissipation relation@xcite .
it is probably reasonable to describe the interaction between the inflaton and the heat bath as a decay " of the inflaton @xcite .
these results support the idea of introducing a damping or friction term into the field equation of motion .
in particular , the friction term with the form in eq .
( 2.3 ) , @xmath17 , is a possible approximation for the dissipation of @xmath3 field in a heat bath environment in the near - equilibrium circumstances . in principle
, @xmath18 can be a function of @xmath3 . in the cases of polynomial interactions between @xmath3 field and bath environment
, one may take the polynomial of @xmath3 for @xmath18 , _
i.e. _ , @xmath19 .
the friction coefficient must be positive definite , hence @xmath20 , and the dissipative index of friction @xmath21 should be zero or even integer if @xmath12 is invariant under the transformation @xmath22 .
the equation of the radiation component ( thermal bath ) is given by the first law of thermodynamics as @xmath23 the temperature of the thermal bath can be calculated by @xmath24 , @xmath25 being the effective number of degrees of freedom at temperature @xmath26 .
the warm inflation scenario is generally defined by a characteristic that the thermal fluctuations of the scalar field dominate over the quantum origin of the initial density perturbations . because the thermal and quantum fluctuations of the scalar field are proportional to @xmath26 and @xmath5 respectively , a necessary condition for warm inflation models is the existence of a radiation component with temperature @xmath27 during the inflationary expansion .
( 2.5 ) is also necessary for maintaining the thermal equilibrium of the radiation component . in general ,
the time scale for the relaxation of a radiation bath is shorter for higher temperature .
accordingly , to have a relaxing time of the bath shorter than the expansion of the universe , a temperature higher than @xmath5 is generally needed . as a consequence of eq .
( 2.5 ) , warm inflation scenario requires that the solutions of eqs .
( 2.1 ) - ( 2.4 ) should contain an inflation era , followed by smooth transition to a radiation - dominated era .
dynamical system analysis also confirmed that for a massive scalar field @xmath28 , the warm inflation solution of eqs .
( 2.1 ) - ( 2.4 ) is very common . a smooth exit from inflation to radiation era
can be established even for a dissipation with @xmath18 as small as @xmath29@xcite .
a typical solution of warm inflation will be given in next section . since warm inflation solution
does not rely on a specific potential , we will employ the popular @xmath30 potential commonly used for the new " inflation models .
it is @xmath31 to have slow - roll solutions , the potential should be flat enough , _
i.e. _ , @xmath32 , where @xmath33 . for models based on the potential of eq .
( 2.6 ) , the existence of a thermal component during inflation seems to be inevitable . in order to maintain the @xmath3 field close to its minimum at the onset of the inflation phase transition ,
a thermal force is generically necessary . in other words ,
there is , at least , a weak coupling between @xmath3 field and other fields contributing to the thermal bath . during the slow roll period of inflation ,
the potential energy of the @xmath3 field is fairly constant , and their kinetic energy is small , so that the interaction between the @xmath3 field with the fields of the thermal bath remains about the same as at the beginning . as such , there is no compelling reason to ignore these interactions . strictly speaking
, we should use a finite temperature effective potential @xmath34 .
however , the correction due to finite temperature is negligible .
the leading temperature correction of the potential ( 2.6 ) is @xmath35 . on the other hand ,
as mentioned above , we have @xmath36 for the flatness of the potential .
therefore , @xmath37 , _
i.e. _ , the influence of the finite temperature effective potential can be ignored when @xmath38 .
now , we try to find warm inflation solutions of eqs .
( 2.1 ) - ( 2.4 ) for weak friction @xmath39 . in this case , eqs . ( 2.1 ) -
( 2.3 ) are actually the same as the standard " new inflation model when @xmath40 namely , we have the slow - roll solution as @xmath41 @xmath42 and @xmath43 where the subscript @xmath44 denotes the starting time of the inflation epoch . during the stage of @xmath45
, it is reasonable to neglect the @xmath46 term in eq .
we have then @xmath47 considering @xmath39 , an approximate solution of @xmath3 can immediately be found as @xmath48 where @xmath49 and @xmath50 is the initial value of the scalar field . substituting solution ( 2.12 ) into eq . ( 2.4 ) , we have the general solution of ( 2.4 ) as @xmath51 where @xmath52 $ ] , @xmath53 , and @xmath54 is the initial radiation density . obviously , the term @xmath55 in eq .
( 2.12 ) describes the blowing away of the initial radiation by the inflationary exponential expansion , and the term @xmath56 is due to the generation of radiation by the @xmath3 field decay . according to eq .
( 2.13 ) , the evolution of the radiation has two phases .
phase 1 covers the period during which the @xmath55 term is dominant , and radiation density drops drastically due to the inflationary expansion . the component of radiation evolves into phase 2 when the @xmath56 term becomes dominant , where the radiation density increases due to the friction of the @xmath3 field .
namely , both heating and inflation are simultaneously underway in phase 2 .
therefore , this phase is actually the era of inflation plus reheating .
the transition from phase 1 to phase 2 occurs at time @xmath57 determined by @xmath58 .
we have @xmath59}{(m+2)\alpha^3h}\cdot \frac { am^4}{\gamma_m\phi_i^{m+2 } } \right\},\ ] ] where @xmath60 .
then the radiation density at the rebound time becomes @xmath61a\exp [ ( m+2)\alpha ht_b].\ ] ] from eqs .
( 2.12 ) and ( 2.13 ) , the radiation density in phase 2 is given by @xmath62 since @xmath63 , eq .
( 2.16 ) can be rewritten as @xmath64 on the other hand , from ( 2.12 ) , we have @xmath65 therefore , in the case of weak dissipation @xmath66 , we have @xmath67 this is consistent with the condition of inflation eq . ( 2.7 ) when eq .
( 2.9 ) holds .
( 2.7 ) and ( 2.9 ) indicate that the inflation will come to an end at time @xmath68 when the energy density of the radiation components , or the kinetic energy of @xmath3 field , @xmath69 , become large enough , and comparable to @xmath70 . from eqs .
( 2.17 ) and ( 2.18 ) , @xmath68 is given by @xmath71 in general , at the time when the phase 2 ends , or a radiation - dominated era starts , the potential energy may not be fully exhausted yet . in this case
, a non - zero potential @xmath72 will remain in the radiation - dominated era , and the process of @xmath3 decaying into light particles is still continuing . however , considering @xmath73 , the right hand side of eq .
( 2.17 ) will always be less than 1 when @xmath74 is less than @xmath75 .
this means that , for weak dissipation , phase 2 can not terminate at @xmath76 , or @xmath77 .
therefore , under weak dissipation , phase 2 will end at the time @xmath68 when the potential energy @xmath12 is completely exhausted , _ i.e. _ , @xmath78 this means that no non - zero @xmath72 remains once the inflation exits to a radiation - dominated era , and the heating of @xmath3 decay also ends at @xmath68 . from eq .
( 2.13 ) , one can find the temperature @xmath26 of the radiation in phases 1 ( @xmath79 ) and 2 ( @xmath80 ) as @xmath81 where @xmath82^{1/4}a^{1/4 } \exp \left[\left(\frac{m+2}{4}\right)\alpha ht_b \right].\ ] ] the temperature @xmath83 at the end of phase 2 is @xmath84 where @xmath68 is given by eq .
( 2.21 ) .
since @xmath85 is increasing with @xmath86 in phase 2 , the condition ( 2.5 ) for warm inflation can be satisfied if @xmath87 , or @xmath88 using eqs .
( 2.17 ) and ( 2.21 ) , condition ( 2.25 ) is realized if @xmath89 namely , @xmath18 can be as small as @xmath90 for @xmath91 gev , and @xmath92 gev .
therefore , the radiation solution ( 2.13 ) , or warm inflation , should be taken into account in a very wide range of dissipation @xmath93 this result is about the same as that given by dynamical system analysis@xcite : a tiny friction @xmath18 may lead the inflaton to a smooth exit directly at the end of the inflation era .
a typical solution of the evolution of radiation temperature @xmath85 is demonstrated in fig .
1 , for which parameters are taken to be @xmath94 gev , @xmath95 gev , @xmath96 and @xmath97 = 100 . actually , @xmath25-factor is a function of @xmath26 in general .
however , as can be seen below , the unknown function @xmath98 has only a slight effect on the problems under investigation .
figure 1 shows that the rebound temperature @xmath99 can be less than @xmath5 . in this case
, the evolution of @xmath85 in phase 2 can be divided into two sectors : @xmath100 for @xmath101 , and @xmath102 for @xmath103 , where @xmath104 is defined by @xmath105 .
we should not consider the solution of radiation to be physical if @xmath100 since it is impossible to maintain a thermalized heat bath with the radiation temperature less than the hawking temperature @xmath5 of an expanding universe .
nevertheless , the solution ( 2.13 ) should be available if @xmath103 .
therefore , one can only consider the period of @xmath106 as the epoch of the warm inflation .
figure 1 also plots the hubble parameter @xmath107 .
the evolution of @xmath107 is about the same as in the standard new inflation model , _
i.e. _ , @xmath108 in both phases 1 and 2 . in fig .
1 , it is evident that the inflation smoothly exits to a radiation era at @xmath68 .
the hubble parameter @xmath107 also evolves from the inflation @xmath109 constant to a radiation regime @xmath110 .
the duration of the warm inflation is represented by ( @xmath111 ) then .
the number of @xmath112-folding growth of the comoving scale factor @xmath113 during the warm inflation is given by @xmath114 one can also formally calculate the number of @xmath112-folds of the growth in phase 2 as @xmath115 and the number of @xmath112-folds of the total growth as @xmath116 it can be found from eqs . ( 2.28 ) - ( 2.30 ) that both @xmath117 and @xmath118 depend on the initial value of the field @xmath50 via @xmath99 , but @xmath119 does not .
the behavior of @xmath26 at the period @xmath103 is completely determined by the competition between the diluting and producing radiation at @xmath80 .
initial information about the radiation has been washed out by the inflationary expansion .
hence , the initial @xmath50 will not lead to uncertainty in our analysis if we are only concerned the problems of warm evolution at the period @xmath106 .
the fluctuations of @xmath3 field can be calculated by the similar way as stochastic inflations@xcite .
recall that the coarse - grained scalar field @xmath3 is actually determined from the decomposition between background and high frequency modes , i.e. @xmath120 where @xmath121 is the scalar field satisfying @xmath122 @xmath123 in eq.(3.1 ) contains all high frequency modes and gives rise to the thermal fluctuations .
since the mass of the field can be ignored for the high frequency modes , we have @xmath124\ ] ] where @xmath125 is comoving wave vector , and modes @xmath126 is given by @xmath127 e^{-ik\tau},\ ] ] and @xmath128 is the conformal time .
eq.(3.3 ) is appropriate in the sense that the self - coupling of the @xmath3 field is negligible .
considering the high frequency modes are mainly determined by the heat bath , this approximation is reasonable .
the window function @xmath129 is properly chosen to filter out the modes at scales larger than the horizon size @xmath130 , _
i.e. _ , @xmath131 , where @xmath132 actually depends on the details of the cut - off function , which may not be step - function - like .
for instance , considering causality , the cut - off function can be soft , and the longest wavelength of fluctuations can be a few times of the size of horizon@xcite ] is the lower limit to the wavenumber of thermal fluctuations . from eqs.(3.1 ) and ( 3.3 ) , with the slow - roll condition , eq.(3.2 ) renders @xmath133 and @xmath134 eq.(3.5 ) can be rewritten as @xmath135}{\delta \phi } + \eta({\bf x } , t)\ ] ] where @xmath136=\int d^3{\bf x } \left [ \frac{1}{2 } ( e^{-ht}\nabla\bar{\phi})^2 + v(\bar{\phi } ) \right]\ ] ] eq .
( 3.7 ) is , in fact , the rate equation of the order parameter @xmath3 of a system with free energy @xmath137 $ ] .
it describes the approach to equilibrium for the system during phase transition . using the expression of free energy ( 3.8 ) , the slow - roll solution ( 2.8 )
can be rewritten as @xmath138}{d \phi}.\ ] ] hence , in the case of weak dissipation ( @xmath39 ) , eq .
( 3.7 ) is essentially the same as the slow - roll solution ( 2.8 ) or eq .
( 3.9 ) but with fluctuations @xmath139 .
the existence of the noise field ensures that the dynamical system properly approaches the global minimum of the inflaton potential @xmath12 .
strictly speaking , both the dissipation @xmath18 and fluctuations @xmath139 are consequences derived from @xmath140 .
they should be considered together .
however , it seems to be reasonable to calculate the fluctuations alone if the dissipation is weak .
unlike ( 3.2 ) , the langevin equation ( 3.7 ) is of first order ( @xmath141 ) due to the slow - roll condition .
generally , thermal fluctuations will cause both growing and decaying modes@xcite .
therefore , the slow - roll condition simplifies the problem from two types of fluctuation modes to one , _ i.e. _ , we can directly calculate the total fluctuation as the superposition of various fluctuations .
it has been shown @xcite that during the eras of dissipations , the growth of the structures in the universe is substantially the same as surface roughening due to stochastic noise .
the evolution of the noise - induced surface roughening is described by the so - called kpz - equation @xcite .
eqs.(3.5 ) or ( 3.7 ) , which includes terms of non - linear drift plus stochastic fluctuations , is a typical kpz - like equation . from eq.(3.6 ) , the two - point correlation function of @xmath142 can be found as @xmath143 \frac{\sin(k_{h}|{\bf x}-{\bf x'}|)}{k_{h}|{\bf x}-{\bf x'}| } \delta(t - t'),\ ] ] where @xmath144 $ ] is the bose factor at temperature @xmath26 .
therefore , when @xmath145 , we have @xmath146 this result can also be directly obtained via the fluctuation - dissipation theorem@xcite . in order to accord with the dissipation terms of eq .
( 3.7 ) , the fluctuation - dissipation theorem requires the ensemble average of @xmath139 to be given by @xmath147 and @xmath148 the variance @xmath149 is determined by @xmath150 where @xmath151 is the volume with hubble radius @xmath130 . in the case of weak dissipation , we then recover the same result as in eq.(3.11 ) , @xmath152 when @xmath153 , we obtain @xmath154 which agrees exactly the result derived from quantum fluctuations of @xmath3-field@xcite
. therefore , the quantum fluctuations of inflationary @xmath3 field are equivalent to the thermal noises stimulated by a thermal bath with the hawking temperature @xmath5 . eqs .
( 3.15 ) and ( 3.16 ) show that the condition ( 2.5 ) is necessary and sufficient for a warm inflation . for long - wavelength modes ,
the @xmath155 term is not negligible .
it may lead to a suppression of correlations on scales larger than @xmath156 .
however , before the inflaton actually rolls down to the global minimum , we have @xmath157 .
the so - called abnormal dissipation of density perturbations @xcite may produce more longer correlation time than @xmath5 .
therefore in phase 2 , _
i.e. _ , the warm inflation phase @xmath158 , the long - wavelength suppression will not substantially change the scenario presented above .
the fluctuations @xmath159 of the @xmath3 field can be found from linearizing eq .
if we only consider the fluctuations @xmath159 crossing outside the horizon , _
i.e. _ , with wavelength @xmath160 , the equation of @xmath159 is @xmath161 for the slow - roll evolution , we have @xmath162 @xcite .
one can ignore the @xmath163 term on the right hand side of eq .
accordingly , the correlation function of the fluctuations is @xmath164 hence @xmath165 thus , in the period @xmath166 the density perturbations on large scales are produced by the thermal fluctuations that leave the horizon with a gaussian - distributed amplitude having a root - mean - square dispersion given by eq .
( 3.19 ) .
principally , the problem of horizon crossing of thermal fluctuations given by eq .
( 3.7 ) is different from the case of quantum fluctuations , because the equations of @xmath5 and @xmath167 , ( 2.1 ) and ( 2.2 ) contain terms in @xmath168 .
however , these terms are insignificant for weak dissipation [ eq .
( 2.19 ) ] in phase 2 .
thus eqs.(2.1 ) and ( 2.2 ) depend only nominally on the evolution of @xmath168 . accordingly , for weak dissipation , the behavior of thermal fluctuations at horizon crossing can be treated by the same way as the evolutions of quantum fluctuations in stochastic inflation . in that theory ,
quantum fluctuations of inflaton are assumed to become classical upon horizon crossing and act as stochastic forces .
obviously , this assumption is not necessary for thermal fluctuations .
moreover , we will show that in phase 2 the thermal stochastic force @xmath169 is contingent upon the comoving scale of perturbations by a power law [ eqs .
( 2.21 ) and ( 3.21 ) ] , and therefore the power spectrum of the thermal fluctuations obeys the power law .
this make it more easier to estimate the constraint quantity in the super - horizon regime . accordingly , the density perturbations at the horizon re - entry epoch are characterized by@xcite @xmath170 all quantities in the right - hand side of eq .
( 3.20 ) are calculated at the time when the relevant perturbations cut across the horizon at the inflationary epoch .
using the solutions of @xmath3 and @xmath168 of warm inflation ( 2.12 ) and ( 2.13 ) , eq . ( 3.20 ) gives @xmath171 where the dimensionless parameter @xmath172 , and @xmath26 is the temperature at the time when the considered perturbations @xmath173 crossing out of the horizon @xmath174 .
( 3.21 ) shows that the density perturbations are insensitive to the @xmath25-factor .
since inflation is immediately followed by the radiation dominated epoch , the comoving scale of a perturbation with crossing over ( the hubble radius ) at time @xmath86 is given by @xmath175 where @xmath176 and @xmath177 are the present cmb temperature and hubble constant respectively .
( 3.22 ) shows that the smaller @xmath86 is , the smaller @xmath125 will be .
this is the so - called first out - last in " of the evolution of density perturbations produced by the inflation .
using eqs .
( 2.22 ) and ( 3.22 ) , the perturbations ( 3.21 ) can be rewritten as @xmath178 where @xmath179 is the wavenumber of perturbations crossing out of horizon at @xmath104 .
it is @xmath180 therefore , the primordial density perturbations produced during warm inflation are of power law with an index @xmath181 .
we may also express the power spectrum of the density perturbations at a given time @xmath86 .
it is @xmath182 where the spectral index @xmath183 is @xmath184 clearly , for @xmath185 , the warm inflation model generates a flat power spectrum @xmath186 , yet the power spectrums will be tilted for @xmath187 . the dissipation models @xmath188 may not be realistic for higher @xmath21 , but we will treat @xmath21 like a free parameter in order to show that the results we concerned actually are not very sensitive to these parameters .
the warm inflation scenario requires that all perturbations on comoving scales equal to or less than the present hubble radius originate in the period of warm inflation .
hence , the longest wavelength of the perturbations ( 3.24 ) , _ i.e. _ , @xmath189 , should be larger than the present hubble radius @xmath190 .
we have then @xmath191 where we have used @xmath192 , as @xmath193 .
( 2.28 ) , the condition ( 3.27 ) gives an upper bound to @xmath194 for a given @xmath21 as @xmath195 thus , the possible area of the index @xmath183 can be found from eq .
( 3.27 ) as @xmath196 therefore , the power spectrum is positive - titled ( _ i.e. _ , @xmath197 ) if @xmath198 , and negative - titled ( @xmath199 ) if @xmath200 .
figure 2 plots the allowed area of @xmath183 as a function of the inflation mass scale @xmath0 . apparently , for @xmath201 gev , the tilt @xmath202 should not be larger than about 0.15 regardless of the values of @xmath21 from 2 to 12 . to calculate the amplitude of the perturbations we rewrite spectrum ( 3.25 ) into @xmath203 where @xmath204 .
@xmath56 is the spectrum amplitude normalized on scale @xmath205 , corresponding to the scale on which the perturbations re - enter the hubble radius @xmath206 at present time . from eqs .
( 3.21 ) , and ( 3.23 ) , we have @xmath207 applying eq .
( 2.21 ) , the radiation temperature at the moment of horizon - crossing , @xmath86 , can be expressed as @xmath208 $ ] . with the help of eq .
( 2.28 ) , we obtain @xmath209n\right\}.\ ] ] on the other hand , using eqs .
( 2.20 ) , ( 2.23 ) and ( 2.28 ) , one has @xmath210 substituting eqs .
( 3.32 ) and ( 3.33 ) into eq . ( 3.31 ) , we have finally @xmath211n\right\}.\ ] ] eq .
( 3.34 ) shows that the amplitude @xmath56 does not contain the unknown @xmath97-factor .
moreover , @xmath194 can be expressed by @xmath183 and @xmath21 through eq .
( 3.26 ) , and @xmath119 can be expressed by @xmath194 and @xmath0 via eq .
( 2.28 ) . therefore
, the amplitude of the initial density perturbations , @xmath56 , is only a function of @xmath0 , @xmath183 , and @xmath21 .
figures 3 and 4 plot the relations between the amplitude @xmath56 and index @xmath183 for various parameters @xmath0 and @xmath21 . in the case of @xmath212 , @xmath186 , the relation of @xmath56 and @xmath194 is plotted in fig .
it can be seen from figs . 3 , 4 and 5 that for either @xmath213 or @xmath200 , the amplitude @xmath56 is significantly dependent on @xmath0 , but not so sensitive to @xmath21 .
namely , the testable @xmath56-@xmath183 relationship is mainly determined by a thermodynamical variable , the energy scale @xmath0 .
this is a thermodynamical " feature .
the relationship between @xmath56 and @xmath119 plotted in figs .
6 and 7 also show this kind of thermodynamical " feature : the @xmath56-@xmath119 relation depends mainly on @xmath0 . for comparison ,
the observed results of @xmath56 and @xmath183 derived from the 4-year cobe - dmr data ( quadrupole moment @xmath214 and @xmath215@xcite ) are plotted in figs . 3 , 4 and 5 .
the observationally allowed @xmath56-@xmath183 range is generally in a good agreement with the predicted @xmath56-@xmath183 curve if @xmath216 gev , regardless the parameter @xmath21 .
figures 3 and 4 also indicate that if the tilt of spectrum @xmath202 is larger than 0.1 , the parameter area of @xmath217 gev will be ruled out .
therefore , the warm inflation seems to fairly well reconcile the initial perturbations with the energy scale of the inflation .
assuming that the inflaton @xmath3-field undergoes a dissipative process with @xmath218 , we have studied the power spectrum of the mass density perturbations . in this analysis ,
we have employed the popular @xmath30 potential .
however , only one parameter , the mass scale of the inflation @xmath0 , is found to be important in predicting the observable features of power spectrum , _ i.e. _ , the amplitude @xmath56 and index @xmath183 .
actually , the warm inflation scenario is based on two thermodynamical requirements : ( a ) the existence of a thermalized heat bath during inflation , and ( b ) that the initial fluctuations are given by the fluctuation - dissipation theorem .
therefore , we believe that the thermodynamical " features @xmath56 and @xmath183 depend only on @xmath0 would be generic for the warm inflation .
this feature is useful for model testing .
hence , the warm inflation can be employed as an effective working model when more precise data about the observable quantities @xmath56 , @xmath183 etc .
become available .
the current observed data of @xmath56 and @xmath183 from cmb are consistent with the warm inflation scenario if the mass scale @xmath0 of the inflation is in the range of @xmath219 gev .
starobinsky , in _ fundamental interactions _
( mgpi press , moscow , 1984 ) , p. 55 ; s. j. rey , nucl . phys . * b284 * , 706 ( 1987 ) ; m. saaski , y. nambu and k. nakao , nucl .
b308 * , 868 , ( 1988 ) ; a. hosaya , m. morikawa , and k. nakayama , int . j. mod . phys . *
a4 * , 2613 ( 1989 ) ; d.s .
salopek and j.r .
bond , phys . rev . * 43d * , 1005 ( 1991 ) and references therein .
|
we study the power spectrum of the mass density perturbations in an inflation scenario that includes thermal dissipation .
we show that the condition on which the thermal fluctuations dominate the primordial density perturbations can easily be realized even for weak dissipation , _
i.e. _ , the rate of dissipation is less than the hubble expansion .
we find that our spectrum of primordial density perturbations follows a power law behavior , and exhibits a `` thermodynamical '' feature the amplitude and power index of the spectrum depend mainly on the thermodynamical variable @xmath0 , the inflation energy scale . comparing this result with the observed temperature fluctuations of the cosmic microwave background , we find that both amplitude and index of the power spectrum can be fairly well fitted if @xmath1 gev .
# 1@xmath2#1
|
one of the fundamental parameters characterising any stellar population is its mass function , and in particular its _ initial _ mass function ( imf ) , defined simply as the number of stars per unit mass that are actually formed in any system . determining the form of the imf is vital to our understanding of star formation in our own and other star - forming galaxies , since it allows us to model the chemical and physical evolution of these systems . despite ongoing debate ( eg .
leitherer 1998 ) it is not clear whether the imf varies widely between galaxies or even within different star formation regions of our own galaxy . since starburst galaxies are generally dusty , ir indicators of the imf are preferable since they provide considerable improvement in sensitivity since the extinction is lower than in the optical , and perhaps as important , any error in the derived extinction value has a correspondingly smaller effect on any observed line ratio when the lines are close in wavelength .
indeed any method that relies on absolute flux values or widely separated line pairs is likely to lead to significant errors in the derived results when the extinction is large .
there are limited methods for directly measuring the current imf from stellar features at these wavelengths , so indirect methods are required .
most work has relied on inferring the nature of the stars present by comparing the ionisation structure of the enveloping nebular gas with the predictions of detailed photoionisation models .
there has been extensive use of strong mid and far - ir forbidden line data in modelling hii regions in this fashion .
however , such data is difficult to obtain for a large sample of galaxies , the atomic data is typically less certain for forbidden transitions and extinction corrections are complicated by the presence of the silicate absorption feature at 9.7@xmath1 m and our poor understanding of the extinction law at longer wavelengths .
consequently there has not been widespread application of that method .
given the problems inherent with the longer wavelength forbidden line data , it was natural that attention should turn instead to the near infrared .
the brightest lines in general below 2.5@xmath1 m are permitted hi and hei lines . in principle
, the comparison of hei and hi lines tests the relative volumes occupied by he@xmath5 and he@xmath6 relative to h@xmath6 , and hence places constraints on the form of the emergent uv radiation field shortwards of the hi lyman limit given the higher ionisation potential of helium .
this is certainly true for any lines that arise solely from recombination .
the brightest hei line longward of 1.1@xmath1 m is the 2.058@xmath1 m hei 2@xmath0p2@xmath0s transition .
this has led to considerable attention being given to the use of the ratio of hei 2@xmath0p2@xmath0s with hi br@xmath3 at 2.166@xmath1 m ( eg . doyon , puxley & joseph 1992 and references therein ) .
however , the 2.058@xmath1 m hei 2@xmath0p2@xmath0s transition is determined by the population in the 2@xmath0p state .
this is largely driven by resonance scattering of the 584 hei 2@xmath0p1@xmath0s transition rather than a simple recombination cascade .
these resonance photons can be absorbed by dust or hydrogen reducing the population in the 2@xmath0p state .
they also depend implicitly on the neutral helium fraction present since they are a resonance from the ground state .
in addition , microscopic velocity structure can lead to a sufficient shift in the wavelength emitted so that the photon can not be reabsorbed by helium at all .
initially it was thought that these secondary effects were relatively small and easily calculable .
however , shields ( 1993 ) showed from a full photoionisation treatment that the strength of the 2@xmath0p2@xmath0s line was much more dependent than previously thought on the neutral helium fraction , and indeed predicted that as the stellar effective temperature , t@xmath7 , increases much beyond 40000k that this line actually gets weaker as most helium in the hii region is turned into he@xmath6 .
more accurate atomic data and photoionisation models have confirmed this general picture ( ferland 1999 ) .
observational confirmation of this prediction was evident from the extragalactic work of lumsden , puxley & doherty ( 1994 ) and doherty et al .
( 1995 ) , and the study of planetary nebulae by depoy & shields ( 1994 ) . however , the former papers deal with complex clusters of ob stars for which we have no accurate knowledge of the effective temperature .
the latter used slitless spectroscopy of extended planetary nebulae , resulting in very low resolution spectra in which only the brightest lines were visible .
therefore , although we know the trend in the observed data appears to be in accord with the photoionisation models , there have been no attempts to really constrain the model results in any detail . crucially , it was still unclear whether or not the hei 2@xmath0p2@xmath0s to hi br@xmath3 ratio can be used as a tracer of the high mass end of the imf .
we therefore decided to carry out observations of objects whose stellar temperature could be defined more accurately than is the case for typical obscured compact hii regions .
we observed a sample of 23 planetary nebulae ( pn ) to derive as full a set as possible of the observable line parameters for both the hei 2@xmath0p2@xmath0s line and hi br@xmath3 .
pn are ideal for this test since they are predominantly excited by only a single central star ( unlike hii regions ) , and a sample covering a wider range of effective temperature is much easier to construct .
our sample of pn was chosen ( i ) to be compact on the sky , ( ii ) to have good quality line measurements in the eso planetary nebula catalogue ( acker et al.1992 ) and ( iii ) to have a spread in ra to make observation easier . where possible we chose those objects with predicted stoy temperatures less than 60000k ( kaler & jacoby 1991 ) in preference to hotter objects , but the final sample contains a selection of pn with predicted temperatures well above this value as well .
the sample is therefore largely heterogeneous , since our selection criteria do not really distinguish , for example , between nearby low temperature and luminosity pn , and more distant high excitation and luminosity pn .
in addition , we did not impose any morphological selection on the sample , so there are objects ranging from spherical to bipolar present . for most of our sources
there are no published peimbert types , and even those with published types are often contradictory . it is likely therefore that at least some of our objects are type i pn , with the consequence that they will probably have enhanced helium abundances ( eg peimbert & torres - peimbert 1983 ) . since we can not distinguish these pn accurately however , we do not split our sample into peimbert types .
this is not a particular problem for this project however , since all we require is a set of pn with a reasonable spread in effective temperature , which are bright enough that we can derive line ratio measurements easily .
we can test for abundance effects by examining the behaviour of the hei 2@xmath0p2@xmath0s line with other helium recombination lines .
the only real problem with the inclusion of pn which are not spherical is that the photoionisation modelling reported in section 5 will not be an ideal match .
we do not expect that this will significantly effect the conclusions we reach however .
in addition to our own near infrared spectra we also sought high quality optical data in order to provide more accurate constraints on the photoionisation models .
for the most part this comes from a search of the literature , and from the electronic catalogue of kaler , shaw & browning ( 1997 ) . for some of the pn in our sample we obtained low resolution 43007400 ccd spectra with the red arm of isis on the william herschel telescope on the night of 30 july 1996 in poor weather .
this allowed us to check the reliability of the published data in the literature as well . in particular ,
we have chosen not to rely on the fluxes from the acker et al.catalogue in our work beyond the aspect of sample selection wherever possible .
the bright lines are often saturated , and the faint hei lines we require in our analysis are often undetected making the catalogue largely unsuitable if there is an alternative .
we also excluded a large amount of the oldest ( pre-1980 ) data since there were clear indications that it disagreed completely with our own ccd data where there were objects in common .
where necessary we placed all data onto a common extinction corrected intensity scale using the extinction law given in seaton ( 1979 ) . since all of the objects ( with the exception of crl 618 ) have low moderate optical extinction , we also made the implicit assumption that @xmath8 in our analyses .
since there are no direct means of determining the extinction from the near infrared data alone , we felt this was the safest option .
assuming @xmath9leads to errors in the observed hei 2@xmath0p2@xmath0s to hi br@xmath3 ratio of @xmath10 or @xmath11 , which are essentially negligible for this sample .
the complete sample is listed in table 1 .
we give the optical references we have used there .
also given there are the averaged fluxes of [ oiii ] 5007 , hei 6678 and heii 4686 relative to h@xmath12 taken from those references .
we attribute 10% errors to the relatively weak 6678line , and 5% errors to the others , to account for potential differences in optical and infrared beamsizes when making comparisons of the actual lines .
these errors are not included in the analysis of the optical data on its own , since there is no aperture effect to correct for . from these data
we have calculated the stoy temperatures ( from the [ oiii]/h@xmath12 ratio and equation 1 of kaler & jacoby 1991 ) , electron densities ( from the [ sii ] 6717/6731 line ratio ) and electron temperature ( from either the [ oiii ] 4363/5007 ratio or the [ nii ] 5755/6548 ratio where these exist ) .
these are also given in the table .
although the density derived from the [ sii ] doublet is potentially unrepresentative of the density in the he@xmath6 region of the nebula , it is the only indicator that is available for the majority of our sample , and we therefore adopt it throughout . unfortunately , as is clear from table 1 , we are lacking in pn with stoy temperatures between 40000 and 60000k .
this was not clear at the time of observation since the predicted values from the eso planetary nebula catalogue ( acker et al . )
did not show this effect .
it is only with the adoption of higher quality , and more strictly global , flux values from the literature that it became obvious .
we have included two objects whose identification as pn is uncertain , m 1 - 78 and crl 618 .
the first of these is sometimes listed as a possible hii region , largely on grounds of excitation and extinction .
however , there are other confirmed pn with similar characteristics so we have included it in our sample anyway .
crl 618 is more strictly a proto - planetary nebula , since the central source is still obscured , and there is a clear bipolar outflow . since it will eventually become a pn however , we include it in our sample .
it should be noted that the optical data for this object is generally from the bipolar lobes , so may not reflect the actual excitation of the core .
we obtained separate low resolution k band spectra ( @xmath13 ) spanning @xmath141.92.5@xmath1 m , and high resolution echelle spectra around the 2.16@xmath1 m hi br@xmath3 and 2.058@xmath1 m hei 2@xmath0p2@xmath0s lines ( @xmath15 ) using cgs4 , the facility infrared spectrograph on the united kingdom infrared telescope .
the data were mainly obtained on the nights of 8 and 9 january 1997 , but additional data were also obtained on 15 july 1995 , 27 and 28 september 1995 , and 4 , 5 and 6 july 1997 . in all cases
we used a two pixel wide slit for the grating spectra , corresponding to 3 arcseconds on the sky .
we used a one pixel wide slit for the echelle data to preserve the highest possible velocity resolution .
proper sampling of the spectrum is obtained by stepping the array a fraction of a pixel between ` sub - exposures ' which are later combined in software to provide the final spectrum .
anamorphic distortion within the spectrograph makes the width of the slit with the echelle grating close to 1 arcsecond .
the separate echelle data of the two lines were obtained primarily for two reasons : first , to derive velocity profiles for the pn to test for the effects of turbulence on the hei 2@xmath0p2@xmath0s to hi br@xmath3 ratio as expected from theory ; second , as a means of correcting the low resolution fluxes obtained around 2.058@xmath1 m where there is a sequence of deep narrow atmospheric absorption lines due to co@xmath16 .
however , the echelle data alone can not be used to derive reliable estimates of the ratio of the hi br@xmath3 and hei 2@xmath0p2@xmath0s lines because we used the narrowest possible slit to maximize the spectral resolution .
this means that any small inaccuracy in positioning the telescope can have a significant impact on the derived line ratio given the compact nature of the pn .
it is clearly preferable to derive ratios from lines observed in the same spectrum , so wherever possible we use the low resolution data for this . the grating spectra were observed in a standard manner , nodding the telescope to keep the source always on the spectrograph slit .
the beam separation was between 20 and 30 arcseconds for all data .
the data were reduced by dividing by an internal flat field , and coadding all the nod pairs .
the data obtained during january 1997 suffered from a problem with the slit mechanism which was tilted markedly relative to the axes of the array .
these data were straightened to make the wavelength perpendicular to the columns of the array .
the spectra were then extracted , and the negative beam subtracted from the initial positive one .
this procedure ensures that residual sky contamination is minimal .
the low resolution data were wavelength calibrated in the standard fashion using an argon arc lamp .
flux calibration of the low resolution data was obtained using observations of a set of a - type stars from the bright star catalogue .
we adopted @xmath17 corrections from johnson ( 1966 ) and used the @xmath18 magnitudes from the bright star catalogue itself .
this technique is reliable in providing approximate absolute fluxes ( accurate to @xmath1430% ) and very good ( @xmath19% ) relative fluxes within each spectra which are sufficient for our requirements .
the br@xmath3 absorption feature present in the a - type standards was interpolated over using a lorenzian line profile .
the echelle data were largely obtained in a similar manner , with the exception of the 1997 data .
the tilted slit in january 1997 meant that more than one echelle order was present in the final image , so that only the central 20 rows were actually useful for our observations .
therefore we adopted the same strategy as used for mapping extended hii regions described in lumsden & hoare ( 1996 ) , and acquired only one ` sky ' frame for each set of up to 5 ` object ' frames .
given the lack of strong sky lines near the emission lines from the pn this procedure provides a reasonable sky subtracted data - set .
we used the same technique in july 1997 for simplicity .
for all echelle data there is also a residual distortion in the wavelength calibration due to the fact that the projection of the slit onto the array is actually curved .
we used oh night sky lines to ` straighten ' the spectra for this and the more extreme distortion caused by the tilted slit .
we also used the oh lines to provide wavelength calibration for the br@xmath3 data .
the hei data were calibrated using an internal argon arc lamp .
we also acquired spectra of standard stars for the echelle to map the atmospheric absorption features .
we used a type standards for the 2.058@xmath1 m data and k type for the 2.166@xmath1 m data , since both are relatively featureless in these wavelength ranges at high resolution . in practice
there are only weak atmospheric absorption features around 2.166@xmath1 m so there is little problem with this data .
however , around 2.058@xmath1 m the wavelength region is strongly attenuated by a sequence of deep narrow ( almost unresolved ) atmospheric absorption features due to co@xmath16 in the upper atmosphere .
although the echelle data can be corrected for this absorption using the standard star observations it reveals a potential problem with the accuracy of the flux calibration in this narrow wavelength range for the low resolution data .
the basic argument is simple : given a narrow emission line as seen in these pn , the actual data as observed through the telescope can range from almost unaffected by the atmospheric absorption to almost completely absorbed .
however , the spectra of the standard stars as measured at low resolution average over these absorption features , so that the correction they give to the low resolution data is wrong in most cases .
we used the basic procedures outlined by doherty et al .
( 1994 ) to correct the observed low resolution hei 2@xmath0p2@xmath0s fluxes for this atmospheric attenuation .
however , rather than using a model of the atmospheric absorption as they did we used the actual observed echelle standard star spectra .
first we correct the echelle spectra for changes in the observed radial velocity due to the time of observation .
the correction is then derived by smoothing the object and standard echelle spectra to the resolution of the grating data .
the smoothed standard star spectrum is used to correct the smoothed pn echelle spectrum , and the flux of the hei line measured .
the original pn echelle spectrum is corrected by the original standard star spectrum as well , and the flux measured of the full resolution data .
the ratio of the measured fluxes from the full resolution data with the smoothed data then give the required correction factor .
the actual values are given in table 2 .
it is worth noting here that no such corrections were applied to the depoy and shields ( 1994 ) data , potentially leading to errors in their results of up to 30% from this factor alone .
the spectra acquired with the echelle grating are shown in figures 1 and 2 .
the full set of low resolution spectra will be published separately in a paper that discusses the other spectral features present . however , a representative sample showing the trend with effective temperature are shown in figure 3 . in all cases
we have summed up the data over the whole object in forming these spectra .
the observed hi br@xmath3 fluxes are listed in table 2 , along with the ratios of the strongest helium lines with br@xmath3 .
the tabulated values of hei 2@xmath0p2@xmath0s against hi br@xmath3 are all corrected for atmospheric absorption using the method given in section 3 .
all ratios are taken from the low resolution data with the exception of the hei 7@xmath20g4@xmath20f transition at 2.16475@xmath1 m , which was measured from the echelle spectra .
this is the strongest of the hei satellite lines to br@xmath3 , and is visible in most of our echelle spectra .
the ratio given for that line is against the observed br@xmath3 flux from our echelle spectra .
this line can also be compared with the other hei lines , since they would all arise from the same region of the pn .
it should not however be compared with the heii line strengths , since they will come from a different region in the pn , and the possible differences in slit position and size again may come into play .
also visible in some of the br@xmath3 echelle spectra are the other hei transitions of 7@xmath21f4@xmath21d at 2.16229@xmath1 m and 7@xmath22f4@xmath22d at 2.16137@xmath1 m , and the heii 148 transition at 2.1653@xmath1 m .
it should be noted that the low resolution br@xmath3 data have not been corrected for any underlying heii emission ( the bright satellite hei lines are sufficiently well resolved not to contribute to the measured fluxes ) .
examination of the echelle spectra shows that these lines are insignificant for most of the sample , and even for the higher excitation pn contribute at most 5% of the flux .
the one exception is hu 1 - 2 , which shows evidence for a bipolar flow and enhanced helium abundance indicating it may be a type i pn . for this case , the heii lines may contribute up to 10% of the observed low resolution br@xmath3 flux .
we also used the echelle data to derive line widths for the hi br@xmath3 and hei 2@xmath0p2@xmath0s lines .
we used gaussian fits in all cases .
the results are given in table 3 .
where the fit deviated noticeably from a single gaussian , we note this below in section 4.3 , and in the table by including the separate gaussian components required to achieve a good overall fit .
we also measured the widths of isolated oh night sky lines to determine the instrumental resolution , since these lines are significantly narrower than this .
the instrumental widths changed by no more than 10% between the different observing sessions ( the worst data being that obtained in january 1997 due to the problems with the slit mechanism ) . in all cases , the hei data is higher resolution than the equivalent br@xmath3 data .
the intrinsic linewidth for each observation was derived assuming the intrinsic and instrumental widths added in quadrature .
the values give in table 3 have been corrected for the instrumental resolution . the observed linear correlation between the [ oiii ] 5007 / h@xmath12 line ratio and the stoy temperature found by kaler & jacoby ( 1991 ) implies that we can look for trends in our data in an entirely model independent fashion by examining the correlation between infrared line ratios and the [ oiii]/h@xmath12 ratio .
the basic results of our survey then are shown in figure 4 .
figure 4(a ) shows the ratio of the hei 7@xmath20g4@xmath20f and br@xmath3 against the [ oiii ] 5007 / h@xmath12 line ratio .
the result is the expected one in the sense that the volume of the he@xmath6 region rises rapidly with effective temperature ( [ oiii ] line strength ) until it fills the entire pn . in figure 4(b ) we show the ratio of hei 2@xmath0p2@xmath0s and br@xmath3 against the [ oiii ] 5007 / h@xmath12 line ratio . here
there is a much less well defined trend with increasing [ oiii]/h@xmath12 ratio , as expected if the results from previous photoionisation models are correct .
in particular we note two key points : this ratio peaks at lower effective temperature than the hei 7@xmath20g4@xmath20f to br@xmath3 ratio and it decreases with increasing [ oiii]/h@xmath12 ratio beyond this peak , both of which are in accord with the models presented by ferland ( 1999 ) .
we can test whether we have introduced any bias in these results from combining the optical data from the literature with our observed infrared data by further examining the behaviour of helium - to - hydrogen line ratios in the optical and infrared separately . in figure 5(a ) we plot the ratio of 6678 hei 3@xmath0d2@xmath0p transition with h@xmath12 against hei 7@xmath20g4@xmath20f with br@xmath3 . smits ( 1996 ) determined values of expected line emissivities for high-@xmath23 states such as the hei 7@xmath20g4@xmath20f lines , as well as the more common optical lines in the general recombination case , including collisional effects amongst the lowest levels .
benjamin , skillman & smits ( 1999 ) expanded this treatment of the collisional effects to higher levels , and detail the resultant emissivities of the better known lines , but otherwise recover the same data for the pure recombination case .
the two solid lines in figure 5(a ) show the full range of possible theoretical values for @xmath24@xmath25 and t@xmath26k from the smits ( 1996 ) data .
all of the 3@xmath0d2@xmath0p and 7@xmath20g4@xmath20f lines should be dominated by recombination processes .
line opacity and collisional excitation , which can affect the lower lying triplet states significantly , should be relatively unimportant here . figure 5(a )
clearly indicates that this is the case , since there is a well defined linear relation between the two ratios which is in reasonable agreement with the predicted theoretical values .
it is interesting to compare this with figure 5(b ) , in which we plot the same optical ratio against the ratio of hei 4@xmath20s3@xmath20p with br@xmath3 . again
the straight lines show the expected theoretical values for the range @xmath24@xmath25 and t@xmath26k ( assuming no collisional excitation ) .
there is a larger theoretical dispersion in this instance , largely due to a stronger dependence of the hei 4@xmath20s3@xmath20p line fluxes on electron temperature . despite this
, there is a clear trend for some points to lie well away from the theoretical values , in the sense that the 4@xmath20s3@xmath20p blend is brighter than expected from recombination alone .
these are clearly objects in which the population of the 4@xmath27s level of hei is enhanced
. this can be due to either collisional excitation ( largely from 2@xmath27s4@xmath22s ) or opacity in the 2@xmath27s@xmath28p series ( leading to corresponding enhancements in the s
p series ) .
the relative contribution of the two can be estimated from the atomic data ( eg lumsden & puxley 1996 ) and for the typical conditions of the pn in this survey they contribute almost equally since it is primarily density that drives both effects .
the role of collisions have also been calculated more exactly for hei emission in general by benjamin , skillman & smits ( 1999 ) as noted above .
they find enhancement of order 50% in the 4@xmath22s3@xmath22p line is possible for the physical conditions being considered , in accord with our data .
the pn that lie away from the linear trend are obvious from table 2 from a comparison of the hei 4@xmath20s3@xmath20p and hei 7@xmath20g4@xmath20f fluxes , since these should be similar in the presence of recombination alone .
however , there does not appear to be a clear correlation between these objects and their electron density as would be expected .
this suggests that the estimated densities from the [ sii ] doublet are not entirely reliable . in particular ,
all of the pn showing heii emission appear to show this effect . in such objects of course
the s@xmath6 zone will lie at the edge of the nebula , which may not be representative of the nebula as a whole . in many cases
there is evidence for higher densities in the literature , as estimated from [ oii ] lines .
this is consistent with the simple model in which the [ sii ] emission arises at the edge of the nebula since o@xmath6 has a higher ionisation potential than s@xmath6 .
as noted in the introduction purely local velocity broadening of the 5842@xmath0s1@xmath0s hei resonance line can reduce the likelihood of absorption by neutral helium ( cf ferland 1999 ) , and hence reduce the ratio of the hei 2@xmath0p2@xmath0s and hi be@xmath3 lines . in the on - the - spot approximation for emission and absorption of this resonance line , only local broadening processes count , since no 584 photon will travel a sufficient distance for a global velocity field to be an important factor .
we therefore need a measure of the local broadening if we are to compare our data with the correct photoionisation model .
therefore we have assumed a simple model to decompose the observed line width into local and global velocity broadening with the pn .
we assume that the intrinsic line width is determined by three possible factors : the expansion velocity of the nebula , the thermal line width , and turbulence , the first of these being the only purely global broadening mechanism .
we also assume that the expansion velocity has a particularly simple form , namely a constant value everywhere in the nebula .
although unrealistic in practice , the variation in expansion velocity is generally sufficiently small that we can ignore it here .
the profile locally will appear gaussian , with the broadening as given by equation 2 below .
along a line of sight directly through the centre of the shell we would then see the classic split line profile composed of two gaussians offset from each other by twice the expansion velocity .
however since the the objects are compact ( mostly unresolved ) on the sky we see total velocity field in the beam .
the velocity profile is ` infilled ' from those regions of the nebula where the velocity field is more perpendicular to the line of sight than tangential , so that the double profile will not be evident in the actual observed data .
if we integrate this velocity profile over the whole extent of the pn , then we can derive the observed profile , which is given by ( for an unresolved source ) @xmath29 @xmath30 where @xmath31 is the emissivity as a function of position in the nebula ( which we assume here is constant ) , @xmath32 is the expansion velocity so that @xmath33 if @xmath34 is the vector defining our line - of - sight , and @xmath35 for a nominal t@xmath36k , and no turbulence , @xmath37kms@xmath38 for hi and @xmath39kms@xmath38 for hei . in the absence of expansion , the line profile will be gaussian , and the broadening is given solely by equation 2 . for most of the objects considered here ,
the actual derived profile is sufficiently similar to a gaussian that we can not uniquely determine the split between turbulence and global expansion .
we can estimate the thermal line width directly from the derived electron temperature in each region to remove that uncertainty . in practice
therefore the best discrimination comes from a comparison of the differences in the observed widths for the hi and hei lines .
we calculated the widths of the lines for varying values of v@xmath40 and v@xmath41 and sought the ` best ' match from this grid of data .
an example for t@xmath36k is shown in figure 6 .
for the hi and hei separately we find the set of curves that match the observed data . these then give the derived v@xmath41 and v@xmath40 values shown in table 4 . since a family of curves is possible in almost all cases only upper limits to v@xmath41 and v@xmath40 can be derived .
clearly , however , these are not joint limits , since when v@xmath41 is large v@xmath40 must be small and vice - versa .
for some of the objects there is a discrepancy between the values derived from the hi and hei lines . in some cases
this is due to the fact that the he@xmath6 and h@xmath6 zones do not overlap because the effective temperature of the exciting star is too low .
for these objects we use the limits on the hei line alone . for a small sample of objects
there are no reliable solutions within the context of our simple model .
comments on these objects are given below , together with other notes on individual pn .
bd+30@xmath423639 : there is considerable structure in bd+30@xmath423639 ( cf sahai & trauger 1998 ) .
the split in the two components seen in the hei line give v@xmath43kms@xmath38 in terms of our simple model , so no turbulent broadening is required .
this expansion velocity is not however compatible with the hi line data .
it seems likely that our simple model is not a good approximation for this source .
however , even given this , there appears little need to invoke a large turbulent velocity component .
crl 618 : the spectra acquired for this paper were centred on the brighter of the two outflow lobes visible in the optical .
the actual hii region is in fact completely obscured at optical wavelengths .
our data agree with previous published near infrared data ( hora , latter & deutsch 1999 ) , which also seems to have been centred on the visible lobe rather than the actual core .
we have further low resolution data that will be published separately of the core showing strong hi emission , and stronger hei emission than present in the lobes .
however , since we have only obtained echelle spectra of the lobe position we only quote the result for the hi and hei line strengths there .
it is possible that most of the emission we detect in the lobes is actually scattered light from the obscured central regions , since crl 618 is known to have significant polarisation .
trammell , dinerstein & goodrich ( 1993 ) show that , in the optical , all the recombination lines and high excitation forbidden lines are scattered into our line of sight .
we therefore do not attempt to determine turbulent or expansion components to the velocity from the observed data ( which would imply very large velocities ) , and for sake of the photoionisation model simply assume v@xmath4410kms@xmath38 .
hu 1 - 2 : this object shows clear signs of a bipolar outflow in the echelle spectra , rather than straightforward expansion , since the velocity field flips from red to blue shifted moving south across the position of the central star .
the double peaked line profile that results is clearly visible in both echelle spectra .
the difference in line width probably reflects a differing acceleration in the he@xmath45 and he@xmath46 zones in the outflow .
the observed widths limit v@xmath4710kms@xmath38 .
m 1 - 1 : only a two component gaussian fit gave a good match to the data .
the velocity separation of the two peaks is @xmath1458kms@xmath38 , indicating an expansion velocity of @xmath1430kms@xmath38 .
the predicted turbulent velocity component is then small , since most of the remaining line broadening must be thermal . because of the weakness of the hei 2@xmath0p2@xmath0s line in the low resolution data we did not observe this object with the echelle at 2.058@xmath1 m .
there are no satellite hei lines visible in the br@xmath3 data either .
ngc 7027 : this is the only one of our sample to show spatially resolvable velocity structure .
the structure most closely resembles an expanding shell in agreement with the radio data of roelfsema et al .
given this we can further limit the velocity components so that v@xmath48kms@xmath38 and v@xmath4710kms@xmath38 .
we used cloudy version 90.05 ( ferland 1996 ) to model the emission line spectra for a grid of pn with differing physical parameters .
the main parameters we considered were effective temperature , central star luminosity , electron density and the velocity structure .
all models used the standard cloudy planetary nebula abundances with their corresponding dust grain abundance .
we considered effective temperatures in the range 25000150000k .
we used two sets of model atmospheres .
first , to ensure that we have the same model for all temperatures , we used a simple blackbody approximation .
then , to test whether a different model might affect the results , we used the standard kurucz ( 1992 ) models with @xmath49 for the more restrictive temperature range of 25000 - 50000k .
the only alternative models of stellar atmospheres potentially applicable to low effective temperature pn in the version of cloudy used are due to mihalas ( 1972 ) .
we did not consider these however when it became clear that the blackbody models were a sufficiently good match to our data ( see section 5.2 ) .
there is observational and theoretical evidence that pn central stars with low effective temperatures all have approximately the same bolometric luminosity .
observational evidence comes from studies of pn in the magellanic clouds ( cf figure 11 in dopita et al .
the theoretical model tracks of vassiliadis & wood ( 1993 ) also show that below 50000k there is little dispersion in the luminosity - effective temperature domain . from vassiliadis & wood ( 1993 )
we adopt a value of 5000 as a mean bolometric luminosity , but consider 3000 and 7000 as well , since these match the spread in the range of observed and theoretical values for pn central stars below about 100000k .
the range of electron density and turbulent / expansion velocity considered are taken directly from the observed data .
we computed models with @xmath503000 , 6000 , 12000 , 24000 and 48000@xmath25 , and v@xmath510 , 5 , 10 and 15kms@xmath38 .
lesser effects that need to be considered are due to metallicity / abundance and the final radius chosen .
this latter parameter is really only important for the pn with hot central stars , since , for small outer radii , the nebula can be density rather than ionisation bounded , resulting in reduced he@xmath6 zones .
this is apparent in some of the observed data for those pn with hot central stars .
since it only affects the high temperature end of the models , we adopt @xmath52pc as an appropriate value for our sample ( cf the values in cahn , kaler & stanghellini 1992 ) , and accept that this may mean that we overestimate the model hei line strength at high effective temperature .
the inner radius has little effect on the strengths of the lines we are interested in , and we adopt @xmath53pc .
we do not consider metallicity effects further here , since there is no direct evidence from the observational data that it is a significant effect .
for example , dddm 1 is a known halo pn , which has lower metallicity than the other pn in the sample , but whose helium - to - hydrogen line ratios show little deviance from the rest of the sample .
our selection used the relation between the [ oiii ] 5007 to h@xmath12 ratio and stoy temperatures for the pn central stars as derived by kaler and jacoby ( 1991 ) to determine approximate stellar effective temperatures for our sample .
however , these temperatures are not completely reliable measures of the actual underlying t@xmath7 , since the method is only sensitive to effective temperatures below 60000k .
this can be seen clearly in figure 7 , where we show the original data used by kaler & jacoby , together with their fit to the same data , and the results from our photoionisation modelling .
it is encouraging that the blackbody stellar models reproduce the spread in the observed data reasonably well . in this case
it appears that the identification of the underlying effective temperature with the stoy temperature is sound up to 60000k .
this is in spite of the fact that a black body is not a good match to the actual spectrum of a pn central star .
however , previous workers have also found that their use gives a good match between model and observation ( cf . the discussion in dopita & meatheringham 1991 ) .
by comparison , the kurucz model atmospheres which are shown as the dashed line in figure 7 do not reproduce the observed relation at all , particularly at the lower end of our effective temperature range ( note also that the kurucz models only exist for t@xmath5450000k ) .
we therefore did not consider the further use of these models .
the primary mechanism giving rise to the spread in [ oiii ] 5007 to h@xmath12 ratio at a fixed temperature is the electron density .
collisional enhancement at a fixed temperature can change the ratio by 30% over the range of densities considered ( the largest value of the ratio occurring at @xmath55@xmath25 ; at higher values the ratio declines again ) , whereas the central star luminosity contributes at most a 15% change , and the velocity structure is essentially irrelevant . as can be seen , the [ oiii]/h@xmath12 ratio rises almost linearly with temperature , as the o@xmath56 region in the nebula increases in size , until about t@xmath5760000k .
however , beyond this temperature , the o@xmath56 region saturates , and eventually decreases with increasing t@xmath7 as o@xmath58 becomes dominant . in addition
, for our sample of compact pn , it is possible that the nebula is density rather than ionisation bounded .
this effect tends to become significant for temperatures above that at which the he@xmath45 region first appears ( @xmath59 ) . for a compact nebula , the available gas reservoir at this point may be less than that required to absorb all the emitted radiation below 912 . in this case , the flux from the lower ionisation species near the edge of the nebula will clearly drop first .
the effect is the same as that which leads to the well known difference between hydrogen and helium zanstra temperatures in pn at high temperature ( cf kaler & jacoby 1991 and references therein ) .
therefore for those pn showing evidence of heii emission we deduced a calibration between the 4686 heii line strength and t@xmath7 from the models .
first , we extended the effective temperature range by calculating a small set of models spanning the range 150000400000k , with l@xmath605000 , @xmath61@xmath25 and v@xmath515kms@xmath38 .
variations in the luminosity and density lead to changes of less than 25% in the 4686 heii to h@xmath12 ratio .
we interpolated these data to find effective temperatures for the real pn .
there was one problem with this approach , since the model heii 4686 to h@xmath12 ratio never increased beyond 0.55 ( near 300000k : again this is a reflection of the fact that the he@xmath45 zone is now becoming density bounded ) .
three of our pn show values larger than this however .
this may be a reflection of either a density bounded nebula or non spherical geometry .
for the sake of this paper we simply ascribe to these pn a lower limit on the effective temperature of 250000k . in figure 8
we show how the hei 2@xmath0p2@xmath0s to hi br@xmath3 ratio varies as a function of effective temperature .
we have explicitly split the dependence on ( a ) density , ( b ) turbulent velocity and ( c ) stellar luminosity to show how the ratio is affected by each .
the results are as expected
. the ratio increases with increasing density because of collisions from the metastable 2@xmath0s state enhancing the population of the 2@xmath0p state .
as the microscopic velocity width increases , the 584 2@xmath0p1@xmath0s resonance line weakens , leading to a decrease in the population of the 2@xmath0p state . as the stellar luminosity increases the ionisation parameter at the inner edge of the cloud also increases , leading to a marginal increase in the he@xmath45 region compared to a lower luminosity star , once that star is hot enough to actually excite this state , and a decline in the fraction of he@xmath5 and he@xmath6 .
since he@xmath5 is required to scatter the 584 photons , the direct consequence is a reduction in the ratio for effective temperatures beyond about 40000k . from figure 8
it is also clear that the models with the ` extreme ' high and low ratios are those with high density , and low turbulent velocity and central star luminosity and low density , and high turbulent velocity and central star luminosity respectively .
since the luminosity dependence is only weak however , we will ignore it in the following , and consider only our mean central star luminosity of l@xmath62 .
all of the strong hi and hei lines are calculated directly by cloudy , but the two weaker k band hei lines are not .
since the 2.16475@xmath1 m 7@xmath20g4@xmath20f line must be dominated by recombination processes , and the 6678 3@xmath0d2@xmath0p is known to be ( eg benjamin et al .
1999 ) , the ratio of these two lines must in theory be roughly constant .
figure 5(a ) shows that this is true observationally .
we can therefore use the data from smits ( 1996 ) to correct the model 6678 flux into a model 2.16475@xmath1 m flux .
the actual correction for t@xmath36k and @xmath63@xmath25 is that the model hei 7@xmath20g4@xmath20f to hi br@xmath3 ratio is 1.23 times the model hei 3@xmath0d2@xmath0p to h@xmath12 ratio ( where we have also used the data on hi from storey & hummer 1995 ) .
this values rises to 1.31 for t@xmath64k and @xmath63@xmath25 .
the value depends only slightly on electron density but much more strongly on electron temperature .
unfortunately , the behaviour of the 2.113@xmath1 m 4@xmath20s3@xmath20p line is not as simple to translate from the cloudy results , given the comments in section 4.2 .
we therefore chose not to compare it to the model , but rather to use its ratio with the 7@xmath20g4@xmath20f line as an indicator of collisional or line opacity effects , and hence of high electron density ( @xmath65@xmath25 ) .
as noted in section 4.2 , these objects can be found in table 2 by a comparison of the hei 4@xmath20s3@xmath20p and 7@xmath20g4@xmath20f fluxes . objects with significantly larger 4@xmath20s3@xmath20p fluxes than 7@xmath20g4@xmath20f fluxes must have enhanced 4@xmath22s populations , and hence high electron density . in figure 9(a ) we plot the observed hei 7@xmath20g4@xmath20f to hi br@xmath3 ratio and in figure 9(b ) the observed hei 6678 3@xmath0d2@xmath0p to h@xmath12 ratio as taken from the literature .
we also plot a typical result from our photoionisation models ( the actual results change by less than 5% as a function of density , luminosity or velocity , and 10% as a function of electron temperature ) .
we have not labelled the individual pn here since clearly both sets of data are in good agreement with the models .
therefore we can be confident that the ratio of a pure recombination hei and hi line does behave in the fashion expected , and that such a ratio could be used as a measure of effective temperature in the 3040000k range .
in figure 10 we plot the observed hei 2@xmath0p2@xmath0s to hi br@xmath3 ratio against effective temperature ( taken to be the greater of t(stoy ) or t(heii ) from table 1 ) .
the solid lines are the model results for the ` extreme ' models for a central star with a luminosity of 5000 as discussed in the previous section .
the lower curve is a model with @xmath66@xmath25 and v@xmath67kms@xmath38 and the upper curve is a model with @xmath68@xmath25 and v@xmath69kms@xmath38 .
we have labelled the individual pn to make identification easy .
we have compressed the temperature scale to show the detail at t@xmath70k , but note that the high excitation pn that are not plotted all lie above the predicted model curve . finally in figure 11 we plot the observed hei 2@xmath0p2@xmath0s to hei 7@xmath20g4@xmath20f ratio , since this removes any metallicity dependence .
we have plotted the same models as in figure 10 . from figure 10
, it is clear that the models systematically underpredict most of the high effective temperature pn .
these objects do show evidence for high density ( either from their [ sii ] ratios , or from the strength of the hei 4@xmath20s3@xmath20p line ) .
some must have low turbulent velocities ( m 1 - 20 for example ) , but there does not appear to be a clear trend of the ratio with this parameter ( remembering from figure 8 that even turbulent velocities at the 10kms@xmath38 level will suppress the hei 2@xmath0p2@xmath0s line by 25% at high t@xmath7 ) . indeed
3 of the sources shown in figure 10 and 11 ( m 1 - 20 , k 3 - 62 and m 1 - 4 ) lie above the maximum predicted theoretical values for the ratios shown , and the same trends also carry through to the higher effective temperature pn not plotted in figures 10 and 11 . from figure 11
it is also clear that enhanced helium abundance can not explain the observed data for these objects , since they also tend to lie above the model predictions there . for effective temperatures near and above 60000k
it is also clear from figure 8 that density has only a small role in determining the hei 2@xmath0p2@xmath0s line strength .
therefore higher density than that used in our modelling is also unlikely to account for these data .
the only obvious explanation for all these results is that the population of the hei 2@xmath0p level is not as small for these pn as predicted by cloudy .
it is not the ionisation structure that is wrong though , since the model does predict the strength of the neutral [ oi ] 6300 line correctly ( at a level between 1 and 5% of h@xmath12 for these pn ) .
therefore it seems that the only alternative is that the hei ly@xmath71 destruction probability as used in cloudy is an overestimate ( see ferland 1999 for a fuller discussion of this factor ) .
the behaviour of the low temperature pn ( those with t@xmath5428000k ) may be easier to understand .
three of the five objects sit within the model bounds , though they again show deviations from the model since bd+30@xmath423639 should have high density and no turbulence , yet is the lowest of the three .
more importantly , two objects lie below the predicted ratio .
for all these objects we must of course be cautious as to whether the temperature scale is accurate . a shift of 3000k in temperature ( the effective error on the kaler & jacoby relation ) , would be sufficient to explain all the data .
it should also be borne in mind that we may have ` errant ' observational values for crl 618 given the comments in section 4.3 .
also these pn are unlikely to be well matched by black body stellar atmospheres ( bd+30@xmath423639 is a wc star after all ) . the predicted ratio from the kurucz models we calculated dropped off more sharply with decreasing effective temperature for example ( in the same sense that the [ oiii]/h@xmath12 ratio declines more sharply than for a black body stellar atmosphere , as shown in figure 7 ) .
finally we come to the group of objects between 28000 and 40000k .
again there seems to be no direct correlation between the limits on the turbulent velocities and the observed line ratio in contradiction with the models .
the two most discrepant points here though are m 1 - 6 and m 1 - 14 . at least for the these
there is observational evidence for moderately high density .
m 1 - 6 may simply have the wrong assigned effective temperature as well , since it is the only object for which we have had to rely on the acker et al .
( 1992 ) catalogue for the 5007 [ oiii ] flux .
in addition , there is some evidence here for an enhanced helium abundance , since figure 11 shows that some of the pn that sit above the maximum model value in figure 10 lie well within the model boundaries of the hei 2@xmath0p2@xmath0s to hei 7@xmath20g4@xmath20f ratio .
m 1 - 6 , m 1 - 9 and m 1 - 14 are possible examples where this is true ( and we note that the first two of these pn are listed as having large helium abundances , and being peimbert type i objects , by shibata & tamura 1985 ) .
we also note however that there are other objects with similar effective temperatures that have even higher observed hei 2@xmath0p2@xmath0s to hi br@xmath3 ratios than m 1 - 14 , such as the ultracompact hii region , g45.12 + 0.13 , which again has high density ( lumsden & puxley 1996 ) .
therefore , two plausible arguments present themselves here : either there is some enhancement to the helium abundance over the 10% assumed in our modelling ( as might be expected for peimbert type i pn ) or the population of the hei 2@xmath0p level is collisionally enhanced above the maximum values derived using cloudy ( perhaps due to the limited number of triplet states considered in cloudy ) . at these values of the effective temperature
however there is no need to invoke a change in the hei ly@xmath71 destruction probability .
it is clear from the results we have presented here that the behaviour of the hei 2@xmath0p2@xmath0s line is sufficiently complex that it is not a reliable indicator of stellar effective temperature .
the variation of 30% or more in both observed and model predictions for a fixed temperature mean that only a very general picture of the effective temperature can be gained from this ratio .
it is possible that in certain restricted circumstances , such as objects with known density and turbulent velocity fields , that it may provide secondary information about the effective temperature however .
figure 11 provides an example of how it might be used in combination with another hei recombination line to constrain the temperature of objects above 40000k , where the hei / hi recombination line ratio alone is no longer varying rapidly .
however , for this to be reliable the discrepancies between model and data seen in our results will need to be explained .
there are factors that we have not accounted for that may help explain some of observations .
a change in the dust - to - gas ratio , clumping within the ionised gas or a density rather than ionisation bounded nebula could plausibly explain some of the discrepancies we see .
we would certainly not recommend the use of this line however as a primary indicator of stellar effective temperature .
by contrast , the behaviour of the hei 7@xmath20g4@xmath20f line , which should be due solely to recombination , is as expected , and in agreement with the models .
although this line is not in itself of particular value , since it is blended with hi br@xmath3 at intermediate spectral resolution , it does show that the use of such hei lines is in accord with the original intent of doyon et al .
in particular , the 1.7007@xmath1 m 4@xmath22d3@xmath22p transition is also expected to be due mostly to recombination , is in a clear part of the atmospheric h band window , and is well separated from other emission lines .
this line has been discussed in the past in this context ( eg , lumsden & puxley 1996 , vanzi & rieke 1997 ) .
we shall discuss it further in the paper presenting our low resolution spectra for this pn sample .
we would like to thank derck smits for providing the machine readable version of his hei predictions from his 1996 paper .
sll acknowledges support from pparc through the award of an advanced research fellowship .
sll also thanks the access to major research facilities program , administered by the australian nuclear science and technology organisation on behalf of the australian government , for travel support for the observations reported here .
the united kingdom infrared telescope is operated by the joint astronomy centre on behalf of pparc .
* figure 1 : * the hi br@xmath3 echelle spectra . the most common of the satelite lines is the 2.16475@xmath1 m hei 7@xmath20g4@xmath20f blend .
also visible in some of the spectra are the other hei transitions of 7@xmath21f4@xmath21d at 2.16229@xmath1 m and 7@xmath22f4@xmath22d at 2.16137@xmath1 m , and the heii 148 transition at 2.1653@xmath1 m .
the spectra are of : ( a ) bd+303639 ; ( b ) crl 618 ; ( c ) dddm 1 ; ( d ) hu 1 - 2 ; ( e ) k 3 - 60 ; ( f ) k 3 - 62 ; ( g ) k 3 - 66 ; ( h ) k 3 - 67 ; ( i ) k 4 - 48 ; ( j ) m 1 - 1 ; ( k ) m 1 - 4 ; ( l ) m 1 - 6 ; ( m ) m 1 - 9 ; ( n ) m 1 - 11 ; ( o ) m 1 - 12 ; ( p ) m 1 - 14 ; ( r ) m 1 - 74 ; ( s ) m 1 - 78 ; ( t ) ngc 7027 ; ( u ) pc 12 ; ( v ) sast 2 - 3 ; ( w ) vy 1 - 1 .
* figure 2 : * the hei 2@xmath0p2@xmath0s echelle spectra .
the spectra are of : ( a ) bd+303639 ; ( b ) crl 618 ; ( c ) dddm 1 ; ( d ) hu 1 - 2 ; ( e ) k 3 - 60 ; ( f ) k 3 - 62 ; ( g ) k 3 - 66 ; ( h ) k 3 - 67 ; ( i ) k 4 - 48 ; ( j ) m 1 - 4 ; ( k ) m 1 - 6 ; ( l ) m 1 - 9 ; ( m ) m 1 - 11 ; ( n ) m 1 - 12 ; ( o ) m 1 - 14 ; ( p ) m 1 - 74 ; ( q ) m 1 - 78 ; ( r ) ngc 7027 ; ( s ) pc 12 ; ( t ) sast 2 - 3 ; ( u ) vy 1 - 1 .
* figure 3 : * typical low resolution spectra of the sample .
the objects shown are ( in order of increasing excitation ) ( a ) m 1 - 12 , ( b ) m 1 - 9 , ( c ) k 3 - 67 and ( d ) m 1 - 1 .
the brightest lines are hi br@xmath3 and br@xmath72 at 2.1661@xmath1 m and 1.9451@xmath1 m respectively , and hei @xmath73s@xmath73p at 2.058@xmath1 m .
the bright line to the red of br@xmath3 in m 1 - 1 is heii 107 at 2.1891@xmath1 m , indicative of the very high central star temperature in this object .
* figure 4 : * the observed ratios of ( a ) 2.166@xmath1 m @xmath74g@xmath75f with br@xmath3 and ( b ) 2.058@xmath1 m @xmath73s@xmath73p with br@xmath3 against 5007 [ oiii ] with h@xmath12 . note the well defined trend of the data in ( a ) against the much more complex behaviour shown in ( b ) .
this illustrates the difference between a hei line arising purely from recombination and the @xmath73s@xmath73p transition .
* figure 5 : * the observed ratios of 6678 3@xmath0f2@xmath0d with h@xmath12 plotted against ( a ) 2.166@xmath1 m @xmath74g@xmath75f with br@xmath3 and ( b ) 2.113@xmath1 m @xmath76s@xmath77p with br@xmath3 .
note how the data in ( a ) follow an essentially linear relation , whereas in ( b ) there is significant scatter in the relation between the lines .
the latter is due to enhancement of the @xmath78s level of hei either through collisions or as a result of significant opacity in the @xmath79s@xmath80p series .
the solid lines in each figure represent the theoretical bounds for the plotted ratios for the range of electron temperature and density considered .
* figure 6 : * the derived value of the full width at half maximum of the integrated emission line using our simple expansion model as a function of v@xmath41 and v@xmath40 . the solid curves represent ( from bottom up ) values of v@xmath41 of 0 , 5 , 10 , 15 , 20 , 25 and 30kms@xmath38 . *
figure 7 : * the observed and predicted behaviour of the 5007 [ oiii ] to h@xmath12 ratio as a function of effective temperature .
the solid lines are generated using simple blackbodies for the model stellar atmospheres .
the dashed lines are generated from kurucz model atmospheres .
the crosses are the data from kaler & jacoby ( 1991 ) , where we have used their tabulated stoy temperatures for t@xmath7 .
the straight line is their best fit relation to this data .
the two curved lines for each set of models are the extreme values from our grid .
the upper line represents the prediction for a central star of luminosity 7000 , with electron density of 24000@xmath25 , the lower a star of luminosity 3000 and electron density of 3000@xmath25 . * figure 8 : * the predicted ratio of the 2.058@xmath1 m hei @xmath73s@xmath73p line with hi br@xmath3 as a function of stellar effective temperature . in ( a ) we show the dependence on density , in ( b ) on turbulent velocity and in ( c ) on central star luminosity . in each case
we held the other two parameters constant at the values of @xmath61@xmath25 , v@xmath81kms@xmath38 and l@xmath62 . in ( a )
the curves correspond to @xmath503000 , 6000 , 12000 , 24000 and 48000@xmath25 from lowest up . in
( b ) the curves correspond to v@xmath510 , 5 , 10 and 15kms@xmath38 from highest down . in ( c )
the curves correspond to l@xmath603000 , 5000 and 7000 from highest down . *
figure 9 : * the observed and predicted behaviour of ( a ) the ratio of the 2.16475@xmath1 m hei @xmath74g@xmath76f line with hi br@xmath3 and ( b ) the 6678 hei 3@xmath0d2@xmath0p with h@xmath12 as a function of effective temperature .
the model plotted has l@xmath62and t@xmath36k .
there is less than 5% variation of the model ratios with either density or turbulent velocity . *
figure 10 : * the observed and predicted behaviour of the ratio of the 2.058@xmath1 m hei @xmath73s@xmath73p line with hi br@xmath3 .
the models plotted are for @xmath68@xmath25 and v@xmath69kms@xmath38 ( upper curve ) and @xmath66@xmath25 and v@xmath67kms@xmath38 ( lower curve ) which are essentially the extreme maxima and minima in our model grids . in both cases l@xmath62 .
t@xmath7 is estimated from the observed 5007 [ oiii ] or 4686 heii to h@xmath12 ratio as discussed in the text .
* figure 11 : * the observed and predicted behaviour of the ratio of the 2.058@xmath1 m hei @xmath73s@xmath73p line with the 2.16475@xmath1 m hei @xmath74g@xmath76f line .
the models plotted are the same as in figure 8 .
|
we have observed a large sample of compact planetary nebulae in the near - infrared to determine how the 2@xmath0p2@xmath0s hei line at 2.058@xmath1 m varies as a function of stellar effective temperature , @xmath2 .
the ratio of this line with hi br@xmath3 at 2.166@xmath1 m has often been used as a measure of the highest @xmath2 present in a stellar cluster , and hence on whether there is a cut - off in the stellar initial mass function at high masses .
however , recent photoionisation modelling has revealed that the behaviour of this line is more complex than previously anticipated .
our work shows that in most aspects the photoionisation models are correct .
in particular , we confirm the weakening of the 2@xmath0p2@xmath0s as @xmath2 increases beyond 40000k .
however , in many cases the model underpredicts the observed ratio when we consider the detailed physical conditions in the individual planetary nebulae .
furthermore , there is evidence that there is still significant 2@xmath0p2@xmath0s hei line emission even in the planetary nebulae with very hot ( @xmath4100000k ) central stars .
it is clear from our work that this ratio can not be considered as a reliable measure of effective temperature on its own .
infrared : ism : lines and bands
ism : planetary nebulae : general
|
we have recently @xcite proposed a new form of nonextensive entropy which depends on a parameter similar to tsallis entropy @xcite , and in a similar limit approaches shannon s classical extensive entropy .
we have shown how the definition for this new form of entropy can arise naturally in terms of mixing of states in a phase cell when the cell is re - scaled , the parameter being a measure of the rescaling , and how shannon s coding theorem @xcite elucidates such an approach . in this paper
we shall adopt a more general attitude and try to develop the statistical mechanics of systems where the entropy is defined almost arbitrarily .
such a designer entropy @xcite may indeed be relevant in a specific context , but we shall not justify here any specific form .
the applicability of the tsallis form which leads to a levy - type pdf found in many physical contexts is now well established @xcite and in the earlier paper we have commented about how our form may also be more relevant in a context that demands a more stiff pdf . in this paper
we concentrate on the use of the pdf to obtain macroscopic quantities for general entropy functions .
central to the development of the statistical mechanics of a system is the definition of the free energy , because it is related to the normalization of the probability distribution function , which in turn controls the behavior of all the macroscopic properties of the ensemble .
hence we shall first establish a general prescription to obtain the free energy , and then determine its value in a simple physical case in terms of the temperature for tsallis and the newer form .
we shall then use it to get the specific heat in both cases and see how it changes with the change of the parameter at different temperatures .
the pdf is found by optimizing the function @xmath0 where @xmath1 is the entropy ( whichever form ) , @xmath2 the internal energy of the ensemble per constituent , @xmath3 is the probability of the @xmath4-th state which has energy @xmath5 , and @xmath6 and @xmath7 are the lagrange multipliers associated with the normalization of the pdf s @xmath3 and the conservation of energy .
let us make the assumption that the entropy @xmath1 is a sum of the components from all the states : @xmath8 where @xmath9 is a generalized function which will in general be different from the shannon form : @xmath10 we get from optimizing @xmath11 @xmath12 with the simple solution @xmath13 the constant of integration vanishing because there can be no contribution to the entropy from a state which has zero probability .
we shall assume that the helmholtz free energy @xmath14 is defined by @xmath15 hence if @xmath1 is nonextensive , and @xmath2 is extensive , then @xmath14 must also be nonextensive . using eq.5 and eq.6
get @xmath16 hence we get the relation for the pdf @xmath17 with the definition @xmath18 here we assume that the function @xmath19 can be inverted
. this may not always be the case for an arbitrary expression for the entropy , at least not in a manageable closed form .
finally , @xmath14 can be obtained from the constraint equation @xmath20 once @xmath14 has been determined , it may be placed in eq.8 to get the pdf @xmath3 for each of the states , all properly normalized , and we can find @xmath2 and its derivative @xmath21 , the specific heat . for the simple system that we shall be concerned with , pressure and volume , or their analogues , will not enter our considerations , and hence we have only one specific heat , @xmath22 , with @xmath7 now identified as the inverse scale of energy , the temperature @xmath23 , @xmath24
the shannon entropy eq.3 immediately gives , using eq.8 and eq.9 @xmath25 so that eq.10 yields the familiar expression for @xmath14 @xmath26 where @xmath27 is the partition function @xmath28 the exponential form of @xmath3 in eq.12 allows the separation of the @xmath14 dependent factor and hence such a simple expression for @xmath14 in the case of shannon entropy , giving us normal extensive thermodynamics . in the tsallis case
we have @xmath29 using the q - logarithm @xmath30 and this is easily seen to lead to , using the symbol @xmath31 for @xmath32 @xmath33 we note that as it is no longer possible to separate out a common @xmath14 dependent factor , we can no longer find an expression for @xmath14 in terms of the partition function in the usual way .
instead it is necessary to solve the normalization equation eq.10 . for a general value of @xmath31 this will give an infinite number of roots , but for values of @xmath31 corresponding to reciprocals of integers , we shall have polynomial equations with a finite number of roots , which too may be complex in general .
we shall later see , at least for the simple example we consider later , that a real and stable root can be found that approaches the shannon value of @xmath14 as @xmath34 , because in that limit the @xmath35 function in the definition of the tsallis entropy also coincides with the natural logarithm .
the entropy we proposed in ref.@xcite is given by @xmath36 with mixing probability @xcite @xmath37 we can also express it as the q - expectation value @xmath38 proceeding as in the case of tsallis entropy we get , with @xmath39 , @xmath40 where @xmath41 is the lambert function defined by @xmath42 like the tsallis case , here too we have to obtain @xmath14 by solving numerically the transcendental eq .
10 . for the specific heat
we get @xmath43 where we have written for brevity @xmath44 and have used the following identities related to the lambert function @xcite .
@xmath45 as @xmath46 for small @xmath47 , we see that for small @xmath31 we effectively get classical pdf and thermodynamics , as in the tsallis case .
so the parameter @xmath31 is again a measure of the departure from the standard statistical mechanics due to nonextensivity .
however , our nonextensivity is different functionally from the tsallis form @xcite , and the values of @xmath31 in the two forms can only be compared in the limit of low @xmath7 .
if we use the power series expansion of @xmath48 @xmath49 and compare that with the power series expansion for @xmath50 in a form of the tsallis @xmath3 similar to the new entropy form @xmath51 we get a cancelation of the parameter @xmath52 of the new entropy and of tsallis parameter @xmath53 in the first order , so that both distributions approach the shannon pdf , as we have already mentioned , but if we demand equality in the second order , we get @xmath54 the third order difference between @xmath55 and @xmath50 is only @xmath56 , and hence the difference between the tsallis form and our form of entropy will be detectable at rather low @xmath23 , i.e. high @xmath7 .
let us consider the simplest nontrivial system , where we have only two energy eigenvalues @xmath57 , as in a spin-1/2 system .
for a noninteracting ensemble of systems we shall expect the standard results @xcite corresponding to shannon entropy @xmath58/\beta \\ s = \log[2 \cosh(\beta e)]- \beta e \tanh ( \beta e ) \\ u = - e \tanh ( \beta e ) \\ c = ( \beta e)^2 /\cosh^2 ( \beta e)\end{aligned}\ ] ] @xmath14 for shannon ( top ) , tsallis with @xmath59 ( middle ) and tsallis with @xmath60 ( bottom),width=302 ] @xmath1 for same three entropy forms ( from bottom to top shannon , tsallis ( 0.1 ) , tsallis ( 0.25)),width=302 ] for the tsallis entropy we shall take @xmath61 and solve numerically .
the values are shown in figs 1 - 4 .
it is seen that tsallis entropy gives very similar shapes for all the variables and for @xmath62 we get a fit much nearer to the shannon form than for @xmath63 .
the specific heat shows the typical schottky form for a two - level system .
@xmath2 for same three entropy forms ( same order as for @xmath1),width=302 ] @xmath21 for same three entropy forms ( peaks bottom to top - shannon , tsallis ( 0.1 ) , tsallis ( 0.25)),width=302 ] though the variables involve both @xmath64 and @xmath65 , after finding @xmath14 we can replace @xmath65 by @xmath66 for faster execution of the numerics . for our new entropy
too we first find the numerical value of @xmath14 from the normalization condition @xmath67 and then use this value to find @xmath2 , @xmath1 , and @xmath21 . in figs .
5 - 8 we show the values corresponding to @xmath68 , which is half the tsallis parameter @xmath69 used , and also the shannon values .
we note that only for the @xmath1 curve we have a perceptible difference between tsallis entropy and our new entropy at values of @xmath7 near @xmath70 .
comparison of @xmath14 for shannon ( apart ) , tsallis with @xmath71 and new entropy for @xmath72 ( superposed),width=302 ] comparison of @xmath1 for same three forms of entropy ( shannon , tsallis ( 0.1 ) , new entropy(0.05 ) - tsallis is just over the new entropy ) .,width=302 ] comparison of @xmath2 for the same three entropies .
shannon is separated , other two overlap.,width=302 ] specific heat c for same three entropies .
again , only shannon is separated.,width=302 ]
we have presented above a simple prescription for finding the important thermodynamic variables for any given form of the entropy as a function of the pdf s of the states .
we note that despite the apparent complexity of the exponential or transcendental equations determining the primary variable , the helmholtz free energy @xmath14 , it is possible to numerically get stable values which approach the expected shannon values in the right limit of the parameter used , both in the tsallis case and in our case of the newly defined entropy . for higher values of the parameter
our entropy would give values varying significantly from the shannon or even the tsallis entropy , and there may be physical situations where that may indeed be the desirable characteristic .
but at low values the corresponding values of the parameters for the two distributions produce virtually completely overlapping graphs
. the form of the entropy may be a reflection of the effective interaction among the constituent systems of the ensemble , the shannon form being the limiting case of the zero interaction case , and tsallis or our form being results of different forms of interaction with @xmath31 standing for a coupling constant
. it may be interesting to investigate more complicated systems that are physically realizable and comparable .
it is noteworthy that most thermodynamic functions we have considered here are not crucially dependent on the form of the entropy with adjusted coupling , though the value of entropy itself may vary significantly in the different formulations .
this probably indicates that the best way to discriminate between the suitability of different definitions of entropy in different contexts may be in comparing quantities that relate most directly to entropy .
f. shafee , `` a new nonextensive entropy '' , nlin.ao/0406044 ( 2004 ) c. tsallis,_j .
_ , * 52 * , 479(1988 ) p. grigolini , c. tsallis and b.j .
west,_chaos , fractals and solitons_,*13 * , 367 ( 2001 ) a.r .
plastino and a. plastino , _ j. phys .
* 27 * , 5707 ( 1994 ) m.a .
nielsen and m. chuang , _ quantum computation and quantum information _
( cambridge u.p . ,
ny , 2000 ) p.t .
landsberg , `` entropies galore '' , _ braz .
j. phys . _
* 29 * , 46 ( 1999 ) c. beck , `` nonextensive statistical mechanics and particle spectra '' , hep - ph/0004225 ( 2000 ) o. sotolongo - costa et al .
, `` a nonextensive approach to dna breaking by ionizing radiation '' , cond - mat/0201289 ( 2002 ) c. wolf , `` equation of state for photons admitting tsallis statistics '' , _ fizika b _ * 11 * , 1 ( 2002 ) r.m .
corless , g.h .
gonnet , d.e.g .
hare , d.j .
jeffrey and d.e .
knuth , `` on the lambert w function '' , u.w .
ontario preprint ( 1996 ) r.k .
pathria , _ statistical mechanics _ , ( butterworth - heinemann , oxford , uk , 1996 ) p.77
|
we consider the problem of defining free energy and other thermodynamic functions when the entropy is given as a general function of the probability distribution , including that for nonextensive forms .
we find that the free energy , which is central to the determination of all other quantities , can be obtained uniquely numerically even when it is the root of a transcendental equation .
in particular we study the cases of the tsallis form and a new form proposed by us recently .
we compare the free energy , the internal energy and the specific heat of a simple system of two energy states for each of these forms .
|
it has been discovered that the edge mode of a fractional quantum hall ( fqh ) system is described as a chiral tomonaga - luttinger ( tl ) liquid @xcite .
when one makes a point contact between the edges , backward scatterings become possible only near that point contact .
such a system is expected to be described as a tl model with a scattering potential at @xmath1 @xcite . for a @xmath2 state ,
the model predicts a @xmath3 dependence of the tunneling conductance near the zero temperature @xcite which is consistent with the recent experiment @xcite . the result does not rely upon whether we start the renormalization group ( rg ) analysis with the picture where quasiparticles tunnel between the edges through the bulk fqh liquid ( quasiparticle tunneling picture ) or the picture where electrons tunnel through the vacuum between the edges of two condensates separated by the tunnel effect ( electron tunneling picture ) @xcite . to describe the bulk fqh liquids with a filling factor
@xmath4 hierarchy constructions are introduced @xcite . at a given filling one
can easily propose many hierarchical constructions .
when we apply the above discussion on the edge mode to the hierarchy states , we encounter a completely novel situation .
let us consider a two - terminal geometry where the bulk fqh liquids between the left and right terminals have upper and lower edges .
each of the edge branches consists of more than two channels . in this case
the tunneling at apoint contact as well as the interaction between the channels become important for the quantum transport at the edges . for @xmath5 and 2/3
there are two channels of the edge modes @xcite which are specified by a set of filling factors ( @xmath6 ) .
however the possibility that there are more edge channels can not be excluded @xcite . for @xmath0 two constructions with @xmath7 and ( 1/3,1/3 ) have been proposed , which correspond respectively to sharp @xcite and smooth confinement @xcite . in the former ( latter ) construction two edge modes in a same branch propagate in the opposite ( same ) direction .
recent numerical calculations based on the composite fermion approach in the hartree approximation confirmed that there is a transition from the macdonald s ( former ) picture to the beenakker s ( latter ) one as the confinement is relaxed @xcite .
so far we have not taken into account the spin degrees of freedom since we have assumed that all the spins are polarized .
however finite - size calculations of the coulomb energy show that the unpolarized spin configuration is preferred when @xmath5 or @xmath8 @xcite .
the result is consistent with the experiment in tilted magnetic fields @xcite .
nevertheless we can realize the spin - polarized fqh states at arbitrary filling by applying the strong parallel magnetic field .
wen has shown that the fqh state has gapless boundary excitations using the gauge invariance of the bulk effective theory @xcite . in his derivation ,
low - energy effective action for the edge mode is determined only from the gauge invariance . in the appendix
we give an alternative derivation .
we describe the edge mode as a self - induced eigenmode of the bulk fqh state @xcite .
it turns out that the existence of maxwell terms in the dual action of the bulk effective theory is essential for our derivation .
we study the effects of backward scatterings at a point contact using the effective action generalized for the two - channel edge modes .
we concentrate on the analysis of @xmath5 and 2/3 fqh edges but our formulation can be applied to any two - channel fqh edges .
we assume that the interaction between the electrons is short - ranged .
the interaction between the upper and lower branches are neglected .
we consider both of the limits where the scattering potential is very weak and very strong . in the weak potential limit
we start the rg analysis from the quasiparticle tunneling picture and study the scaling behavior of the deviation of the conductance from the quantized value .
the situation corresponds to a `` high '' temperature .
as the potential becomes strong , the tunneling through the potential barrier becomes rare and we can employ the dilute instanton gas approximation ( diga ) .
as the potential further increases , it is better to start with the electron tunneling picture .
the question is whether or not the scaling behaviors of the tunneling conductance are smoothly connected from one picture to the other .
the plan of the paper is the followings . in sec .
ii the model for the edge mode is introduced . in sec .
iii the quasiparticle tunneling picture is explained , and the rg analysis in the weak and strong potential limit is given .
electron tunneling picture is studied in sec .
v is devoted to discussion and conclusions .
our model describes a two - terminal hall bar geometry where a two - dimensional electron system between the left and right terminals has upper and lower edges .
low - energy excitations of the system are the modes which are localized near the edges since bulk fqh state has an energy gap . in our model ,
edge modes of both upper and lower branches consist of two channels ( 1-channel and 2-channel ) . in the following , \(a )
we assume that long - range coulomb interactions are screened away and electrons in the fqh edges have only intra - branch short - range interactions .
\(b ) we consider only the backward scatterings at a point contact , since forward scatterings are relevant but do not open an energy gap @xcite , and since backward scatterings except at a point contact are forbidden as a result of momentum conservation .
\(c ) we also assume that the spins are fully polarized .
this may not be true for @xmath5 and @xmath8 .
this issue will be discussed in future publications .
the lagrangian density ( a7 ) in the appendix is the starting point of our following analysis .
we generalize it to the two - channel case as @xmath9\ ] ] short - range electron - electron interactions have the form @xmath10 where @xmath11 s are intra - channel interactions : @xmath12 for @xmath13 and @xmath14 is the inter - channel interaction : @xmath15 .
all @xmath16 s are positive since we consider the repulsive coulomb interaction .
the most general form of the edge action for the multi - channel fqh states has the form @xmath17 in our two - channel model , since two matrices @xmath18 and @xmath19 are given by @xmath20,\ \ \
v = \left [ \begin{array}{cc } v_1/\nu_1 & g_{12 } \\
g_{12 } & v_2/\nu_2 \end{array } \right]\ ] ] where @xmath21 , the action ( 2 ) reduces to the following special form @xmath22 where @xmath23 .
@xmath24 is a 4 by 4 matrix which has the form @xmath25\ ] ] where _ i _ is the 2@xmath262 identity matrix , and @xmath27\ ] ] with @xmath28 for @xmath29 .
when we introduce backward scatterings between the edges by making a point contact , the system is described as the tl model with a scattering potential at @xmath1 . since here
we do not need the detailed structure of the scattering potential @xmath30 $ ] , we leave the analysis of it to the next section .
all we need here is that the total action has the form @xmath31 $ ] .
the partition function @xmath32 of the system can be written in terms of the field @xmath33 as @xmath34-\int d\tau u[\theta(\tau)]}\ ] ] where @xmath35 = s_0 + i\int d\tau \lambda^t(\tau)\left(\theta(\tau ) - \phi(\tau , x=0)\right)\ ] ] with @xmath36 , @xmath37 by introducing the auxiliary field @xmath38 so that we may obtain @xmath33 after integrating out the @xmath38 s .
we perform successive integrations as @xmath39 } = \int{\cal d}\lambda e^{-s_0[\theta,\lambda ] } = e^{-s_0[\theta]}$ ] where @xmath40 $ ] has the form @xmath41 = \frac{1}{\beta}\sum_{\omega } \left ( \lambda^t(-\omega)b(\omega)\lambda(\omega ) + i\lambda^t(-\omega)\theta(\omega ) \right)\ ] ] with @xmath42 where @xmath43 is the adjugate matrix of @xmath44 .
the matrix @xmath45 can be explicitly evaluated .
the determinant of @xmath24 is given by @xmath46 where @xmath47 with @xmath48 .
it is equivalent to @xmath49 where @xmath50 .
@xmath51 are positive constants which satisfy @xmath52 it turns out that finite components are @xmath53 , @xmath54 , and @xmath55 .
the other components vanish since their integrands have odd powers of k. we obtain @xmath56 where @xmath57 , @xmath58 . rearranging the fields as @xmath59 , @xmath60 , we can write the action ( 9 ) in the form @xmath40 = s_0^{(+)}[\theta^{(+)},\lambda^{(+ ) } ] + s_0^{(-)}[\theta^{(-)},\lambda^{(-)}]$ ] , where @xmath61 = \frac{1}{\beta}\sum_{\omega } \left ( \frac{\pi}{|\omega|}\lambda^{(\pm)t}(-\omega)p \lambda^{(\pm)}(\omega ) + i\lambda^{\pm t}(-\omega)\theta^{\pm}(\omega ) \right)\ ] ] the matrix p is given by @xmath62 & ( \eta>0 ) \\ \\ \left [ \begin{array}{rr } \nu_1p & q \\ q & -\nu_2p \end{array } \right ] & ( \eta<0 ) \end{array } \right.\ ] ] integrating out @xmath63 s , we finally obtain the effective action for @xmath64 as @xmath65 = s_0^{(+)}[\theta^{(+)}]+s_0^{(-)}[\theta^{(-)}]$ ] with @xmath66 = \frac{1}{4\pi\beta}\sum_{\omega } partition function of the system now reduces to @xmath67- \int d\tau u[\theta(\tau)]}$ ] .
note that @xmath68 and @xmath69 are decoupled in @xmath70 . in later sections
, we will show that the scaling dimensions of the tunneling amplitudes are universal if both of the channels propagate in the same direction .
the stability of the system is crucial for this universality .
the stability of the system requires that the hamiltonian should be positive definite . in order to proceed
it is convenient to transform the action into a representation in which both @xmath18 and @xmath19 are diagonal .
this can be accomplished by performing the following transformations @xcite : @xmath71 , where the first step is the diagonalization of the @xmath18 matrix : @xmath72 with @xmath73 .
the second step is given by @xmath74 and @xmath75 where @xmath76 with @xmath77 .
in the final step we choose as @xmath78 and @xmath79 with @xmath80 . in the @xmath81-representation
the action ( 2 ) has the form in which all the channels are decoupled , @xmath82.\ ] ] the situation is similar for the hamiltonian @xmath83 where @xmath84 , and @xmath85 satisfy the following commutation relations , @xmath86= \pm\frac{k}{2\pi}\left(k^{-1}\right)_{ij},\ \ \ [ \tilde{\rho}_i^{(u , l)}(k ) , \tilde{\rho}_{j}^{(u , l)}(-k)]= \pm\frac{k}{2\pi}\sigma_i\delta_{ij}\ ] ] the stability of the system requires that all @xmath87 s should be positive . in our two - channel model , @xmath18 and @xmath19 are given in ( 3 ) .
since @xmath18 has been already been diagonalized , @xmath88 is chosen to be an identity .
the choice of @xmath89 is trivial .
finally we must choose @xmath90 such that @xmath78 .
hence it is enough to study the following two cases @xcite .
( i)@xmath91 in this case , the matrix @xmath92 is the identity except its sign , @xmath93 where @xmath94 .
the matrix @xmath90 can be chosen as @xmath95 $ ] where @xmath64 must satisfy @xmath96 .
it can be shown that @xmath97 and @xmath98 are two different solutions of the quadratic equation @xmath99 .
note that the discriminant @xmath100 is always positive , which ensure the real solutions for @xmath101 and @xmath102 .
hence the stability condition reduces to @xmath103 and @xmath104 . in this case , the matrix @xmath105 given in ( 11 ) becomes diagonal and has only positive universal components .
recall that @xmath50 with @xmath48 . in the present case since both @xmath106 and @xmath107 are positive , @xmath108 .
hence both channels propagate in the same direction .
( ii)@xmath109 in this case , the matrix @xmath90 which satisfies @xmath78 has the form @xmath110 $ ] where @xmath111 .
it is shown that @xmath97 and @xmath98 satisfy the following quadratic equation respectively : @xmath112 .
the requirement of stability is that both @xmath97 and @xmath98 should have a positive solution , which is equivalent to @xmath113 .
the matrix @xmath105 is non - universal and has off - diagonal components . in the present case since both @xmath106 and @xmath107 are negative , @xmath114 .
hence both channels propagate in the opposite direction .
in this section , we first consider the limit where the scattering potential is very weak , i.e. we start the rg analysis from the situation where the upper and lower branches are separated by the bulk fqh liquid .
quasiparticle tunnelings are possible only near the point contact since only there not only upper and lower branches approach but also the translational symmetry breaks down . performing the rg transformation , we calculate exactly the scaling dimensions of the scattering amplitudes . next we consider the opposite limit where the scattering potential is very strong . in this limit , the electron transport can be viewed as the tunneling of the phase @xmath64 from a potential minimum to an adjacent minimum .
this process corresponds to an instanton or an anti - instanton . by the duality mapping we transform the original model to an analogous model in the weak potential limit and rg analysis of the tunneling amplitude
is given .
it will be shown that the scaling dimensions are universal when @xmath91 , while it depends on the interaction when @xmath109 . in case ( i )
most of the processes are relevant , but for the ( 1,@xmath1151/3 ) state only @xmath116 is positive , which results in a different situation for the tunneling at low temperature .
now we analyze in detail the backward scatterings at a point contact .
the scattering potential at @xmath1 has the form @xmath117 where @xmath118 corresponds to the intra - channel quasiparticle tunneling process : @xmath119 where @xmath120 is the electron operator on the upper ( lower ) edge . in terms of the bose field
@xmath121 the electron operator can be written as @xmath122 a similar formula is given in ( a5 ) in the appendix .
the other term , which will be proven to be important for our following discussions , describes inter - channel scatterings . for @xmath91 , it has the form @xmath123 where @xmath124 for @xmath109 , @xmath125 where @xmath126 here @xmath127 and @xmath128 , and we assumed that @xmath129 with @xmath130 being an odd integer .
@xmath131 corresponds to inter - branch ( intra - branch ) tunnelings for @xmath91 ( @xmath109 ) . in the following we also assume @xmath132 for @xmath91 and @xmath133 for @xmath109 .
hence for @xmath91 @xmath134 and for @xmath109 @xmath135 we assume that all @xmath136 s are positive but the result does not depend on the signs of @xmath136 s .
we first consider the limit where the scattering potential is very weak .
we calculate the scaling dimensions of the scattering amplitudes by recursively integrating out the high - frequency modes @xcite @xcite @xcite . at zero temperature
, the partition function can be written as @xmath137- \int d\tau u_{\lambda}[\theta_{\lambda}(\tau)]}$ ] .
here we introduced a cutoff @xmath138 such that @xmath139 the unperturbed action @xmath140 $ ] has the form @xmath140 = s_0^{(+)}[\theta_{\lambda}^{(+ ) } ] + s_0^{(-)}[\theta_{\lambda}^{(-)}]$ ] where @xmath141 = \frac{1}{4\pi\beta}\sum_{\omega } p^{-1}\theta_{\lambda}^{(\pm)}(\omega)\ ] ] we divide the field @xmath142 into slow and fast modes as @xmath143 such that @xmath144 integrating out the fast modes , we obtain an effective action for the slow modes @xmath145 : @xmath146-\int d\tau u(\mu)[\theta_{\mu}(\tau)]}$ ] . here
we keep only the first order terms with respect to the scattering potential u since they are most relevant .
we define the scaling dimension @xmath147 as @xmath148 for @xmath149 .
these are determined by the equations @xmath150 for @xmath13 , and @xmath151 .
@xmath152 s are the correlation functions of the fast modes : @xmath153 where @xmath154 is a smoothing function with @xmath155 for @xmath156 and @xmath157 for @xmath158 @xcite . @xmath147 s are universal only when @xmath91 corresponding to the form of matrix @xmath105 given in ( 15 ) .
the tunneling conductance scales at a `` high '' temperature as @xmath159}\ ] ] where @xmath160 is the quantized hall conductance .
the exponent @xmath147 s are listed in table 1 , and explicit values are presented for @xmath5 and 2/3 cases in table 2 .
here we consider the limit where the scattering potential is very strong . in this limit
, the electron transport can be viewed as the tunneling from a potential minimum to its adjacent one , and tunneling matrix elements , i.e. instanton fugacities are the expansion parameters . using the dilute instanton gas approximation ( diga ) , we transform the original model to its dual theory in the weak potential limit and apply to it the rg analysis in the preceding subsection . in determining the minima of the scattering potential
@xmath161 $ ] at @xmath1 , only @xmath162 , @xmath163 , and @xmath164 ( @xmath165 ) for @xmath91 ( @xmath109 ) are independent . after integrating out the other degree of freedom , the partition function has the form @xmath166- \int d\tau u[\theta_1,\theta_2,\theta_3]}$ ] where @xmath167 = \frac{1}{4\pi\beta}\sum_{\omega } @xmath168 and q is a 3 by 3 matrix which has the form @xmath169 where @xmath170 is defined as @xmath171 the scattering potential ( 22 - 25 ) now has the form @xmath172 where @xmath173 hence the scattering potential has minima at @xmath174 where @xmath175 are integers which satisfy the condition that @xmath176 is ( a)even for even n and ( b ) odd for odd n. it follows that the minima of the potential form a double - layer lattice in the three - dimensional @xmath177-space .
the typical layers of that lattice are the equi-@xmath178 planes with ( a ) @xmath179 and ( b ) @xmath180 . using the diga we show below that the partition function in the strong potential limit is mapped to an analogous one in the weak potential limit @xcite .
in contrast to the single - channel case , it turns out that the dual theory is not identical to the original one . in the diga
the field @xmath181 can be written as @xmath182 where @xmath183 s are listed in table 3 .
@xmath184 s are linear combinations of instantons ( @xmath185 ) and anti - instantons ( @xmath185 ) : @xmath186 where @xmath185 and @xmath187 are respectively the topological charge and the central imaginary time of the k - th instanton in the j - th species .
@xmath188 is a one - instanton solution at @xmath189 the width of which should be neglected compared to @xmath190 in the diga .
the instanton or anti - instanton in @xmath191 which we call those of the @xmath192-th species corresponds to a tunneling process from @xmath193 to @xmath194 .
here we took into account only the nine species listed in table 3 since they are most likely to be relevant .
tunneling of an instanton or an anti - instanton along the @xmath195-axis , @xmath196-axis , and @xmath178-axis can be interpreted physically as a pulse of 1-channel current , 2-channel current and the total charge density fluctuation respectively .
the bosonization of current and density is given for example in the appendix , where they are expressed as @xmath197 and @xmath198 .
we may write the fourier transform of @xmath181 as @xmath199 where we used the approximation @xmath200 .
the partition function now has the form @xmath201}$ ] where @xmath202 s satisfy ( 36 ) .
@xmath203 denotes a `` summation '' over instanton configurations , or explicitly , @xmath204 where @xmath205 is the instanton fugacity , i.e. the tunneling matrix elements from @xmath206 to @xmath207 .
we consider the gaussian integral over the``dual fields '' : @xmath208 where @xmath209 and insert it into the `` summation '' in the partition function .
next we perform the shift transformation @xmath210 which cancels the original action @xmath211 $ ] .
we can write the partition function in the form @xmath212- i\sum_{j=1}^{9}\sum_{k=1}^{n_j}e_{jk } \sum_{i=1}^{3}c_{ij}\tilde{\theta}_i(\tau_{jk } ) \right]\ ] ] where @xmath213 = \frac{1}{4\pi\beta}\sum_{\omega } taking the `` summation '' over instanton configurations , we can finally write the partition function in terms of the dual action @xmath214 $ ] as @xmath215}$ ] where @xmath214 = \tilde{s}_0[\tilde{\theta}]+\int d\tau y[\tilde{\theta}(\tau)]$ ] with `` the scattering potential '' @xmath216 = \sum_{j=1}^{9}2y_j\cos \left ( \sum_{i=1}^{3}c_{ij}\tilde{\theta}_i \right)\ ] ] it turns out that the unperturbed part @xmath217 $ ] of the dual action can be identified with the original @xmath65 $ ] by the correspondences @xmath218 and @xmath219 .
in contrast to the single - channel case , the potential term @xmath220 $ ] is analogous but not identical to the original potential @xmath161 $ ] in spite of the correspondence @xmath221 .
it is straight - forward to calculate the scaling dimensions @xmath222 of the tunneling matrix element @xmath205 , which is defined by @xmath223 these are calculated using the equations @xmath224 where @xmath225 s are the correlation functions of the fast modes : @xmath226 the results are listed in table 4 .
the explicit values of @xmath222 for some specific edge constructions are shown in table 5 . in the following
we give a detailed discussion on the temperature dependence of the tunneling conductance in the dilute instanton gas ( dig ) picture .
first we consider the case where both of the edge channels propagate in the same direction . in sec .
ii we have shown that @xmath147 s are universal only for @xmath91 , which is also the case with @xmath227 s ( table 4 ) .
table 5 of @xmath227 s tells us that all instanton fugacities scale to zero when @xmath91 ( i ) . in this case
we can determine the temperature dependence of the tunneling conductance using dig picture : @xmath228 where @xmath229 $ ] .
it should also be noted that the inter - channel tunnelings play a crucial role in determining @xmath230 . in table 6
the explicit values of @xmath230 is shown , for comparison , in the presence and in the absence of inter - channel tunnelings . here
we consider the ( 1,@xmath1151/3 ) state which belongs to case ( ii ) : @xmath109 .
as was shown in table 4 the exponents depend on the interaction between the channels , and not all of the instanton fugacities scale to zero , i.e. the one ( @xmath231 ) which corresponds to @xmath232 scales to a larger value . in sec .
ii we have shown that the scattering potential @xmath131 which suppress the fluctuation of @xmath232 is irrelevant .
hence at both strong and weak potential limit @xmath69 is not pinned .
in contrast to the neutral mode @xmath232 all @xmath205 s which describe the instanton tunneling in the @xmath233 or @xmath234 direction scale to zero , i.e. both @xmath233 and @xmath234 are fixed .
however it is shown in sec .
ii that only @xmath234 is pinned at weak potential limit .
these two pictures in the strong and weak potential limits are not smoothly connected .
they can be understood by considering a rg flow diagram of the type shown in fig .
1 , which was constructed in the following way . the inter - channel tunnelings are irrelevant at both of the strong and weak potential limit for the ( 1,@xmath1151/3 ) state and can be neglected .
it is enough to study only the intra - channel tunnelings . in the following we study three dig pictures .
first we consider the situation where only @xmath233 is pinned .
we start with the action @xmath235 + \int_0^{\beta}d\tau \left(u_1[\theta_1^{(+)}]+u_2[\theta_2^{(+)}]\right)\ ] ] where @xmath236 was given in ( 16 ) , and @xmath237 and @xmath238 were introduced in ( 35 ) .
the scaling dimension of @xmath239 can be calculated as @xmath240 irrespective of the strength of @xmath241 .
the scattering potential @xmath239 is considered to be relevant also in the dig picture .
on the other hand on the axis @xmath242 the effective action for @xmath233 is obtained by integrating out @xmath234 : @xmath243 where we used the relation @xmath57 , @xmath58 , @xmath47 and @xmath48 .
we can apply diga to the action ( 47 ) to obtain its dual version @xmath244 where @xmath245 is the instanton fugacity , the scaling dimension of which is obtained as @xmath246 since @xmath247 for the @xmath248 state the instanton fugacity @xmath245 is relevant . in terms of the original scattering potential @xmath241 it tends to decrease in the large-@xmath241 region . recall that @xmath241 is irrelevant in the small-@xmath241 region .
we conclude that both the original and the dual picture belong to the domain ( a ) of fig
next we consider the situation where only @xmath234 is pinned . in this case
the same procedure gives us @xmath249 where @xmath250 .
it turns out that the scattering potential @xmath241 is irrelevant but @xmath239 tends to increase , which indicates that @xmath251 is an atractive fixed point in the @xmath252-plane .
finally we consider the case where both @xmath233 and @xmath234 are pinned .
we start with the action ( 45 ) and perform the duality mapping to obtain @xmath253 where p is a @xmath254 matrix given in ( 15 ) .
the scaling dimensions of @xmath245 and @xmath255 are calculated as @xmath256 where we used the relation @xmath257 .
it turns out that both of the instanton fugacities are irrelevant . in terms of the original scattering potential ,
both @xmath241 and @xmath239 tend to increase , i.e. @xmath258 is another atractive fixed point in the @xmath252-plane .
it is evident that this picture is not smoothly connected to the above two dig pictures where only one of the scattering potential is strong .
therefore we expect that there are two domains in the rg flow diagram in terms of the original scattering potential @xmath241 and @xmath239 . @xmath251 and @xmath258 are the atractive fixed points in the domain ( a ) and ( b ) of fig .
1 , respectively .
in the latter half of the preceding section , we consider the limit of strong scattering potential in terms of duality mapping .
there it was revealed that the tunneling amplitude through the potential is irrelevant in most of the cases
. then it would be more appropriate to start with the picture where the bulk fqh state is divided into left and right condensates and electrons tunnel between the left and right edges through a insulating region at the point contact .
first we study the electron tunneling ( et ) picture for @xmath91 . as was pointed out
before , the situation is different for the ( 1,@xmath1151/3 ) state . in this case
we must consider two different et pictures both of which correspond to two atractive fixed points in the @xmath252-plane , respectively .
these subjects will be discussed toward the end of this section . at a point contact ,
two channels are close to each other within the magnetic length , the single - channel electron operator @xmath259 on the left ( right ) branch is no longer the most general one .
the general electron operator @xmath260 may contain charge transfers between the edges of the same branch and takes the following form @xcite @xmath261 where the elementary charge transfer @xmath262 is given by @xmath263\ ] ] the electron tunneling from the left to right edge is expressed by the operator @xmath264 .
first we consider the case where the two channels propagate in the same direction .
the electron tunneling through a point contact can be described using the concept of tunneling hamiltonian @xcite @xmath265 where field operators on the left and right edges anticommute , which in the present picture corresponds to the fact that two edges are those of independent incompressible fluids .
we consider a nonequilibrium situation where the chemical potentials on each of the edges are different by the applied voltage : @xmath266 .
the tunneling current @xmath267 from the left to right edge is given by the average value of the operator @xmath268 where @xmath269 is the number of particles on the left edge .
since we concentrate on the linear response of @xmath267 with respect to @xmath270 , we can evaluate the average using kubo formula : @xmath271\rangle$ ] where @xmath272 . in terms of the fourier transform of the retarded green s function @xmath273\rangle$ ]
we can write the current as @xmath274 $ ] .
hence the scaling dimension @xmath275 of the tunneling conductance in the electron tunneling picture is equal to that of @xmath276 @xcite : @xmath277 we can determine the scaling dimension of @xmath278 by applying the rg analysis to the partition function , which can be written in terms of @xmath279 as @xmath280-s_0^{(r)}[\theta^{(r ) } ] -\int d\tau h_t[\theta^{(l)}(\tau),\theta^{(r)}(\tau)]}\ ] ] where @xmath281 = \frac{1}{4\pi\beta}\sum_{\omega } \left [ \begin{array}{cc } 2/|\nu_1| & 0 \\ 0 & 2/|\nu_2| \end{array } \right ] \theta^{(l , r)}(\omega).\ ] ] with @xmath282 .
here we have integrated out the fields except at the point contact . in the present picture @xmath283 and @xmath284
are not canonically conjugate but are the independent degrees of freedom .
next we introduce a cutoff @xmath138 and apply the rg analysis in the preceding section . performing the integral over the fast modes , we obtain @xmath285\ ] ] where @xmath286 is @xmath287 using @xmath288
we can write @xmath275 as @xmath289 -{1\over\nu_3}\right)-2.\ ] ] it is shown that the minimum value is obtained when @xmath290=1 for @xmath291 , @xmath290=1 or 2 for @xmath292 and @xmath290=2 or 3 for @xmath293 , respectively . hence @xmath275 s are evaluated as @xmath294 for the ( 1,@xmath1151/3 ) state we must consider two different et pictures . in sec .
iii we have shown that there are two attractive fixed points in the @xmath295-plane , @xmath251 and @xmath296 .
we can consider the et picture in the vicinity of these points . in sec .
iii we performed the rg analysis near @xmath297 , where we discuss the inter - channel tunneling process @xmath131 as well as @xmath298 and @xmath299 . here
we concentrate on the analysis of intra - channel tunnelings @xmath300 and @xmath301 since the rg flow of @xmath69 have turned out to be trivial . for @xmath302
we start with the situation where only the 2-channel is reflected . in this case
we slightly modify the discussion in sec .
iii as @xmath303 since we consider the electron tunneling from the left to right branch for the 2-channel .
we obtain @xmath304 where @xmath305 as @xmath306 . we confirm that @xmath251 is a stable fixed point .
recall that in the dig picture the scaling dimension of instanton fugacity @xmath245 in the present region is given by @xmath307 which is smaller than @xmath308 since @xmath309 .
we predict the existence of a crossover between the two pictures . for @xmath310
we start the rg analysis with the picture where both of the channels are reflected : @xmath311 , @xmath312 .
we obtain @xmath313 , @xmath314 where @xmath315 and @xmath316 as @xmath306 .
it is confirmed that the point @xmath296 is another stable fixed point . in this case
the scaling dimensions of the tunneling amplitudes are precisely equal to those of the instanton fugacities in the dig picture .
hence we predict no crossover between the two pictures .
the whole @xmath252-plane is divided into two domains separated by a boundary ( the dotted line in fig.1 ) , each of which has the attractive fixed point @xmath251 and @xmath296 , respectively .
in this paper we have shown that \(a ) scaling dimensions are universal only when both channels propagate in the same direction , which is guaranteed by requiring the stability of the system .
\(b ) in the presence of inter - channel tunnelings , the quasiparticle tunneling picture and the electron tunneling picture give different temperature dependence of the tunneling conductance and there exists a crossover between the two pictures .
\(c ) for the ( 1 , @xmath1151/3 ) state the phase diagram is divided into two domains giving different temperature dependence of the conductance .
finally we explicitly write down our prediction for the temperature dependence of the tunneling conductance through a point contact in the presence of interactions .
first we consider the case where @xmath91 ( i ) . in the case of
@xmath293 we can read @xmath317=-4/5 $ ] , @xmath318 and @xmath319 from the tables of exponents . hence @xmath320 similarly we obtain for the ( 1/3,1/15 ) state @xmath321 and for the ( 1/3 , 1/3 ) state @xmath322 next we consider the ( 1 , @xmath1151/3 ) state which belongs to case ( ii ) : @xmath109 . as was pointed out in the last section , our prediction is different for that state .
there is an ambiguity with the quantized value of the hall conductance at `` high '' temperature when two channels propagate in the opposite directions . in this case a generalized landauer formula @xcite gives different quantized values for the two - terminal conductance @xmath323 and the four - terminal hall conductance @xmath324 , which are both different from @xmath325 @xcite .
for @xmath326 they are calculated respectively as @xcite @xmath327 however we take the viewpoint that landauer formula is no longer valid in this case and the conductance is quantized : @xmath328 , which is experimentally observed . as was studied in secs .
iii and iv the rg flow diagram in the ( @xmath329)-plane has two stable fixed points , which means that there are two different behaviors of the conductance depending on the initial value of @xmath330 .
first we study the domain ( a ) of fig . 1 where @xmath331 .
if the initial value of @xmath241 is not so large only the quasihole @xmath332 channel becomes reflected by the scattering potential as the temperature is lowered .
the other ( @xmath333 integral quantum hall ) edge is robust against backward scatterings even at zero temperature . at zero temperature
the conductance is again quantized . to obtain the quantized value , landauer formula for the two - terminal @xcite and four - terminal conductance
was applied @xcite .
similarly the two - terminal conductance can also be calculated @xcite .
both @xmath323 and @xmath324 give the same value @xmath334 as @xmath335 , which is identified with @xmath336 . at low temperature
only @xmath233 is relevant for the electron tunneling . in this case
the landauer formula is valid since the single - channel model is free from the complications on the equilibration between the channels @xcite .
the tunneling conductance behaves as @xmath337 where @xmath57 , @xmath47 and @xmath48 .
the signs of the temperature dependent terms reflect the fact that the @xmath338 edge is that of a quasihole droplet .
we predict the existence of crossover among the three pictures .
next we study the domain ( b ) of fig . 1 where @xmath339 . if the initial value of @xmath241 is sufficiently large both @xmath241 and @xmath239 scale to infinity . in this case
the scaling dimension of the instanton fugacity @xmath245 in the dig picture and that of the electron tunneling amplitude @xmath340 are identical , i.e. bith give @xmath341 .
the tunneling conductance behaves at low temperature as @xmath342 where @xmath343 is a positive exponent if @xmath344 is finite but goes to zero as @xmath306 .
we expect that the exponent is small and there is no crossover between the et and the dig picture . in this case
we also predict that even at `` high '' temperature the tunneling conductance is not quantized since this domain is not smoothly connected to the origin in the @xmath252-plane .
in summary we have shown that the structure of the hierarchical edge modes gives a variety of behaviors of the conductance through a point contact .
this may give a clue to identify the edge structure in terms of the experiment .
we are grateful to a. furusaki for useful discussions .
this work was supported by a grand - in - aid for scientific research no .
04240103 from the ministry of education , science and culture of japan .
we start with the lagrangian density @xmath345 where @xmath346 is a boson field and @xmath347 is a chern - simon gauge field .
we assume that the interaction between the bosons is short - ranged : @xmath348 where @xmath349 is a chemical potential and @xmath350 comes from the coulomb interaction .
it is clear that the action is minimized by the trivial constant solution @xmath351 with @xmath352 and @xmath353 .
using the equation of motion for @xmath347 and integrating over the density fluctuation @xmath354 , we can write ( a1 ) as @xmath355 where @xmath356 and @xmath357 . the single - valued part @xmath358 of the phase fluctuation @xmath64 is the bogoliubov mode of the bose condensate whose sound velocity is given by @xmath359 , while the multi - valued part @xmath360 is the massive vortex excitation which , we assume , is absent in the following discussions . introducing the stratonovich - hubbard variable @xmath361 , we can write ( a2 ) in the form @xmath362 where @xmath363 is the fluctuation of the boson current from @xmath364 .
integration over @xmath358 gives the continuity equation for the boson current @xmath365 which is automatically satisfied by writing @xmath363 in terms of the auxiliary gauge field @xmath366 as @xmath367 .
after integrating out the gauge field fluctuation @xmath368 , we obtain @xmath369 where @xmath370 and @xmath371 .
we consider a finite sample with both upper and lower edges .
if two edges are well separated , we can assume that they are independent , and are described respectively by that of a semi - infinite plane @xmath372 ( @xmath373 ) for the upper ( lower ) edge .
we first turn off the additional electromagnetic field @xmath374 .
the edge mode is a self - induced eigenmode of the fqh liquid which localizes near the boundary @xmath375 as @xmath376 for @xmath377 , where @xmath378 is the one - dimensional ( 1d ) current , and @xmath379 ( @xmath380 ) for the upper ( lower ) edge . in the absence of
@xmath374 we can assume @xmath381 which is consistent with the boundary condition that the current does not flow outside the sample .
note that only with that condition the action remains gauge invariant @xcite .
hence the current conservation reduces to @xmath382 which is satisfied by introducing the field @xmath383 for the upper ( lower ) edge such that @xmath384 and @xmath385 .
we substitute these current densities into the equation of motion for @xmath366 derived from ( a4 ) .
a solution is obtained when @xmath386 with @xmath387 , and @xmath388 .
the electron operator on the edges are written in terms of @xmath383 as @xmath389 which guarantees the relation between the 2d current and the phase of the electron operator : @xmath390 @xcite . putting the 2d current into ( a4 ) , we obtain a lagrangian density for a 1d boson : @xmath391 where @xmath392 .
it should be noted that the existence of the maxwell terms in the lagrangian density ( a4 ) is essential for our derivation .
if two edges are well separated , we can write the total 1d lagrangian density as a sum two chiral ones . it will be convenient to write the 1d euclidean lagrangian density as @xmath393 where @xmath394 . using the canonical quantization procedure
, we can show @xmath395=\pm\nu k/2\pi$ ] which is consistent with the anomaly equation @xmath396 @xcite .
the whole @xmath252-plane is divided into two domains separated by a boundary ( the dotted line ) , each of which has the attracive fixed point @xmath251 ( a ) and @xmath296 ( b ) , respectively .
the former ( a ) corresponds to the picture where only quasihole channel is reflected , and the latter ( b ) to the one where both channels are reflected .
|
we study the effect of backward scatterings in the tunneling at a point contact between the edges of a second level hierarchical fractional quantum hall states
. a universal scaling dimension of the tunneling conductance is obtained only when both of the edge channels propagate in the same direction .
it is shown that the quasiparticle tunneling picture and the electron tunneling picture give different scaling behaviors of the conductances , which indicates the existence of a crossover between the two pictures .
when the direction of two edge - channels are opposite , e.g. in the case of macdonald s edge construction for the @xmath0 state , the phase diagram is divided into two domains giving different temperature dependence of the conductance .
|
b0540 - 69 is a young , rotationally powered pulsar in the large magellanic cloud ( lmc ) that is similar to the crab pulsar in many ways . with a rotation frequency @xmath2 of almost 20 hz ,
it is one of the most rapidly rotating young pulsars known .
its spin - down luminosity ( @xmath3 ) of @xmath4 erg @xmath5 , where @xmath6 is the moment of inertia of the neutron star ( ns ) in units of @xmath7 g @xmath8 , is also among the largest for all pulsars .
the spin - down rate of a pulsar is often described in terms of a braking index @xmath9 , in which @xmath10 . the slowdown of b0540 - 69 is relatively stable for a young pulsar , and it is one of only 8 young pulsars for which a braking index has been reliably measured ( lyne 2015 ) .
it has a characteristic spin - down age ( @xmath11 ) of @xmath12 years .
pulsars are remarkably stable rotators , and the deviations from clocklike precision can provide information about the structure and processes at work in the neutron star and its magnetosphere .
sudden changes in @xmath2 and @xmath0 ( `` glitches '' ) are occasionally seen in pulsars , especially those with ages between @xmath13 and @xmath14 years ( yu 2013 ) .
glitches are thought to occur in rotationally powered pulsars when angular momentum is transferred from a more rapidly rotating component of the ns to the outer crust ( e.g. , anderson & itoh 1975 ; franco 2000 ) .
a different kind of rapid change is seen in some pulsars that affects the spin - down rate . in their examination of long - term monitoring of 366 pulsars
, lyne ( 2010 ) found that many pulsars rapidly switch between two different spin - down rates .
the time scales for the transitions are quasi - periodic , with typical time scales of years .
`` intermittent pulsars '' are extreme examples of such pulsars .
they transition from a radio - on state to radio - off state with a simultaneous change in the spin - down rate .
the best studied example is psr b1931 + 24 ( kramer 2006 ; young 2013 ) for which multiple transitions on a timescale of weeks from a radio bright state with @xmath15 hz s@xmath16 and a radio quiet state with @xmath17 hz s@xmath16 have been observed .
similar behavior on longer time scales has been reported for psr j1832 + 0029 ( lorimer 2012 ) and psr j1841 - 0500 ( camilo 2012 ) .
all three pulsars show rapid transitions between two states with stable spin - down rates , large differences in the spin - down rates , large changes in the radio flux , and no simultaneous change in @xmath2 . unlike most explanations for glitches , models for these state changes have emphasized changes in the pulsar s magnetosphere ( e.g. , li 2012 ) .
none of the intermittent pulsars have been detected in the optical , x - ray , or gamma - ray bands , but somewhat similar behavior has been seen for the gamma - ray pulsar psr j2021 + 4026 ( allafort 2013 ) . a 4% increase in @xmath0 was seen with a simultaneous decrease of 18% in the flux above 100 mev with a time scale for the transition of less than a week . since there is no radio counterpart , any change in radio flux is not known .
b0540 - 69 has been extensively studied since its discovery with the ( seward 1984 ) . a small glitch for b0540 - 69
was reported by zhang ( 2001 ) with a relative change in @xmath2 of @xmath18 and @xmath0 of @xmath19 .
the reality of this glitch was disputed by cusumano ( 2003 ) and later supported by livingstone ( 2005 ) .
ferdman ( 2015 ) examined 15.8 years of data from and report a second glitch with a relative change in @xmath2 of @xmath20 and @xmath0 of @xmath21 .
both these changes in @xmath0 are orders of magnitude smaller than we report in this paper .
no other glitches have been reported in the extensive monitoring of the pulsar .
optical pulsations are also seen ( mignani 2010 and references therein ) .
manchester ( 1993 ) discovered radio pulsations with an above average luminosity at 640 mhz of 1200 mjy kpc@xmath22 .
in this paper we present and discuss new timing analysis of b0540 - 69 as observed with and details of the observations and the results of the temporal analysis are reported in section 2 , and interpretations are given in section 3 .
finally , section 4 is a summary .
b0540 - 69 was observed with the proportional counter array ( pca ) on board the _ rossi x - ray timing explorer _ ( ) ( bradt 1993 ) during a 12.5-year campaign to monitor the nearby psr 0537 .
the final observation was on dec .
31 , 2011 .
we concentrate on observations covering the final 140 days of the campaign .
results from observations have been reported by zhang ( 2001 ) , livingstone ( 2005 ) and ferdman ( 2015 ) .
the pca is composed of five co - aligned xenon detectors ( proportional counter units ( pcus ) ) with a total area of @xmath23 @xmath24 .
individual pcus were routinely turned off and on to reduce the number of high - voltage breakdowns .
additional observations were made with the x - ray telescope ( xrt ) instrument ( burrows 2005 ) on the _ swift gamma - ray burst explorer _
( gehrels 2004 ) starting in feb . , 2015 .
table [ obs_log ] is a log of the observations .
data reduction for both missions followed very similar procedures .
x - ray events were screened to maximize the signal - to - noise ratio , and then photon arrival times were corrected to the solar system barycenter with the ftool faxbary for and barycorr for using the _ hubble space telescope _
position of @xmath25 , @xmath26 ( j2000.0 ) ( mignani 2010 ) .
the mid - point of each observation was chosen as the epoch , and then the best period was determined by comparing multiple folded light curves using efsearch . since the folded light curve is approximately a sine wave ( cusumano 2003 ) , the phase at the epoch was determined by fitting a sine wave to the folded light curve produced with efold and using the phase of the peak of the sine wave .
the resulting frequencies and phases were fit to the usual truncated taylor series expansion of the phase and its time derivatives @xmath2 , @xmath0 , and @xmath27 .
uncertainties quoted in this paper are given at the 90% confidence level unless otherwise noted .
for multi - parameter fits , the other parameters are allowed to vary when calculating the limits .
the pca is sensitive to x - rays in the 2 to 60 kev band with moderate ( @xmath28 ) resolution .
each event is time - tagged on the spacecraft with an accuracy better than @xmath29 s ( rots 1998 ) . to improve the signal - to - noise ratio , only events in the first xenon layer in the energy range of @xmath30 kev
were included .
figure [ joint_residuals ] shows the frequency residuals for the last 14 observations relative to the best - fit ephemeris model for the first 12 of the observations .
the value of @xmath27 is set to @xmath31 , the best - fit value for observations between oct .
2010 and dec .
the model , whose parameters are given in table [ ephemeris ] , provides a very good fit to the first 12 observations with rms phase residuals of @xmath321% , but the final two observations deviate dramatically .
a linear fit to the frequencies for the final two observations indicates that @xmath0 has changed by @xmath33 hz s@xmath16 with a statistical uncertainty of @xmath34 hz s@xmath16 at some time between the observations on dec .
3 and dec . 17 .
this is an increase of @xmath35 in the rate at which the pulsar s spin is slowing .
this value is consistent with , but less accurate than , the value determined in section 2.2 using both and data .
we note that our value for @xmath27 is @xmath36 lower than the values reported by ferdman ( 2015 ) for longer intervals of data .
reprocessing the data using the ferdman values for @xmath27 demonstrates that revising @xmath27 would have a negligible effect on our results and would not change our conclusions .
if the sudden change in @xmath0 were due to a glitch , then a simultaneous change in @xmath2 would be expected .
surveys of glitches in other pulsars find that the amplitude of a glitch in @xmath2 correlates with the amplitude of the glitch in @xmath0 for pulsars in general ( espinoza 2011 ) and for the crab pulsar in particular ( lyne 2015 ) .
the correlation for pulsars in general would indicate an accompanying glitch in b0540 - 69 larger than @xmath37 hz , and the correlation for the crab would indicate an accompanying glitch of @xmath38 hz .
there is no indication of an accompanying change in @xmath2 , and we place limits on its possible size by assuming an instantaneous glitch with no recovery and extrapolating the fit to the post - glitch frequencies back toward the final pre - glitch observation . with these assumptions the largest glitch would occur immediately after the final pre - glitch observation with a size of @xmath39 hz .
later glitches would be smaller until at 200 ks after final pre - glitch observation there would be no glitch in @xmath2 . large glitches in @xmath0
typically decay away on time scales from about a day to weeks .
since the new value for @xmath0 is consistent with that later seen with ( section [ xrt ] ) , there is also no indication of such a recovery .
we conclude that the sudden change in @xmath0 is almost certainly a state transition rather than a glitch .
we will refer to the state before the transition as the `` low spin - down '' state and the state after the transition as the `` high spin - down '' state .
there is also no indication of a change in the pulsed x - ray emission from b0540 - 69 .
figure [ pulsed_rate ] shows the average pulsed count rate ( @xmath40 , where a is the relative sine wave amplitude , c is the total number of counts detected , @xmath41 is the exposure time , and @xmath42 is the average number of pcus that were on during the observation ) . the average pulsed count rate for the two observations after the transition ( @xmath43 ) is consistent with the average for the first 12 observations ( @xmath44 ) .
further the folded light curves for these two observations show no significant change . to confirm the continued nominal performance of the pca including the final two observations , we processed the same observations for psr 0537 following the procedures given in marshall ( 2004 ) .
after the glitch near mjd 55815 , the final 12 observations are very well fit using the standard form for the ephemeris with an rms phase residual of @xmath45 .
we conclude that the unexpected values of @xmath2 for b0540 - 69 are not due to any instrumental effect .
the x - ray telescope started observations of b0540 - 69 on feb .
17 , 2015 , as part of a target of opportunity campaign .
the xrt is a focusing x - ray telescope with a ccd detector with an effective bandpass of 0.3 - 10 kev .
all observations were made using the window timing ( wt ) mode in which the central 200 ccd columns are continously read out , providing a time resolution of 1.7 ms .
the xrt data were reduced with the standard software ( xrtpipeline v0.13.1 ) applying the default filtering and screening criteria ( heasoft 6.16 ) , using the 20140709 update to the xrt caldb files and the 20150428 update to the clock correction file .
the phase and frequency were measured for each continuous viewing interval ( a `` snapshot '' ) .
each of the entries in table [ obs_log ] except the first consists of two snapshots separated by about 96 minutes ( the duration of a
the first entry has a single snapshot .
we obtained the best ephemeris by determining a model for the closely spaced observations and then refining the model as more widely separated observations were added .
all the models assumed a value of @xmath31 for @xmath27 based on measurements with before the state transition in 2011 .
this parameter makes a very small contribution to the model because the duration of the current
campaign is only 65 days .
the best - fit model was confirmed using a large grid search of possible combinations of @xmath2 and @xmath0 .
the parameters of the best - fit model and the 90% confidence uncertainties are given in table [ ephemeris ] .
the model provides a good fit to the data with rms phase residuals of @xmath323% .
figure [ joint_residuals ] , which shows the frequency residuals relative to the ephemeris before the transition , displays the overall history of @xmath2 .
we used two methods to determine @xmath0 after the state transition .
the first method uses the best - fit model for the xrt data ( table [ ephemeris ] ) with a value of @xmath46 at the xrt epoch . after adjusting by @xmath47 to account for the effect of @xmath27 , this is a change in @xmath0 from the value before the transition ( table [ ephemeris ] ) of @xmath48 .
the second method compares the average value of @xmath2 in the final two observations ( @xmath49 hz @xmath50 hz at mjd 55919.71 ) with the best - fit value for the xrt data of @xmath51 hz @xmath52 hz . the resulting @xmath0 is @xmath53 s@xmath54 at mjd 56498.62 . after adjusting by @xmath55 to account for the effect of @xmath27 ,
the result is @xmath56 s@xmath54 at the xrt epoch , which is larger than result from the first method by @xmath57 s@xmath54 .
this discrepancy is shown in figure [ joint_residuals ] in which the extrapolation of the xrt ephemeris back to the final two observations predicts a value for @xmath2 that is @xmath58 larger than that observed .
the uncertainty of the extrapolated values due to the uncertainty in the model s @xmath0 is only @xmath59 hz .
the discrepancy suggests that the evolution of the high spin - down state is more complicated than our simple model .
the second method indicates a change in @xmath0 from the value before the transition of @xmath60 .
since the value of @xmath27 after the state transition is uncertain and the statistical uncertainties are small , we use the value of the adjustments due to @xmath27 to estimate the uncertainty .
the relative change in @xmath0 is then @xmath61 and @xmath62 for the first and second methods respectively .
we adopt @xmath63 , the average of the two methods with an uncertainty that encompasses the individual error bars , as our best estimate of the relative change in @xmath0 .
the folded light curve is similar to that reported from other x - ray observations ( e.g. , cusumano 2003 ) . although the actual light curve is more complicated , a model of a constant plus a gaussian provides a good description of the xrt data .
the best - fit standard deviation for the gaussian is @xmath64 , which is consistent with a fit to the folded light curve for the data before the state transition .
b0540 - 69 has been extensively monitored by numerous observatories in the more than 30 years since its discovery including _ ginga _ observations spanning 4.4 years ( deeter 1999 ) and the 12.5 years of observations with .
more than 10 measurements of the braking index for the pulsar , which require long spans of data , have been reported ( cusumano 2003 , who list previous measurements ; ferdman 2015 ) .
values for the braking index range from 1.81 ( zhang 2001 ) to 2.74 ( gelman & hasinger 1990 ) .
none of these studies found a sudden change in @xmath0 comparable to the one reported here .
the glitches in @xmath0 reported by zhang ( 2001 ) and ferdman ( 2015 ) are @xmath322000 times smaller , and knowledge of the pulsar s phase is maintained throughout the 12.5 years of monitoring until the final two observations .
when discovered with the ( seward 1984 ) , b0540 - 69 had a @xmath0 of @xmath65 hz s@xmath16 , which indicates that the pulsar was in the low spin - down state in 1979 and 1980 .
the lack of published values of @xmath0 indicative of the high spin - down state suggests that b0540 - 69 remained in the low spin - down state the vast majority of the time , if not all the time , until the transition in late 2011 .
the sudden change in the spin - down rate is most likely associated with a change in the magnetosphere of the ns .
one possibility is a global change in the conductivity of the magnetosphere .
li ( 2012 ) developed solutions for pulsar magnetospheres with finite resistivity that mitigate some of the limitations of the vacuum and force - free models .
the dependence of the spin - down luminosity was calculated for a wide range of the conductivity parameter @xmath66 ( expressed in terms of the pulsar angular rotation rate @xmath67 ) and inclination angle @xmath68 using a three - dimensional numerical code . for broad ranges of @xmath68 ( 0@xmath69 to 90@xmath70 and the conductivity ( @xmath71 from 0.04 to 4.0 ) , a 36% increase in the luminosity ( and by extension in @xmath72 ) can be achieved with an increase in @xmath66 of less than a factor of 6 .
there are maximum values of @xmath71 , ranging from 40 at @xmath73 to 1.3 at @xmath74 above which it is no longer possible to explain an increase in the luminosity with an increase in @xmath66 . since for large values of @xmath75 , the component of the electric field parallel to the magnetic field @xmath76 ( kalapotharakos 2014 ) , an increase in @xmath66 may reduce the gamma - ray flux from the pulsar .
the long - term light curve of b0540 - 69 in the energy range of 200 mev to 100 gev ( fermi - lat collaboration 2015 ) shows no indication of change in the gamma - ray flux with an upper limit of 30% .
detailed modelling , such as that done by brambilla ( 2015 ) for other fermi pulsars , is needed to understand the implication of this constraint on parameters for b0540 - 69 , but this is beyond the scope of this paper .
another possible explanation for the spin - down transition is a change in torque due to a change in the plasma outflow from the polar caps .
kramer ( 2006 ) suggested this to explain the state transitions in the intermittent pulsar psr b1931 + 24 .
unlike psr b1931 + 24 , b0540 - 69 appears to be radio bright in its low spin - down state , but its radio flux may significantly increase or its pulse shape may change in the high spin - down state . following young
we estimate the change in the charge density @xmath77 as @xmath78 c m@xmath79 , assuming a pulsar radius of 10 km and i@xmath80 of @xmath81 .
this estimate of 0.78 c m@xmath79 is close to the goldreich - julian charge density for b0540 - 69 of 1.09 c m@xmath79 .
an increase in @xmath77 predicts an increase in the radio flux , which can be tested with radio observations of the high spin - down state .
it also predicts a reduction in the braking index as the relative importance of magnetic dipole radiation decreases . a change in the amount of open poloidal magnetic flux would also change the spin - down luminosity , which is expected to be proportional to @xmath82 where @xmath83 is the radius beyond which the magnetic field lines become open ( contopoulos 2007 ) .
an 36% increase in the spin - down luminosity with no change in @xmath2 would require a 15% decrease in @xmath83 .
the spin - down rate of b0540 - 69 increased suddenly by 36% ( a change in @xmath0 of @xmath84 hz s@xmath16 ) between dec .
3 and dec . 17 , 2011 .
observations with in 2015 show that this change in spin - down rate has persisted for more than 3 years .
such a large persistent change in the spin - down rate has never been reported for b0540 - 69 .
there is no indication of a simultaneous change in the pulse shape in x - rays , the pulsed x - ray luminosity , or the total gamma - ray luminosity .
we interpret this change as a transition between two stable states of the pulsar , similar to state transitions in other pulsars .
b0540 - 69 extends the class of state changing pulsars to include a very young and luminous example .
the transitions in b0540 - 69 appear to be rare with the low spin - down state probably lasting more than 30 years .
future work can test predicted changes in the radio flux and the braking index in the high spin - down state .
we gratefully acknowledge the anonymous referee for suggestions that improved the paper .
this work made use of data supplied by the uk swift science data centre at the university of leicester and the high energy astrophysics science archive research center , provided by nasa s goddard space flight center .
allafort , a. , 2013 , , 777 , l2 anderson , p. w. , & itoh , n. 1975 , nature , 256 , 25 bradt , h. v. , rothschild , r. e. , & swank , j. h. 1993 , a&as , 97 , 355 brambilla , g. , 2015 , , 804 , 84 burrows , d. n. , 2005 , space science reviews , 120 , 165 camilo , f. , 2012 , , 746 , 63 contopoulos , i. , 2007 , a&a , 466 , 301 cusumano , g. , 2003 , a&a , 402 , 647 deeter , j. e. , nagase , f. & boynton , p. e. 1999 , , 512 , 300 espinoza , c. m. , 2011 , , 414 , 1679 ferdman , r. d. , 2015 , , submitted fermi - lat collaboration , 2015 , in prep .
franco , l. , link , b. , & epstein , r. i. 2000 , , 543 , 987 gehrels , n. , 2004 , , 611 , 1005 kalapotharakos , c. , harding , a. & kazanas , d. 2014 , , 793 , 97 kramer , m. , 2006 , science , 312 , 549 li , j. , spitkovsky , a. & tchekhovskoy , a. 2012 , , 746 , 60 link , b. , & epstein , r. 1997 , , 478 , l91 livingstone , m. a. , 2005 , , 633 , 1100 lorimer , d. r. , 2012 , , 758 , 141 lyne , a. g. 2010 , science , 329 , 408 lyne , a. g. 2015 , , 446 , 857 manchester , r. n. , 1993 , , 403 , l29 marshall , f. e. , 2004 , , 603 , 682 mignani , r. p. , 2010 ,
a&a , 515 , 110 gelman , h. & hasinger , g. 1990 , , 353 , l21 rots , a. h. , 1998 , , 501 , 749 seward , f. d. , harnden , f. r. & helfand , d. j. 1984 , , 287 , l19 young , n. j. , 2013 , , 429 , 2569 yu , m. , 2013 , , 429 , 688 zhang , w. , 2001 , , 554 , l177 llccr & 96023 - 01 - 17 & 2011 aug 13 & @xmath85 & @xmath86 + & 96023 - 01 - 42 & 2011 sep 06 & @xmath87 & @xmath88 + & 96023 - 01 - 43 & 2011 sep 14 & @xmath89 & @xmath90 + & 96023 - 01 - 44 & 2011 sep 18 & @xmath91 & @xmath92 + & 96023 - 01 - 20 & 2011 sep 23 & @xmath93 & @xmath92 + & 96023 - 01 - 45 & 2011 sep 26 & @xmath94 & @xmath95 + & 96023 - 01 - 46 & 2011 sep 26 & @xmath94 & @xmath96 + & 96023 - 01 - 21 & 2011 oct 09 & @xmath97 & @xmath98 + & 96023 - 01 - 22 & 2011 oct 22 & @xmath99 & @xmath92 + & 96023 - 01 - 23 & 2011 nov 04 & @xmath100 & @xmath92 + & 96023 - 01 - 24 & 2011 nov 18 & @xmath101 & @xmath88 + & 96023 - 01 - 25 & 2011 dec 03 & @xmath102 & @xmath86 + & 96023 - 01 - 26 & 2011 dec 17 & @xmath103 & @xmath104 + & 96023 - 01 - 19 & 2011 dec 31 & @xmath105 & @xmath106 + & 00033603002 & 2015 feb 17 & @xmath107 & @xmath108 + & 00033603004 & 2015 feb 23 & @xmath109 & @xmath110 + & 00033603005 & 2015 feb 25 & @xmath111 & @xmath112 + & 00033603006 & 2015
feb 25 & @xmath111 & @xmath113 + & 00033603007 & 2015 mar 11 & @xmath114 & @xmath110 + & 00033603008 & 2015 apr 11 & @xmath115 & @xmath116 + & 00033603009 & 2015 apr 13 & @xmath117 & @xmath116 + & 00033603010 & 2015 apr 23 & @xmath118 & @xmath119 + lcc epoch ( modified julian date ) & 55892.352587464 & 57077.537772223 + phase & 0.000 ( 15 ) & 0.000 ( 23 ) + @xmath2 ( hz ) & 19.72655182 ( 1 ) & 19.70077383 ( 4 ) + @xmath0 ( @xmath120 ) & -1.86322 ( 2 ) & -2.52871 ( 8) + @xmath27 ( @xmath121 ) & 3.249 ( fixed ) & 3.249 ( fixed ) +
|
we report the discovery of a large , sudden , and persistent increase in the spin - down rate of b0540 - 69 , a young pulsar in the large magellanic cloud , using observations from the and satellites .
the relative increase in the spin - down rate @xmath0 of @xmath1 is unprecedented for b0540 - 69 .
no accompanying change in the spin rate is seen , and no change is seen in the pulsed x - ray emission from b0540 - 69 following the change in the spin - down rate .
such large relative changes in the spin - down rate are seen in the recently discovered class of `` intermittent pulsars '' , and we compare the properties of b0540 - 69 to such pulsars .
we consider possible changes in the magnetosphere of the pulsar that could cause such a large change in the spin - down rate .
|
baados , silk and west pointed out that a rapidly rotating black hole may act as a particle accelerator to arbitrarily high energy @xcite .
we call this acceleration mechanism the baados - silk - west ( bsw ) mechanism or bsw process .
there are several discussions on the bsw process @xcite . from a purely relativistic point of view
, it is suggested in @xcite that an arbitrarily high energy collision induced by the bsw process implies that an extremal kerr black hole is linearly unstable .
if we consider the free - fall test particles with an arbitrarily small mass from the far region , we expect that the gravity induced by those test particles will be well described by a linear perturbation around the background spacetime .
if one of the test particles has the fine - tuned angular momentum for given energy such that the bsw process works well , an arbitrarily high energy particle collision can occur near the horizon . in principle , the center of mass ( cm ) energy can be as large as the mass energy of the background black hole , where such intense gravity can not be described by a linear perturbation around the background black hole .
thus , it seems that we can construct a perturbation which can be described by a linear perturbation around an extremal kerr black hole initially , but it can not be after time evolution , so that an extremal kerr black hole is linearly unstable .
recently , it is shown that the second - order derivative of a test field on the horizon of an extremal black hole blows up in late times if the test field has a nonzero value on the horizon as an initial condition @xcite .
unlike in the case of a nonextremal black hole , the test field can stay on the horizon for a long time and grow up . in this sense
, an extremal black hole has an instability on the horizon .
we would like to point out that there are some analogies between these two instabilities , both of which imply that the extremal black hole horizon is linearly unstable .
first , it seems that both phenomena come from the vanishing surface gravity of the extremal horizon . in this case ,
an observer on the horizon does not feel gravitational force , and hence a particle or a field can stay in the vicinity of the horizon .
note that it is essential for the bsw process that there exists a geodesic particle which stays in the vicinity of the horizon .
next , a wave with sufficiently short wavelength behaves like a test particle .
so , there should be a natural analogy between the dynamics of test particles and test fields .
therefore , we suggest that there is a close relationship between the bsw process and the instability of fields on extremal black holes .
according to the previous works @xcite , for an arbitrarily high energy particle collision to occur by the bsw process , we need to impose at least two conditions : ( i ) the black hole horizon is extremal and ( ii ) the parameters of either of the two colliding particles are fine - tuned .
we can derive these two conditions by imposing that a particle can asymptote to the event horizon in infinite proper time .
the condition ( ii ) is known as the critical particle condition which is a relation among energy , angular momentum and electric charge of the particle @xcite . since both the radial velocity and the acceleration of such a particle asymptote to zero
as it approaches the horizon , i.e. , @xmath0 where @xmath1 is the proper time of the particle and @xmath2 is the radius of the horizon , we might regard this condition as `` force balance '' conditions in an approach to the horizon .
it is proposed that the critical particle condition can be naturally realized if we consider an innermost stable circular orbit ( isco ) particle . in the case of a near extremal kerr black hole ,
the radius of the isco is close to the horizon and such a particle satisfies @xmath3 and @xmath4 in the vicinity of the horizon which is similar to the conditions ( [ eq : forcebalance1 ] ) and ( [ eq : forcebalance2 ] ) .
if we consider a collision between an isco particle and another free - fall particle around an extremal kerr black hole , an arbitrarily high energy particle collision can also occur in the extremal limit as shown in @xcite .
we can expect that there is a relation between the approach of the isco to the horizon and the bsw process . in this paper , to understand the nature of the bsw process deeper , we study high energy particle collision around higher dimensional extremal black holes .
the effective potential for the radial coordinate @xmath5 roughly has the form of @xmath6 , where @xmath7 is the mass of the black hole and @xmath8 is the conserved angular momentum for a particle in a @xmath9-dimensional spacetime .
the first term is the gravitational potential and the second term is the potential of centrifugal force . in higher dimensions @xmath10 ,
the power of gravitational force @xmath11 is not greater than that of centrifugal force @xmath12 . in that case , the effective potential does not have a local minimum ; namely , there is no stable circular orbit in contrast with four dimensions .
if the existence of isco is important for the bsw process , we can expect that high energy particle collision can not occur around higher dimensional extremal black holes . on the other hand , recently , it was shown that there are linear instabilities of test fields on the horizon of higher dimensional extremal black holes @xcite .
if there exists an analogy between instabilities of the test field and the bsw process , we can expect that a high energy particle collision can also occur in higher dimensions .
there is also a discussion for the case of charged particles around an extremal reissner - nordstrm black hole , which can be considered as a similar system for a free - fall test particle around an extremal kerr black hole @xcite . in this case
, we can understand the critical particle condition for a radially moving charged particle as the force balance between gravitational force and coulomb force . in the case of higher dimensions , we can expect high energy collision between two charged particles can also occur around an extremal charged black hole .
this is because gravitational force and coulomb force have the same power law of radial distance in higher dimensional spacetimes , and the situation is the same as the four - dimensional case .
this paper is organized as follows . in sec .
[ secii ] , we review the motion of a test particle around myers - perry black hole spacetimes and investigate a cm energy for the colliding two particles . we show that the collision with an arbitrarily high cm energy can occur in the case of higher dimensions . in secs .
[ seciii ] and [ seciv ] , we discuss whether critical particles can reach the event horizon by investigating effective potentials of test particles . as a typical case , we also investigate the cases of five and six dimensions in detail . in sec .
[ secv ] , we study the case of @xmath9-dimensional charged black holes .
section [ secvi ] is devoted to the summary and discussion . in this paper
we use the units in which the light speed @xmath13 .
in this section , first , we review the motion of a test particle in the @xmath9-dimensional myers - perry black hole spacetime @xcite .
next , we investigate the collision of two particles falling into the black hole and derive a cm energy of them . we describe the myers - perry spacetime in @xmath14 dimensions in the boyer - lindquist coordinates with the spin parameters @xmath15 . here
@xmath17 where @xmath18 is the floor function or the greatest integer that is less than or equal to @xmath19 . for even dimensions ,
one of the spin parameter @xmath20 should vanish because of an extra unpaired spatial coordinate .
we will adopt the convention that @xmath21 vanishes for even dimensions .
we can assume @xmath22 for @xmath23 without loss of generality . in this paper , for simplicity , we focus on the case where the spin parameters @xmath24 take only one value @xmath25 or take only two different values @xmath25 and @xmath26 .
we adopt the convention @xmath25 for @xmath27 and @xmath28 for @xmath29 , where @xmath30 .
we will assume that the parameter @xmath31 does not vanish . under the assumptions and convention
, we can calculate the hamilton - jacobi equation and a cm energy of colliding two particles around a myers - perry black hole in a unified way regardless of the parity of the spacetime dimensions . in the even dimensional case
, the black hole has @xmath32 equal angular momenta if we set @xmath33 ( or @xmath34 ) because of the spin parameter @xmath35 . in the odd dimensional case ,
the black hole has @xmath36 equal angular momenta if we set @xmath37 . the line element in the higher dimensional myers - perry black hole is explicitly expressed by the following form : @xmath38^{2}\nonumber\\ & & + ( r^{2}+a^{2})\sin^{2}\theta \sum^{q-1}_{i=1}\left ( \prod^{i-1}_{k=1 } \sin^{2}\alpha_{k } \right ) d\alpha_{i}^{2 } + ( r^{2}+b^{2})\cos^{2}\theta \sum^{p-1}_{j=1}\left ( \prod^{j-1}_{k=1 } \sin^{2}\beta_{k } \right ) d\beta_{j}^{2}\nonumber\\ & & + ( r^{2}+a^{2})\sin^{2}\theta \sum^{q}_{i=1 } \left ( \prod^{q - i}_{k=1 } \sin^{2}\alpha_{k } \right ) \cos^{2}\alpha_{q - i+1 } d\phi_{i}^{2}\nonumber\\ & & + ( r^{2}+b^{2})\cos^{2}\theta \sum^{p-\epsilon}_{j=1 } \left ( \prod^{p - j}_{k=1 } \sin^{2}\beta_{k } \right ) \cos^{2}\beta_{p - j+1 } d\phi_{j+q}^{2},\qquad\quad \end{aligned}\ ] ] where @xmath39 is @xmath40 and @xmath41 for even and odd @xmath9 dimensions , respectively
, @xmath42 is the azimuthal angular coordinates which take values in the range @xmath43 , @xmath5 is the radial coordinate , @xmath44 is the temporal coordinate , @xmath45 is the angular coordinate which takes a value in the range @xmath46 , @xmath47 and @xmath48 are two sets of spherical polar coordinates , @xmath7 is the mass of the black hole and @xmath49 and where @xmath50 and @xmath51 for any function @xmath52 .
if the equation @xmath53 has at least one positive root , there exists an event horizon at @xmath54 , where @xmath2 is the largest root . for later convenience , we define the functions @xmath55 , @xmath56 and @xmath57 as @xmath58 and @xmath59 respectively .
we note that the function @xmath60 becomes zero at the event horizon @xmath54 , i.e. @xmath61 these functions satisfy the following equations : @xmath62 see appendix a for more details on the line element .
we discuss the motion of a test particle with the rest mass @xmath63 in the @xmath9-dimensional myers - perry black hole spacetime by using the hamilton - jacobi method .
we define the action @xmath64 as a function of the coordinates @xmath65 and the parameter @xmath66 for massive particle , where @xmath1 correspond to the proper time .
the hamilton - jacobi equation is given by @xmath67 where @xmath68 is the hamiltonian of a particle and @xmath69 is the conjugate momentum given by @xmath70 we can write @xmath71 in the following form of separation of variables because of the cyclic coordinates @xmath44 and @xmath72 : @xmath73 where @xmath74 , @xmath75 for @xmath76 and @xmath77 for @xmath78 are constants and @xmath79 and @xmath80 are functions of the coordinates @xmath5 and @xmath45 , respectively .
for even dimensions , one of the conserved angular momenta @xmath8 and @xmath81 should vanish because of an extra unpaired spatial coordinate .
here we have adopted the convention that @xmath82 vanishes for even dimensions .
the constant @xmath83 corresponds to the energy of the particle and @xmath84 and @xmath85 correspond to the angular momenta of the particle .
we can separate the hamilton - jacobi equation ( [ eq : s5_hamilton_explitcit ] ) , which is given in appendix b , for @xmath5 and @xmath45 as @xmath86 and @xmath87 respectively , where @xmath88 , @xmath89 and @xmath90 are separation constants and we have defined the function @xmath91 as @xmath92 see eqs .
( [ eq : s5_j1 ] ) and ( [ eq : s5_l1 ] ) for the definitions of @xmath89 and @xmath90 , respectively , in appendix b. from these equations , we obtain @xmath93 with @xmath94 @xmath95,\end{aligned}\ ] ] where @xmath96 and @xmath97 .
note that we can choose the values of @xmath98 and @xmath99 independently and @xmath100 and @xmath101 must be satisfied for allowed particle motions . from eq .
( [ eq : s5_theta ] ) , the condition for @xmath88 is written as @xmath102 from the forward - in - time condition @xmath103 , we can see the condition @xmath104 must be satisfied outside the event horizon . by taking the limit @xmath105 ,
this condition yields @xmath106 equation ( [ eq : drdlambda ] ) can be transformed to the familiar form as the energy equation , which is given by @xmath107 where @xmath108 is the effective potential for a particle , which is defined as @xmath109 with @xmath110 .
\quad \end{aligned}\ ] ] we obtain @xmath111 at large distances @xmath112 .
therefore , a particle with @xmath113 is bound by the black hole or prohibited .
we will discuss the effective potential in odd and even dimensions in detail in secs .
iii and vi , respectively .
we define a critical particle as a particle which satisfies the condition @xmath114 in fact , this condition is obtained by solving the condition @xmath115 namely , the radial velocity becomes zero at the horizon . for @xmath116 or @xmath37 , the function @xmath91 in eq .
( [ mathcale ] ) for a critical particle can be written in a simple form as @xmath117 we consider the collision of particles @xmath118 and @xmath119 whose rest masses and momenta are given by @xmath120 and @xmath121 and @xmath122 and @xmath123 , respectively .
the cm energy @xmath124 for the collision of the two particles at a collision point @xmath5 is obtained by @xmath125 where the subscripts @xmath118 and @xmath119 denote the quantities for the particles @xmath118 and @xmath119 , respectively , and where @xmath126 and @xmath127 are independent ; see eqs .
( [ eq : s5_an])-([eq : s5_bp-1 ] ) in appendix b for the definition of @xmath128 and @xmath129 .
we can see that @xmath124 takes a finite value outside the event horizon for colliding particles with finite conserved quantities .
there is a possibility that the center of mass energy @xmath124 can be arbitrarily large only in the limit where the collision point is arbitrarily close to the horizon . on the event horizon
, @xmath130 must be satisfied .
thus , we assume @xmath130 . using lhospital s rule with respect to @xmath131 in the near horizon limit @xmath132
, we obtain @xmath133 where @xmath134 where @xmath135 , @xmath136 and @xmath137 are @xmath88 , @xmath138 and @xmath139 for the particle @xmath140 or @xmath119 . here
we have used eq .
( [ eq : s5_forward_time ] ) .
thus , the cm energy for the collision of the two particles in the near horizon limit @xmath132 is obtained by @xmath141 in the case of the black hole , this shows that a collision with an arbitrarily high cm energy can occur if and only if one of the two particles is critical and the other is noncritical . for the prospect of collisions with high cm energy , it is important to study whether critical particles can reach the event horizon .
thus , we will discuss the effective potentials for critical particles in secs .
iii and iv .
in this section , we will investigate the effective potentials for the critical particles and test whether critical particles can reach the event horizon in the odd dimensional myers - perry black hole spacetime with equal angular momenta @xmath37 . in this case , the effective potential for critical particles is given by @xmath142 where @xmath143 r^{2}\delta(r ) -2m(r^{2}-r_{+}^{2})^{2}e^{2 } , \quad \quad \label{todd } \\
\delta & = & \frac{1}{r^2 } ( r^2 + a^2)^n - 2m.\end{aligned}\ ] ] on the event horizon @xmath144 , the effective potential vanishes , @xmath145 because of the condition for the event horizon @xmath146 .
for the critical particle to reach the event horizon from @xmath147 , the first derivative of the effective potential @xmath148 should be zero or negative .
the function @xmath149 at the horizon becomes @xmath150 r_+^{2 } \delta^\prime ( r_+).\end{aligned}\ ] ] since @xmath55 takes a positive value outside the horizon , @xmath151 should be zero or positive . in the case of @xmath152 , @xmath153 must be satisfied .
this case corresponds to the so - called multiple scattering scenario @xcite . in this paper
, we mainly focus on the case @xmath154 . in this case , the function @xmath155 becomes zero at the horizon , i.e. , @xmath156 and a particle asymptotes to the event horizon in infinite proper time . from eq .
( [ eq : s5_extremal_odd_dimension ] ) , we obtain @xmath157 we can see that the myers - perry black hole must have an extremal rotation .
note that we can obtain the same conclusion for the case of even dimensions in a way similar to the above discussion . from eqs .
( [ eq : s5_horizon_odd_dimension ] ) and ( [ eq : s5_extremal_odd_dimension ] ) , we can see that both the radial velocity and the acceleration of the test particle vanish at the horizon , i.e. , @xmath158 and @xmath159 this implies that the critical particle satisfies the force balance conditions ( [ eq : forcebalance1 ] ) and ( [ eq : forcebalance2 ] ) in the case of the extremal horizon .
it is useful to classify the effective potential with the sign of the second derivative @xmath160 and @xmath161 , where @xmath162 .
\end{aligned}\ ] ] in the case @xmath163 and @xmath164 , @xmath165 , and hence @xmath166 for a critical particle should be negative at least in the vicinity of the horizon ; the critical particle is allowed to exist at large distances from the black hole and we refer to this class as class ia .
this class is most interesting because critical particles can reach from the infinity to the event horizon if @xmath165 is negative everywhere outside the event horizon .
these critical particles can collide with a noncritical particle near the event horizon with an arbitrarily high cm energy . in the case
@xmath163 and @xmath167 , @xmath165 , and hence @xmath166 for a critical particle should be negative at least in the vicinity of the horizon ; the critical particle is always bounded by the black hole and we refer to this class as class ib . in classes ia and ib , from the inequality ( [ eq : s5_k_min ] ) and @xmath168 , @xmath88 should satisfy the condition @xmath169 the case @xmath170 is the marginal case , and we refer to this class as class ii . in four dimensions ( @xmath171 ) , this case corresponds to the last - stable - orbit collision @xcite .
suppose @xmath172 .
then , the critical particle can not reach the horizon without multiple scattering because of the potential barrier and we refer to this class as class iii . for the critical particle to reach the horizon from the far region ,
the effective potential should satisfy @xmath173 outside the horizon .
though it is difficult to show it in general , we can prove it in the case of the massless particle or a highly relativistic particle with @xmath174 . since the function @xmath175 is an increasing function of @xmath88 as seen in eq .
( [ todd ] ) , we only have to show @xmath176 for the maximum value of @xmath177 in eq .
( [ krange1 ] ) . in this case
, @xmath175 becomes @xmath178 as shown in appendix [ vnegativeodd ] the function @xmath179 takes a negative value outside the horizon @xmath147 . for simplicity
, we consider the case where the critical particle has only one nonvanishing angular momentum .
that is , we assume @xmath180 where @xmath181 , @xmath182 , , @xmath183 , @xmath184 , @xmath182 , , @xmath185 and @xmath186 , @xmath182 , , @xmath187 . see appendix b for the definitions of @xmath128 , @xmath129 , @xmath188 and @xmath189 . from eqs .
( [ eq : s5_jq ] ) , ( [ eq : s5_lp ] ) , ( [ eq : s5_an])-([eq : s5_bp-1 ] ) and ( [ eq : s5_one_conserved_angular_momentum ] ) , we obtain @xmath190 and @xmath191 thus , the critical particle must satisfy @xmath192 the above results ( [ eq : j1_one_conserved_angular_momentum])-([eq : critical_one_conserved_angular_momentum ] ) are not only true in odd dimensions but also in even dimensions under the assumption ( [ eq : s5_one_conserved_angular_momentum ] ) . under the assumption ( [ eq : s5_one_conserved_angular_momentum ] ) , from the inequality ( [ eq : s5_k_inequality_all_dimension ] ) , we can find @xmath88 which satisfies the inequality ( [ eq : s5_k_inequality_all_dimension ] ) if the following inequality is satisfied : @xmath193 therefore , we can find @xmath88 if @xmath194 satisfies @xmath195 and @xmath45 satisfies @xmath196 on the equatorial plane @xmath197 , we will find @xmath88 in all odd dimensions if @xmath194 satisfies @xmath198 .
this shows that the particle must be massless or highly relativistic for @xmath88 to exist in the large dimension limit @xmath199 . as the simplest example of odd dimensional cases
, we will consider the motion of the critical particle in the extremal five - dimensional myers - perry black hole spacetime ( @xmath200 ) with equal angular momenta ( @xmath37 ) @xcite and investigate the bsw effect there .
this case is given by setting @xmath201 and @xmath202 .
the explicit form of the line element of the myers - perry black hole spacetime is given in appendix c. the event horizon is obtained by @xmath203 from eqs .
( [ eq : s5_critical ] ) and ( [ eq : s5_5dim_horizon_extremal_degenerate ] ) and @xmath204 , we get @xmath205 from eqs .
( [ eq : alpha_q_beta_p ] ) , ( [ eq : s5_5dim_critical_extremal_degenerate ] ) , ( [ eq : s5_jq ] ) and ( [ eq : s5_lp ] ) and @xmath204 , we obtain @xmath206 and @xmath207 the effective potential of the critical particle is given by @xmath208 where @xmath209.\end{aligned}\ ] ] figure 1 shows the examples of the effective potential @xmath166 . in fig .
1 , the curves of classes ia , ib , ii and iii correspond to the cases @xmath163 and @xmath164 , @xmath163 and @xmath167 , @xmath170 and @xmath163 , respectively . for a critical particle in the five - dimensional and extremal myers - perry black hole spacetime with equal angular momenta .
the solid ( red ) , dashed ( green ) , dot - spaced ( blue ) and dotted ( purple ) curves denote the effective potential for the critical particle in classes ia ( @xmath210 , @xmath211 ) , ib ( @xmath210 , @xmath212 ) , ii ( @xmath213 , @xmath214 , @xmath211 ) and iii ( @xmath213 , @xmath215 , @xmath211 ) , respectively . ,
width=328 ] we consider classes ia and ib and therefore assume @xmath216 , where @xmath217.\end{aligned}\ ] ] for five dimensions , the inequality ( [ eq : s5_k_inequality_all_dimension ] ) is described as @xmath218 in the case @xmath219 with @xmath220 or @xmath221 with @xmath222 , the inequality ( [ eq : s6_k_inequality ] ) is satisfied and we can find @xmath88 which has a finite value if and only if @xmath223 . on the other hand , in the case @xmath219 with @xmath224 or @xmath221 with @xmath225 , we can not get @xmath88 .
next , we consider the case where @xmath226 is neither @xmath41 nor @xmath40 . in this case , @xmath88 which satisfies the inequality ( [ eq : s6_k_inequality ] ) exists if the following quadratic inequality for @xmath226 is satisfied : @xmath227 where @xmath228 .
the inequality ( [ eq : s6_sin_square_inequality ] ) is expressed by @xmath229 for @xmath167 , we easily see that the inequality ( [ eq : s6_sin_square_inequality2 ] ) is not satisfied . for @xmath230 , the inequality ( [ eq : s6_sin_square_inequality ] )
has the solution @xmath45 satisfying @xmath231 , where @xmath232}}{2(3-\alpha ) } \nonumber\\\end{aligned}\ ] ] and the range of @xmath233 is given by @xmath234 we plot @xmath235 in fig .
2 . versus @xmath228 for a critical particle in the five - dimensional and extremal myers - perry black hole spacetime with equal angular momenta .
the solid ( red ) , dashed ( green ) , dot - spaced ( blue ) , dotted ( purple ) and dash - dotted ( cyan ) curves denote @xmath236 and @xmath237 for @xmath238 , @xmath236 and @xmath237 for @xmath239 and @xmath235 for @xmath240 , respectively . in the region @xmath241 including @xmath242 and @xmath243 , collisions with arbitrarily high cm energy occur .
, width=328 ] therefore , we can find @xmath88 in the region @xmath231 for @xmath230 and @xmath244 .
collisions with an arbitrarily high cm energy occur within this region .
in this section we discuss the effective potential for a critical particle in the even dimensional and extremal myers - perry black hole spacetime with the equal angular momenta @xmath25 for @xmath245 and @xmath246 . in this case
we have to set @xmath34 and @xmath33 .
notice @xmath247 in the even dimensional case because of the convention @xmath248 and eq .
( [ eq : s5_lp ] ) .
the effective potential for a critical particle is given by @xmath249 where @xmath250 r\delta -2mr\frac{(r^{2}-r_{+}^{2})^{2}}{r^{2}+a^{2}}e^{2 } \label{tevendim}\end{aligned}\ ] ] in the same way as for odd dimensions , we can show that the black hole must have the extremal rotation so that critical particles reach the event horizon , where the force balance conditions ( [ eq : forcebalance1 ] ) and ( [ eq : forcebalance2 ] ) hold . from the condition for the event horizon @xmath251 and the condition for the extremal rotation @xmath252
, we obtain @xmath253 we concentrate on classes ia and ib , and hence @xmath168 , where @xmath254 ( 2m)^{\frac{2n-4}{2n-3}}(2n-2)^{\frac{-n+2}{2n-3}}2(2n-3 ) .
\qquad \end{aligned}\ ] ] from the inequality ( [ eq : s5_k_min ] ) and @xmath168 , @xmath88 should satisfy the condition @xmath255 we can find @xmath88 if we get @xmath226 which satisfies the inequality @xmath256
\sin^{2}\theta + \frac{j_{1}^{2}}{r_{+}^{2}e^{2}}\leq 0 .
\qquad \end{aligned}\ ] ] we assume the one conserved angular momentum case ( [ eq : s5_one_conserved_angular_momentum ] ) for simplicity . under the assumption ( [ eq : s5_one_conserved_angular_momentum ] ) , we also obtain eqs .
( [ eq : j1_one_conserved_angular_momentum])-([eq : critical_one_conserved_angular_momentum ] ) in even dimensions and the inequality ( [ eq : s5_inequality_even ] ) yields @xmath257 \sin^{2}\theta + 4(n-1)^{2 } \leq 0 .
\qquad \end{aligned}\ ] ] for @xmath238 , from the inequality ( [ eq : s5_inequality_even_one_conserved_angular_momentum ] ) , we obtain the condition @xmath258 thus , we can find @xmath88 for massless particles such as photons or highly relativistic massive particles with @xmath259 in all even dimensions in the region where @xmath260 satisfies the condition @xmath261 where @xmath262 on the equatorial plane @xmath219 , from the inequality ( [ eq : s5_inequality_even_one_conserved_angular_momentum ] ) , we see that @xmath88 will be found in all even dimensions if @xmath263 is satisfied . collisions with an arbitrarily high cm energy occur within this region . in the case of massless or highly relativistic massive particles with @xmath264
, we can show @xmath265 outside the horizon @xmath266 , which is similar to the case of odd dimensions .
the proof is given in appendix [ vnegativeeven ] as the simplest example of the higher and even dimensional case , we will consider the effective potential for a critical particle in the six - dimensional and extremal myers - perry black hole spacetime with the equal angular momenta .
we have to set @xmath267 , @xmath34 and @xmath268 . for simplicity
, we assume @xmath269 and @xmath270 . see eq .
( [ eq : s5_an ] ) for the definition of @xmath271 in appendix b. the inequality ( [ eq : s5_inequality_even ] ) is reduced to @xmath272 equations ( [ eq : s5_jq ] ) and ( [ eq : alpha_q_beta_p ] ) imply @xmath273 equations ( [ eq : s5_a_even_p1 ] ) and ( [ eq : s5_horizon_even_p1 ] ) imply @xmath274 from eqs .
( [ eq : s5_critical ] ) and ( [ eq : s5_a_horizon_m_6d ] ) , the critical particle should satisfy @xmath275 the effective potential for the critical particle is explicitly expressed by @xmath276 \right .
\nonumber\\ & & \left .
\left [ ( r^{2}+a^{2})^{2}-\frac{16a^{3}r}{3\sqrt{3 } } \right ] -\frac{16a^{3}r}{27\sqrt{3}}\frac{(3r^{2}-a^{2})^{2}e^{2}}{r^{2}+a^{2 } } \right\}. \qquad \end{aligned}\ ] ] figure 3 shows the examples of the effective potential @xmath277 . for critical particles in the six - dimensional and extremal myers - perry black hole spacetime with equal angular momenta .
the solid ( red ) , dashed ( green ) , dot - spaced ( blue ) and dotted ( purple ) curves denote the effective potentials for the critical particles of classes ia ( @xmath210 , @xmath211 , @xmath278 ) , ib ( @xmath210 , @xmath212 , @xmath278 ) , ii ( @xmath213 , @xmath279 , @xmath211 , @xmath278 ) and iii ( @xmath213 , @xmath280 , @xmath211 , @xmath278 ) , respectively .
, width=328 ] by substituting eqs .
( [ eq : s5_j_phi_6d ] ) and ( [ eq : s5_critical_6d ] ) into the inequality ( [ eq : s5_inequality_6d ] ) , we obtain the inequality @xmath281 where @xmath282 is defined by @xmath283 the inequality ( [ eq : s5_inequality_critical_6d ] ) is expressed by @xmath284 for @xmath285 , the inequality ( [ eq : s5_inequality_critical2_6d ] ) is satisfied only if @xmath219 and @xmath286 . for @xmath167 , the inequality ( [ eq : s5_inequality_critical2_6d ] ) is not satisfied and therefore we can not find any integral constant @xmath88 which satisfies the inequality ( [ eq : s5_inequarity_k ] ) . in the case
@xmath287 , the inequality ( [ eq : s5_inequality_critical_6d ] ) is satisfied in the region @xmath288 where @xmath289 for @xmath290 we plot @xmath291 in fig .
the collision of a critical particle and a noncritical particle with an arbitrarily high cm energy can occur within this region . versus @xmath292 for a critical particle in the six - dimensional and extremal myers - perry black hole spacetime with the equal angular momenta .
the solid ( red ) , dashed ( green ) and dotted ( blue ) curves denote @xmath291 for @xmath238 , @xmath293 and @xmath294 , respectively . in the region @xmath295 , collisions with an arbitrarily high cm energy can occur .
, width=328 ]
in this section , we investigate a cm energy for the collision of two charged particles in a higher dimensional charged black hole spacetime .
we will show that the cm energy for a critical charged particle and a noncritical charged particle can become arbitrarily large in the near horizon limit and that critical charged particles can reach the event horizon from the infinity in the higher dimensional and extremal charged black hole spacetime , which is similar to the case of four dimensions @xcite .
we consider the @xmath9-dimensional charged black hole spacetime whose line element and gauge 1-form are given by @xcite @xmath296 and @xmath297 respectively , where @xmath298 @xmath299 @xmath300 is @xmath9-dimensional newton s constant and @xmath7 and @xmath301 are the mass and the electric charge of the black hole .
there exists an event horizon if @xmath302 is satisfied . in particular , the horizon is the extremal for @xmath303 .
the event horizon @xmath304 is obtained by @xmath305 without loss of generality , we can set the orbit of the test charged particle on the equator because of the spherical symmetry of the spacetime . in this case
, it is sufficient to consider the reduced line element @xmath306 where we have set the coordinates @xmath307 and @xmath308 .
the hamiltonian of a charged particle with a charge @xmath309 is given by @xmath310 where @xmath311 is the canonical momentum of a charged particle conjugate to the coordinates @xmath312 . from the hamiltonian equation
@xmath313 the @xmath9-momentum @xmath314 of a charged particle is given by @xmath315 from the normalization of the @xmath9-momentum of the charged particle @xmath316 where @xmath63 is the rest mass of the charged particle , the energy equation is obtained by @xmath317 where @xmath318 is the effective potential for the charged particle , @xmath319 is the affine parameter , @xmath320 and @xmath321 and @xmath322 are both constants .
we obtain @xmath323 where @xmath97 . for simplicity
, we consider the case where the charged particle moves in the radial direction by setting @xmath324 . from the forward - in - time condition @xmath103 ,
@xmath325 should be satisfied outside the event horizon .
we define a critical particle in the @xmath9-dimensional charged black hole spacetime as the particle with @xmath326 and hence @xmath327 . therefore , the critical particle satisfies @xmath328 and hence eq .
( [ eq : forcebalance1 ] ) .
we will consider the collision of charged particles @xmath118 and @xmath119 falling into the black hole or @xmath130 , where @xmath329 and @xmath329 are @xmath99 for particles @xmath118 and @xmath119 , respectively .
the cm energy @xmath124 for the collision of the two particles is defined by @xmath330 where @xmath331 , @xmath332 , @xmath333 and @xmath334 are @xmath183 , @xmath63 , @xmath19 and @xmath335 for the particle @xmath140 or @xmath119 , respectively .
using lhospital s rule with respect to @xmath5 in the near horizon limit @xmath132 , we obtain @xmath336 thus , we obtain the cm energy for the collision of two particles in the near horizon limit @xmath132 @xmath337 this shows that the collision with an arbitrarily high cm energy can occur if and only if one of the two particles is critical and the other is noncritical . in the nonextremal case , the critical particle can not reach the event horizon because @xmath338 vanishes and @xmath339 is positive .
thus , the collision with an arbitrarily high cm energy does not occur without multiple scattering in this case .
hereafter we focus on the extremal case , i.e. @xmath340 . in this case , from eqs .
( [ eq : f ] ) , ( [ eq : event_horizon ] ) , ( [ eq : effective_potential ] ) and ( [ eq : x ] ) , we obtain @xmath341 and @xmath342 respectively . here
we have used @xmath343 .
therefore , the effective potential for the critical charged particle is negative outside the event horizon and therefore critical charged particles reach the horizon from infinity .
this shows that critical particles can collide with noncritical particles near the event horizon with an arbitrarily high cm energy .
a critical particle in the extremal charged black hole spacetime satisfies @xmath344 and @xmath345 thus , we obtain @xmath328 and @xmath346 and hence the force balance conditions ( [ eq : forcebalance1 ] ) and ( [ eq : forcebalance2 ] ) are satisfied .
in this paper , we have investigated the collision of two particles around higher dimensional black holes .
we have assumed that the spin parameters of the myers - perry black hole take only two different values at most and test particles with one conserved angular momentum for simplicity .
we have shown that the collision with an arbitrarily high center of mass energy can occur in the vicinity of the event horizon if either of the two colliding particles satisfies the critical particle condition .
if we consider a massless particle or a highly relativistic particle with @xmath347 as a simple case , we can see that the critical particle can reach the event horizon from a distant region as shown in appendixes [ vnegativeodd ] and [ vnegativeeven ] . in the cases of five and six dimensions
, the motion of particles has been investigated in detail for general parameters . as expected in the discussion of analogies between the bsw process and the test - field instability of the extremal black hole in sec .
[ intro ] , an arbitrarily high energy particle collision can occur in the vicinity of the horizon in higher dimensional extremal black hole spacetimes . by considering a critical particle with an arbitrarily small rest mass and the other noncritical particle around an extremal myers - perry black hole , we can construct a perturbation induced by gravity of these two particles which is expected to be well described by linear perturbation around the background spacetime before the collision
. however , if the collision occurs in the vicinity of the event horizon , the center of mass energy can be arbitrarily large , and such intense gravity can not be described in the linear level .
since a field with a short wavelength behaves like a test particle , there holds analogies between the phenomena of test particles and fields .
for example , it is well known that there exists an analogy between the penrose process and the superradiance around rotating black holes .
similarly , we conjecture that the bsw process around an extremal black hole suggests the existence of the linear instability of test fields on the extremal black hole . on the other hand ,
our results suggest that the existence of isco is not essential for the bsw process .
this is because the bsw process works well in higher dimensions as shown in this paper , though it is known that isco does not exist around the higher dimensional black holes . for the rest of this section
, we discuss gravitational backreaction .
recently , murata _
et al . _ studied the test - field instabilities of an extremal reissner - nordstrom black hole , taking gravitational backreaction into account @xcite .
they found that the linearly developed perturbation does not decay even with backreaction but the spacetime evolves to a time - dependent extremal black hole if the initial perturbation is fine - tuned , while the final state is a nonextremal reissner - nordstrom black hole if the initial perturbation is generic .
one can expect that the analogy between the bsw process and the test - field instability may hold even if the gravitational backreaction is taken into account .
it is left as a future work to check this analogy because we currently have a quite limited knowledge about the fate of the bsw process in the presence of the gravitational backreaction @xcite . while this paper was being prepared for submission , an interesting paper @xcite by abdujabbarov _ et al .
_ appeared , where the case of five - dimensional myers - perry black holes was discussed . here
we briefly comment on their claims .
they deal with the center of mass energy of colliding particles around five - dimensional myers - perry black holes not only with equal angular momenta but also with unequal ones . in particular , in the case of @xmath348 and @xmath116 , they showed that the center of mass energy diverges at @xmath349 without the fine - tuned angular momenta of particles .
we point out , however , that as is noted in @xcite , @xmath350 and @xmath351 in the case of @xmath348 and @xmath116 do not correspond to a regular horizon but to a naked conical singularity .
there also exists a naked curvature singularity at @xmath350 and @xmath352 in the case of @xmath348 and @xmath116 @xcite .
hence , this diverging energy should be interpreted in a different context .
abdujabbarov _ et al . _
@xcite also claimed that the center of mass energy of particles which do not have the fine - tuned angular momenta may diverge near the extremal five - dimensional black holes with or without equal angular momenta based on numerical results .
however , eq .
( [ eq : cme ] ) in the current paper clearly disproves this claim .
the authors would like to thank t. kobayashi , h. maeda , t. houri and k. yajima for valuable comments and discussion . especially , the authors thank k. murata for his useful comments about the instabilities of extremal black holes .
is supported by the jsps grant - in - aid for scientific research no .
t.h . was partially supported by the grant - in - aid no .
23654082 for scientific research fund of the ministry of education , culture , sports , science and technology , japan .
was also supported by rikkyo university special fund for research .
in appendix a , we review the line element in the myers - perry black hole spacetime @xcite .
to describe the myers - perry spacetime in @xmath9 @xmath354 dimensions , we will use the boyer - lindquist coordinates , which consist of @xmath16 @xmath17 coordinates @xmath355 with the constraint @xmath356 @xmath357 azimuthal angular coordinates @xmath42 , the radial coordinate @xmath5 and the temporal coordinate @xmath44 .
notice @xmath358 for odd @xmath9 and @xmath359 for even @xmath9 .
the line element in the @xmath9-dimensional myers - perry spacetime in the boyer - lindquist coordinates is given by @xmath360 with @xmath361 for even dimensions , there is an extra unpaired spatial coordinate and we choose the coordinate as @xmath362 .
thus , the spin parameter @xmath21 should vanish for even dimensions .
the components of the inverse metric are obtained by @xmath363 we consider the case where the spin parameters @xmath24 take only two values at most .
see the second and third paragraphs in sec .
ii for our assumptions and conventions about the spin parameters @xmath24 in details .
we can express @xmath364 by @xmath365 for @xmath27 and @xmath366 for @xmath29 , where @xmath367 and @xmath368 satisfy the constraints @xmath369 and @xmath370 respectively .
if we introduce the two sets of spherical polar coordinates @xmath47 and @xmath48 , we can write @xmath371 and @xmath372 as @xmath373 where we use the following rules : @xmath374 and @xmath51 for any function @xmath52 . in the use of the coordinates @xmath45 ,
@xmath375 and @xmath376 , the @xmath377 sector metric @xmath378 in the line element ( [ fulllineelement ] ) can be written as @xmath379 the components of the inverse metric are given by @xmath380 where @xmath381 and @xmath382 run over @xmath40 , @xmath182 , , @xmath187 and @xmath383 where @xmath381 and @xmath382 run over @xmath40 , @xmath182 ,
, @xmath185 .
we obtain the explicit expression for the hamilton - jacobi equation @xcite @xmath386 we can separate @xmath384 and @xmath385 from the hamilton - jacobi equation ( [ eq : s5_hamilton_explitcit ] ) and introduce separation constants @xmath89 and @xmath90 which satisfy @xmath387 and @xmath388 respectively . in addition , we can separate the coordinates @xmath389 , @xmath390 , , @xmath391 inductively from eq .
( [ eq : s5_j1 ] ) and get @xmath392 \nonumber\\ & = & \sum^{q-(w-1)}_{i=1}\frac{\prod^{w-1}_{l=1 } \sin^{2}\alpha_{l}\phi_{i}^{2}}{\prod^{q - i}_{k=1 } \sin^{2}\alpha_{k}\cos^{2}\alpha_{q - i+1 } } + \sum^{q-1}_{i = w}\frac{\prod^{w-1}_{l=1 } \sin^{2}\alpha_{l}}{\prod^{i-1}_{k=1 } \sin^{2}\alpha_{k}}\left ( \frac{ds_{\alpha_{i}}}{d\alpha_{i } } \right)^{2},\\\label{eq : s5_jq-1 } j_{q-1}^{2 } & = & \sin^{2}\alpha_{q-2}\left [ j_{q-2}^{2}-\frac{\phi_{3}^{2}}{\cos^{2}\alpha_{q-2}}-\left ( \frac{ds_{\alpha_{q-2}}}{d\alpha_{q-2 } } \right)^{2 } \right ] \nonumber\\ & = & \frac{\phi_{2}^{2}}{\cos^{2}\alpha_{q-1}}+\frac{\phi_{1}^{2}}{\cos^{2}\alpha_{q}}+\left ( \frac{ds_{\alpha_{q-1}}}{d\alpha_{q-1 } } \right)^{2},\\\label{eq : s5_jq } j_{q}^{2 } & = & j_{q-1}^{2}-\frac{\phi_{2}^{2}}{\cos^{2}\alpha_{q-1}}-\left ( \frac{ds_{\alpha_{q-1}}}{d\alpha_{q-1 } } \right)^{2 } \nonumber\\ & = & \frac{\phi_{1}^{2}}{\cos^{2}\alpha_{q}},\end{aligned}\ ] ] where @xmath393 are separation constants . by changing @xmath394 , @xmath395 , @xmath375 and @xmath309 in eqs .
( [ eq : s5_jn])-([eq : s5_jq ] ) into @xmath396 , @xmath397 , @xmath376 and @xmath183 , similar equations for @xmath398 are obtained as @xmath399 \nonumber\\ & = & \sum^{p-(w-1)}_{i=1}\frac{\prod^{w-1}_{l=1 } \sin^{2}\beta_{l}\psi_{i}^{2}}{\prod^{p - i}_{k=1 } \sin^{2}\beta_{k}\cos^{2}\beta_{p - i+1 } } + \sum^{p-1}_{i = w}\frac{\prod^{w-1}_{l=1 } \sin^{2}\beta_{l}}{\prod^{i-1}_{k=1 } \sin^{2}\beta_{k}}\left ( \frac{ds_{\beta_{i}}}{d\beta_{i } } \right)^{2},\\\label{eq : s5_lp-1 } l_{p-1}^{2 } & = & \sin^{2}\beta_{p-2}\left [ l_{p-2}^{2}-\frac{\psi_{3}^{2}}{\cos^{2}\beta_{p-2}}-\left ( \frac{ds_{\beta_{p-2}}}{d\beta_{p-2 } } \right)^{2 } \right ] \nonumber\\ & = & \frac{\psi_{2}^{2}}{\cos^{2}\beta_{p-1}}+\frac{\psi_{1}^{2}}{\cos^{2}\beta_{p}}+\left ( \frac{ds_{\beta_{p-1}}}{d\beta_{p-1 } } \right)^{2},\\\label{eq : s5_lp } l_{p}^{2 } & = & l_{p-1}^{2}-\frac{\psi_{2}^{2}}{\cos^{2}\beta_{p-1}}-\left ( \frac{ds_{\beta_{p-1}}}{d\beta_{p-1 } } \right)^{2 } \nonumber\\ & = & \frac{\psi_{1}^{2}}{\cos^{2}\beta_{p}},\end{aligned}\ ] ] where @xmath400 are separation constants .
thus , we obtain @xmath401 where @xmath126 and @xmath402 are independent and @xmath403
the line element in the five - dimensional myers - perry black hole spacetime in the boyer - lindquist coordinates is given by @xmath404 the spacetime has killing vectors @xmath405 for stationarity and @xmath406 and @xmath407 for axial symmetries .
the line element is invariant under the transformation @xmath408 the event horizon exists at @xmath409 , where @xmath410,\end{aligned}\ ] ] if and only if @xmath411 note that @xmath412 vanishes .
the angular velocities of the event horizon , @xmath413 and @xmath414 , are given by @xmath415 and @xmath416 , respectively .
the extremal condition is expressed by @xmath417 and the event horizon in the extremal case is given by @xmath418
in the four - dimensional extremal kerr black hole case , we have @xmath202 ( @xmath201 ) .
the inequality ( [ eq : s5_inequality_even ] ) is reduced to @xmath419 equations ( [ eq : alpha_q_beta_p ] ) and ( [ eq : s5_jq ] ) imply @xmath420 equations ( [ eq : s5_a_even_p1 ] ) and ( [ eq : s5_horizon_even_p1 ] ) imply @xmath421 from eqs .
( [ eq : critical_one_conserved_angular_momentum ] ) and ( [ eq : s5_a_horizon_m_4d ] ) , the critical particle should satisfy @xmath422 by substituting eqs .
( [ eq : s5_j_phi_4d ] ) and ( [ eq : s5_critical_4d ] ) into the inequality ( [ eq : s5_inequality_4d ] ) , we obtain the inequality @xmath423 which coincides with the inequality ( 4.5 ) of harada and kimura @xcite .
in this appendix , we show that the function @xmath424 in eq .
( [ tkmax ] ) takes a negative value outside the horizon in odd dimensions .
hereafter we denote @xmath424 as @xmath175 for simplicity .
we define nondimensional radial coordinate @xmath425 as @xmath426 . in this coordinate
, the horizon locates at @xmath427 from eq .
( [ aodddim ] ) . using eqs .
( [ aodddim ] ) and ( [ rpodddim ] ) , the function @xmath175 becomes @xmath428.\end{aligned}\ ] ] we can easily show that @xmath175 and its first , second and third derivatives take a zero or negative value at the horizon @xmath427 , @xmath429 the fourth derivative of @xmath175 becomes @xmath430.\end{aligned}\ ] ] we can see this takes a negative value outside the horizon @xmath431 .
thus we can say @xmath176 , namely @xmath265 outside the horizon .
in this section , we show @xmath265 , namely the function @xmath175 in eq .
( [ tevendim ] ) takes a negative value outside the horizon for a massless critical particle with one conserved angular momentum ( [ eq : s5_one_conserved_angular_momentum ] ) around even dimensional myers - perry black holes with equal angular momenta .
similar to the discussion in odd dimensions , we only have to show @xmath432 in the case of @xmath177 in eq .
( [ krangeeven ] ) . in this case , the function @xmath175 becomes @xmath433 -2^{n+1 } ( n-1)^{n-1 } \tilde{r } \right\ } , \qquad\end{aligned}\ ] ] where @xmath434 and we have used eqs .
( [ eq : critical_one_conserved_angular_momentum ] ) , ( [ eq : s5_a_even_p1 ] ) and ( [ eq : s5_horizon_even_p1 ] ) .
we can easily show @xmath175 and its first , second , third and fourth derivatives take a zero or negative value on the horizon @xmath427 , @xmath435 \le 0,\\ \frac{d^4t}{d\tilde{r}^4 } \bigg|_{\tilde{r } = 1 } & = & - \frac{r_+^{2 n } e^2 } { 2n - 3 } 2^{n+1 } ( n-1)^{n-4 } \left [ 3 + 25(n-2 ) + 60 ( n-2)^2 + 44(n-2)^3 \right ] \le 0.\end{aligned}\ ] ] the fifth derivative of @xmath175 becomes @xmath436 \nonumber\\ & & + ( \tilde{r } -1)^2 \left [ 480 + 6480(n-2)+13200(n-2)^2 + 8640(n-2)^3 \right ] \nonumber\\ & & + ( \tilde{r } -1)^3 \left [ 864 + 6256(n-2)+10064(n-2)^2 + 5856(n-2)^3 \right ] \nonumber\\ & & + ( \tilde{r } -1)^4 \left [ 792 + 4268(n-2)+5812(n-2)^2 + 2808(n-2)^3 \right ] \nonumber\\ & & + ( \tilde{r } -1)^5 \left [ 480 + 2048(n-2)+2336(n-2)^2 + 896(n-2)^3 \right ] \nonumber\\ & & + ( \tilde{r } -1)^6 \left [ 192 + 696(n-2)+688(n-2)^2 + 224(n-2)^3 \right ] \nonumber\\ & & + ( \tilde{r } -1)^7 \left [ 48 + 152 ( n-2)+128(n-2)^2 + 32(n-2)^3 \right ] \nonumber\\ & & \left .
+ ( \tilde{r } -1)^8 \left [ 6 + 19 ( n-2)+16(n-2)^2 + 4(n-2)^3\right ] \right\}.\end{aligned}\ ] ] in the curly brackets , the function is written in the form of the taylor expansion around the horizon @xmath427 .
since all the coefficients of the taylor expansion are manifestly non - negative because @xmath437 , we can see @xmath438 takes a negative value outside the horizon @xmath431 .
so we can say @xmath432 , namely @xmath265 outside the horizon .
wei , y .- x . liu , h. guo and c .- e .
fu , phys .
d * 82 * , 103005 ( 2010 ) .
o. b. zaslavskii , jetp lett . * 92 * , 571 ( 2010 ) [ pisma zh .
fiz . * 92 * , 635 ( 2010 ) ] .
m. kimura , k .-
nakao and h. tagoshi , phys .
d * 83 * , 044013 ( 2011 ) .
t. harada and m. kimura , phys .
d * 83 * , 024002 ( 2011 ) .
e. berti , v. cardoso , l. gualtieri , f. pretorius and u. sperhake , phys .
lett . * 103 * , 239001 ( 2009 ) .
t. jacobson and t. p. sotiriou , phys .
lett . * 104 * , 021101 ( 2010 ) .
a. a. grib and y. v. pavlov , astropart .
phys . * 34 * , 581 ( 2011 ) .
k. lake , phys .
lett . * 104 * , 211102 ( 2010 ) [ erratum - ibid . *
104 * , 259903 ( 2010 ) ] .
m. banados , b. hassanain , j. silk and s. m. west , phys . rev .
d * 83 * , 023004 ( 2011 ) .
o. b. zaslavskii , class .
* 28 * , 105010 ( 2011 ) . m. patil , p. s. joshi and d. malafarina , phys .
d * 83 * , 064007 ( 2011 ) .
t. harada and m. kimura , phys .
d * 83 * , 084041 ( 2011 ) .
m. patil and p. s. joshi , phys .
d * 84 * , 104001 ( 2011 ) .
m. patil , p. s. joshi , m. kimura and k .-
nakao , phys .
d * 86 * , 084023 ( 2012 ) .
t. harada and m. kimura , phys .
d * 84 * , 124032 ( 2011 ) .
t. igata , t. harada and m. kimura , phys .
d * 85 * , 104028 ( 2012 ) .
t. harada , h. nemoto and u. miyamoto , phys .
d * 86 * , 024027 ( 2012 ) [ erratum - ibid .
d * 86 * , 069902 ( 2012 ) ] .
s. t. mcwilliams , phys .
lett . * 110 * , 011102 ( 2013 ) .
h. nemoto , u. miyamoto , t. harada and t. kokubu , phys .
d * 87 * , 127502 ( 2013 ) .
a. galajinsky , phys . rev .
d * 88 * , 027505 ( 2013 ) .
o. b. zaslavskii , phys .
lett . * 111 * , 079001 ( 2013 ) . k. -i .
nakao , m. kimura , m. patil and p. s. joshi , phys .
d * 87 * , 104033 ( 2013 ) .
s. aretakis , commun .
* 307 * , 17 ( 2011 ) . s. aretakis , ann .
henri poincar * 12 * , 1491 ( 2011 ) .
s. aretakis , j. funct .
anal . *
263 * , 2770 ( 2012 ) .
s. aretakis , arxiv:1206.6598 [ gr - qc ] .
s. aretakis , class .
* 30 * , 095010 ( 2013 ) .
s. aretakis , phys .
d * 87 * , 084052 ( 2013 ) .
k. murata , h. s. reall and n. tanahashi , class .
quant . grav .
* 30 * , 235007 ( 2013 ) .
k. murata , class .
grav . * 30 * , 075002 ( 2013 ) .
r. c. myers and m. j. perry , ann .
( n.y . ) * 172 * , 304 ( 1986 ) .
m. vasudevan , k. a. stevens and d. n. page , class .
* 22 * , 1469 ( 2005 ) .
r. emparan and h. s. reall , living rev .
rel . * 11 * , 6 ( 2008 ) . v. p. frolov and d. stojkovic , phys .
d * 67 * , 084004 ( 2003 ) .
v. p. frolov and d. stojkovic , phys .
d * 68 * , 064011 ( 2003 ) .
a. n. aliev and v. p. frolov , phys .
d * 69 * , 084022 ( 2004 ) .
|
we study high energy particle collisions around higher dimensional black holes .
it is shown that the center of mass energy can be arbitrarily large in the vicinity of the event horizon like the baados , silk and west ( bsw ) process in four dimensions if the following two conditions are satisfied : ( i ) the horizon is extremal and ( ii ) the parameters of either of the two colliding particles are fine - tuned , which is called a critical particle condition . we also show that a test particle which satisfies the critical particle condition can reach the event horizon from a distant region for a simple case .
finally , we discuss the relation between the bsw process and the linear instability of test fields around extremal black holes , the latter of which has been recently found by aretakis @xcite .
|
the ability to controllably fabricate , manipulate and examine structures on the sub - micrometer scale has let to the observation of quantum phenomena in electron transport such as , e.g. , universal conductance fluctuations ( ucf ) in chaotic billiards and weak localization ( wl ) , which dominate transport at the nanoscale.@xcite by reducing the characteristic system size below the electronic inelastic mean free path , transport enters the so - called ballistic regime.@xcite ballistic electron transport is a prime candidate for semiclassical descriptions @xcite where the classical trajectories carry an amplitude which reflects the stability of the classical orbits and a phase which contains the classical action and accounts for quantum interference.@xcite on a more fundamental level , the semiclassical framework provides a conceptually powerful bridge between classical and quantum mechanics allowing an intuitive approach to quantum mechanics and quantum chaos in general , and to transport through open quantum dots or so - called quantum billiards in particular.@xcite + several semiclassical approximations ( scas ) based on the approximation of the constant - energy green s function for propagation in a billiard have been proposed and compared with numerical quantum transport calculations or experiment.@xcite while many qualitative features could be well reproduced , quantitative agreement on a system - specific level has remained a challenge .
+ one underlying difficulty is the multi - scale nature of the quantum - to - classical transition for transport through open quantum dots . for the semiclassical approximation to hold
, the de broglie wavelength @xmath0 should be vanishingly small compared to all characteristic dimensions of the device .
such asymptotic theories have been successfully employed to reproduce , upon ensemble averaging , random matrix theory ( rmt ) results for chaotic cavities ( see e.g. ref . ) .
a quantitative comparison on a system - specific level with full quantum calculations or experiments is , however , only possible in the non - asymptotic regime where @xmath0 is small compared to the linear dimension @xmath1 of the dot , @xmath2 , but still comparable to the width of the lead ( or quantum wire ) @xmath3 , @xmath4 .
moreover , for billiards with sharp edges the proper asymptotic limit is , rigorously , out of reach .
the present theory addresses this non - asymptotic semiclassical regime , often also referred to as the `` near '' semiclassical regime .
we show that the proper inclusion of diffractive contributions allows to quantitatively reproduce quantum calculations .
the diffractive coupling between classical paths gives rise to pseudo - paths that are missing in the standard sca and are the key to remedy many of the deficiencies of semiclassical approximations .
+ we show in the present communication that this pseudo - path semiclassical approximation ( psca ) can reach quantitative agreement with full quantum simulations provided a high - order diffraction theory for the coupling between classical paths is used . for the scattering at the leads we develop an approximation involving elements of both the uniform theory of diffraction ( utd)@xcite and the geometric theory of diffraction ( gtd)@xcite referred to in the following as the gtd - utd approximation . with these ingredients
good agreement with quantum simulations is found .
+ one key conceptual insight is the unambiguous identification of the paths that contribute to quantum transport .
we apply the present theory to a circle - shaped regular quantum dot for which the enumeration of paths , more precisely of path bundles , is easily possible .
unlike for chaotic dots , for which the exponential proliferation of contributing paths as a function of the pathlength makes their unique identification difficult , for regular systems their enumeration and identification is straightforward up to large pathlengths .
+ the circular billiard is depicted in fig .
[ fig : circle ] .
the leads are attached at right angle and have equal width @xmath3 . in order to probe the local topology of the cavity we require sufficiently long dwell times such that @xmath5 .
the wavelength of the electron @xmath0 fulfills @xmath6 for the semiclassical limit to hold inside the cavity .
however , as in experimental or numerical studies of quantum billiards @xmath4 .
our semiclassical theory can thus be quantitatively compared with quantum mechanical numerical calculations for the circular billiard .
+ this paper is organized as following : in sec .
[ transport ] we review both the standard sca as well as the psca .
these approximations differ by the different path sets entering the corresponding green s function . in sec .
[ sec : diff ] we introduce the gtd - utd diffraction approximation which is a key to the quantitative agreement between the psca and quantum mechanics for transport properties
. the important role of pseudo - paths and a proper diffraction theory is demonstrated on the level of quantum mechanical length - area spectra @xcite in sec .
[ sec : paths ] . finally , we compare in sec .
[ sec : res ] the semiclassical predictions for a variety of quantum transport properties that play a key role in the understanding of the quantum - to - classical crossover , in particular conductance fluctuations ( cf ) , weak localization ( wl ) and quantum ( non - thermal ) shot noise , with quantum calculations . with perpendicular leads of equal width @xmath3 .
in accordance with previous work@xcite on the circular and the stadium billiard of equal area @xmath7 , we choose @xmath8 and @xmath9.,width=188 ]
we consider ballistic transport through quantum billiards for which transport properties are determined by the wave number @xmath10 and magnetic field @xmath11 dependent quantum mechanical hamiltonian @xmath12-matrix .
dissipative or dephasing processes are neglected .
( we will return to the effect of decoherence below . )
moreover , we refer to a `` clean '' billiard when elastic scattering at a disorder potential in the interior of the structure is absent . in this limit
, the @xmath12-matrix represents elastic scattering at the boundaries of the billiard only .
+ the @xmath12 matrix elements @xmath13 describe the scattering from transverse mode @xmath14 in lead @xmath15 to mode @xmath16 in lead @xmath17 .
we denote the transmission amplitudes from lead @xmath18 to lead @xmath19 as @xmath20 and the reflection amplitudes back into lead 1 as @xmath21 .
according to the landauer formula the conductance @xmath22 of a quantum billiard is directly proportional to the total transmission @xmath23 , @xmath24 where @xmath25 is the number of open modes in the leads having equal width @xmath3 .
the @xmath12 matrix elements can be determined by a projection of the lead modes @xmath26 ( @xmath27 is the transverse coordinate in the lead ) onto the green s function @xmath28 for propagation from @xmath27 to @xmath29 at the cavity lead junction of lead i and lead j , respectively .
the @xmath12-matrix elements at @xmath30 are given by the fisher - lee equations @xcite @xmath31 where @xmath32 ( @xmath33 ) is the longitudinal wave number in lead @xmath18 ( lead @xmath19 ) .
the prefactor @xmath34 is due to flux normalization .
we use atomic units ( @xmath35 ) . at zero magnetic field
the mode wavefunctions take the form @xmath36 for non - zero magnetic field @xmath37 , eqs .
( [ eq : tnm ] , [ eq : modes ] ) have to be modified ( see , e.g. , ref . and references therein ) .
we use the modular recursive green s function method to calculate the exact quantum mechanical @xmath12-matrix elements for a given @xmath38 and @xmath39 ( for details see ref . ) .
the starting point of semiclassical approximations are the @xmath12-matrix elements [ eq .
( [ eq : tnm ] ) ] . in a first step ,
the quantum mechanical green s function @xmath40 is replaced by a corresponding semiclassical expression which represents the fourier - laplace transform in stationary phase approximation of the semiclassical limit of the feynman propagator .
it describes propagation along classical paths with fixed energy .
depending on the class of paths included , a hierarchy of different semiclassical approximations to the green s function results .
these are to be distinguished from the different level of additional approximations employed in the evaluation of the integral [ eq . ( [ eq : tnm ] ) ] which projects the green s function onto the lead function .
the latter gives rise to another set of semiclassical approximations to the @xmath12-matrix elements .
+ we focus first on the replacement of @xmath40 by a semiclassical approximation .
for the latter we consider two different levels of approximation , the standard semiclassical approximations ( sca ) and the pseudo - path semiclassical approximation ( psca ) . both result from the stationary phase approximation ( spa ) to the full feynman propagator reducing the continuous set of paths entering feynman s path integral to a discrete subset of paths .
assuming that the classical action @xmath12 is much larger than @xmath41 and that well localized and separated stationary points with @xmath42 exist , the standard sca contains exclusively classical paths .
however , near sharp edges of the cavity or near the cavity - lead junctions , the de broglie wavelength is not negligibly small and the spa will fail .
this leads to diffractive corrections which can be taken into account within the framework of the pseudo - path semiclassical approximation ( psca ) .
one of its salient features is that the basic notion of a propagator consisting of a sum over a discrete set of paths is preserved .
diffraction effects simply appear as additional contributions to the path sum [ see sec . [ transport ] ( c ) ] .
+ the classical action for an electron moving along a path @xmath43 is given by @xmath44 , where @xmath45 is the length and @xmath46 is the directed enclosed area of the path .
@xmath46 can have both positive and negative values depending on the rotational direction of the path .
in _ all _ our semiclassical calculations ( standard sca as well as the psca ) the magnetic field enters only via the aharonov - bohm phase @xmath47 . the curvature of the paths as well as the effect of non - zero magnetic field on the diffraction at the lead ( introduced in sec . [
sec : diff ] ) is neglected since we consider the regime of weak magnetic fields ( @xmath48 with @xmath49 being the radius of the circular cavity ) . the green s function within the standard sca entering eq .
( [ eq : tnm ] ) is given by : @xmath50 } , \label{eq : gssca}\end{aligned}\ ] ] where @xmath51 is the deflection factor ( a measure for the divergence of nearby trajectories ) and @xmath52 is the maslov index .
the deflection factor is defined as @xmath53 where @xmath54 ( @xmath27 ) is the longitudinal ( transverse ) component of the trajectory s starting point ( @xmath15 ) and end point ( @xmath17 ) , respectively .
the maslov index increases by two for every reflection at the hard wall boundary of the billiard and by one when passing a focal point along the trajectory .
( [ eq : gssca ] ) contains a sum over all _ classical paths _ @xmath43 connecting the entrance lead @xmath15 with the exit lead @xmath17 ( see sec .
[ sec : claspaths ] ) . + evaluation of eq .
( [ eq : tnm ] ) with @xmath55 , @xmath56 proceeds either numerically@xcite or analytically by invoking another set of spas .
it was recognized from very early on that the spa as applied to eq .
( [ eq:6 ] ) is poorly justified in the non - asymptotic regime when @xmath57 .
therefore , various diffraction integral approximations have been proposed.@xcite the transmission amplitudes then take the form : @xmath58}\nonumber\\&\times & c_m(\theta_1,k , d ) , \label{eq : tnmssca}\end{aligned}\ ] ] expressed in terms of diffraction coefficients @xmath59 describing the diffractive coupling from the entrance lead mode @xmath14 into the dot and from the dot into the exit lead mode @xmath16 . deviating from previous calculations ,
we introduce for @xmath60 a combination of keller s geometric theory of diffraction ( gtd ) @xcite and the `` uniform theory of diffraction '' ( utd).@xcite the derivation of @xmath61 within the gtd - utd is given in app .
[ appb ] . the inclusion of diffraction effects in terms of diffraction coefficients in eq .
( [ eq : tnmssca ] ) preserves the structure of the semiclassical transmission amplitude in terms of a discrete sum over paths contributing to transport .
the diffraction coefficients provide @xmath62- and @xmath38-dependent weighting factors for each path contributing to the transmission from mode @xmath14 to mode @xmath16 . within the framework of the diffractive couplings into and out of the leads ,
the entrance and exit leads are treated as point scatterers.@xcite each path bundle which connects the entrance and the exit lead is replaced by an appropriately weighted representative path q connecting the center of the entrance lead with the center of the exit lead . consequently one can replace the deflection factor as given in eq . ( [ deflec ] ) by its value in the closed circular billiard @xmath63 , where @xmath45 is the length of the path . + the diffractive lead - dot couplings in eq .
( [ eq : tnmssca ] ) should be distinguished from diffractive corrections included in the propagation in the interior of the billiard .
we refer to eq .
( [ eq : tnmssca ] ) as the standard sca while inclusion of diffractive corrections in the billiard corresponds to the psca .
classical paths in a regular billiard ( such as the circle ) feature a highly ordered structure of their length and enclosed area distribution ( see fig . [
fig : class ] ) .
the branch structure of the `` length - area '' distribution of paths connecting the entrance with the exit point is a specialty of the circular cavity . along each branch
the number of bounces off the wall increases by one from one path to its next higher neighbor .
the points of convergence of each branch mark those paths that bounce off the wall infinitely many times and thus run exactly along the cavity boundary . in the limit @xmath64 ( where @xmath62 is the entrance angle as given in fig .
[ fig : circle ] ) each branch contains an infinite number of paths . in our numerical calculations ( see sec . [
sec : res ] ) paths near this cluster point effectively do not contribute as they are cut off by vanishing diffraction coefficients @xmath65 .
+ the distribution of paths eventually reflected back to the entrance point [ fig .
[ fig : class ] ( a ) ] is symmetrically distributed relative to the @xmath66 axis due to time - reversal symmetry .
every path has a counterpart of equal length but opposite sign of the enclosed area .
the lowest left and right branches consist of polygons with a number of revolutions @xmath67 in the cavity .
the polygons can be characterized by an angle @xmath68 where @xmath69 is the number of corners . along each branch
@xmath69 increases by one from one path to the next .
the paths of the next higher branches revolve twice ( @xmath70 ) around the circle and @xmath71 .
all higher branches can be described analogously .
+ the branches of transmitted paths are not symmetric relative to the @xmath72 axis but show a clear off - set ( fig .
[ fig : class ] ( b ) ) .
path pairs with similar length do not have , in general , the same topology but differ in the number of bounces off the hard wall boundary . as a consequence ,
these path pairs have different maslov indices and thus do not interfere constructively .
[ fig : class ] also illustrates the difference between the open and closed billiard , i.e. , the effect of `` path shadowing '' or suppression of longer paths due to their prior exit from the structure .
all paths that would be geometrically reflected off the closed lead in the closed system are missing in the open billiard .
the difference between the closed and open billiard is particularly evident in reflection since all paths with four - fold symmetry leave the cavity via lead 2 before being reflected back to lead 1 .
this is system - specific for the circular billiard with perpendicular leads ( see fig . [
fig : circle ] ) .
+ the distinctly different path distributions for transmission and reflection point already to a clear structural deficiency of the sca . many quantum properties of transport are a consequence of the intrinsic coupling between transmission and reflection .
the standard sca does not incorporate this quantum aspect of non - locality .
classically , the path sets of transmission and reflection are disjoint .
the distribution of transmitted paths is markedly different from the one of reflected paths . in quantum transport
transmitted and reflected paths are intertwined and must share the information on the relative phases .
a semiclassical theory that reproduces quantum features must therefore allow for coupling between the path sets associated with transmission and reflection .
this is the key property of pseudo - paths discussed in the following .
the pseudo - path semiclassical approximation @xcite goes beyond the standard sca by systematically including diffractive corrections into the propagation in the interior of the billiard . in the present case ,
diffractive corrections arise from multiple back scattering , i.e. , internal reflections at the leads .
we point to the conceptual similarity to `` pseudo - orbits '' introduced in ref . for the study of the density of states in a closed wedge billiard and to `` hikami boxes''@xcite introduced for elastic scattering at short - ranged potentials in the interior of a diffusive quantum dot .
+ in line with multiple scattering theory the pseudo - path semiclassical green s function can be expressed in terms of a ( semiclassical ) dyson equation,@xcite @xmath73 in the present case , @xmath55 plays the role of the unperturbed green s function and the perturbation `` potential '' @xmath74 accounts for the ( internal ) diffractive scatterings at the lead opening . the unperturbed propagation inside the cavity @xmath75 is equivalent to the free propagation in the `` closed '' system .
+ for a two - terminal system the perturbation potential @xmath74 is given by @xmath76 where @xmath77 and @xmath78 describe the diffractive scattering off lead 1 and lead 2 , respectively .
the semiclassical dyson equation , eq .
( [ dysonpsc ] ) , reads @xmath79 to first order , the psca to the green s function ( denoted by @xmath80 ) connecting lead @xmath15 with lead @xmath17 includes terms in @xmath74 of the form , @xmath81 eq . ( [ eq:11 ] )
may serve as example to illustrate the physics entering eq .
( [ dysmat ] ) .
it describes propagation from lead @xmath15 to lead @xmath17 via one intermediate visit to lead @xmath82 where diffractive internal back scattering with amplitude @xmath83 takes place .
the path from lead @xmath15 to lead @xmath82 , @xmath84 as well as from lead @xmath82 to lead @xmath17 , @xmath85 , are classical paths described by @xmath55 .
diffractive scattering couples the incident path @xmath84 with angle @xmath86 to the exiting path @xmath85 with angle @xmath87 thereby coupling two disjoint subsets of classical paths and generating a first - order pseudo - path .
the determination of the diffraction coefficient @xmath88 will be discussed in more detail in sec .
[ sec : diff ] . from a conceptual point of view , the pseudo - path semiclassical approximation [ eq . ( [ dysmat ] ) ]
is closely related to the diagrammatic perturbation theory.@xcite both leads 1 and 2 ( i.e. , @xmath77 and @xmath78 ) act much like `` hikami boxes'',@xcite where electrons can not be described semiclassically since the characteristic potential length scale ( in our case the sharp edges of the leads ) is smaller than the electron wavelength @xmath0 .
thus , the wave nature of electrons has to be taken into account as diffraction at the point scatterers allowing classically distinct path sets to mix .
this is crucial for the quantum corrections to the transmission and reflection amplitudes and repairs some of the deficiencies of the standard sca ( see sec .
[ sec : res ] ) . ) .
the number of first - order pseudo - paths up to length @xmath89 is larger by more than a factor of @xmath90 compared to the number of classical paths.,width=264 ] within the psca pseudo - paths are formed by joining classical paths together via diffraction . with each increasing order of the psca
the length - area distribution gets more and more densely filled with paths , or equivalently , the number of paths increases with the order of the psca ( fig .
[ fig : class1kink ] ) .
pseudo - paths form product sets of classical paths , e.g. , joining a given classical path for transmission with a classical path contributing to reflection ( or in reverse order ) forms the set of transmitted pseudo - paths to first order .
reflected first - order pseudo - paths result from joining two classical transmitted paths or two classical reflected paths .
this coupling allows to recover the non - locality of quantum transport .
higher - order pseudo - paths are constructed analogously . with increasing order and increasing ( combined ) pathlength the total number of pseudo - paths exponentially proliferates .
this is in sharp contrast to the power - law growth of purely classical paths for regular systems and explains why the effect of diffractive scattering is more likely visible in regular than in chaotic systems where the exponential proliferation of classical orbits may mask the diffractive contributions .
+ the length - area distribution of first order pseudo - paths contained in eq .
( [ dysmat ] ) contributing to reflection is ( of course ) still symmetric [ fig .
[ fig : class1kink ] ( a ) ] .
in addition , new branches appear with classical and pseudo - path partners of approximately equal length .
the change in the branch structure is more dramatic in the spectrum of transmitted paths [ fig .
[ fig : class1kink ] ( b ) ] .
the pseudo - paths lead to a `` symmetrization '' of the length - area distribution .
the symmetrization results primarily from paths which _ change _ their _ rotational direction _ through diffractive scattering [ see the pseudo - path in the inset of fig . [
fig : class1kink ] ( b ) ] .
branches of classical paths are now completed by symmetric pseudo - path `` partner '' branches of approximately equal length and different enclosed area .
+ the weight and the phase of interfering paths are strongly influenced by the diffraction coefficient @xmath91 . as will be demonstrated in sec .
[ sec : res ] , the previously employed fraunhofer theory of diffraction @xcite is not sufficiently accurate as to give quantitatively reliable results for transport properties .
the same holds for the kirchhoff theory of diffraction @xcite which is closely related to the fraunhofer theory of diffraction and gives similar results for the diffraction coefficients ( see fig .
[ fig : diff_qm ] in the following section ) . for a quantitative agreement between the psca and quantum mechanics , it is thus necessary to go beyond low - order diffraction approximations and use a more sophisticated theory of diffraction .
we present such a theory in the following section .
reaches the orifice under an angle @xmath62 and is backscattered into a path @xmath92 that leaves with an angle @xmath93 ( the angles @xmath62 and @xmath93 as depicted in the figure have opposite signs ) .
the dashed lines denote that @xmath94 and @xmath95 are in the far field region.,width=264 ] the contribution of a given classical path with pathlength @xmath45 to the standard semiclassical green s function , @xmath55 , is @xmath96 eq . ( [ eq : gfree ] ) is the basic building block entering the dyson eq .
( [ dysmat ] ) together with the diffraction coefficient @xmath97 . in line with the far - field approximation underlying diffraction theory ,
@xmath97 is assumed to be independent of the length of the path approaching or exiting the diffractive scattering region . +
a successful application of the pseudo - path semiclassical approximation outlined in the preceding subsection requires the determination of accurate diffraction coefficients @xmath91 .
different approximations have been used in the past for the inclusion of diffraction effects : kirchhof diffraction approximation ( kda),@xcite fraunhofer diffraction approximation ( fda),@xcite geometric theory of diffraction ( gtd),@xcite and the `` uniform theory of diffraction '' ( utd).@xcite we have developed a new theory for the reflection at open lead mouths by combining the gtd with the utd ( the gtd - utd ) to take into account paths that scatter multiple times between the edges of the leads , ( see appendix [ appa ] ) . with this theory
we obtain diffraction coefficients in excellent agreement to quantum mechanics .
+ consider , as a test case , the diffractive scattering ( fig . [ fig : diff_sketch ] ) at the lead mouth described by the first - order term [ eq .
( [ eq:11 ] ) ] , @xmath98 the incoming path @xmath43 is incident at angle @xmath62 ( measured with respect to the surface normal ) and is diffractively scattered into angle @xmath93 under which path @xmath92 leaves the scattering region . we compare ( fig .
[ fig : diff_qm ] ) the present gtd - utd theory with the fda , the kda and exact quantum mechanical ( qm ) calculations.@xcite even for a typical @xmath38 value in the low mode regime ( @xmath992.5@xmath100 ) , the agreement between the gtd - utd and the exact qm calculations is very good whereas both the fda and the kda display clear deviations from the qm values . while these deviations do not appear dramatic at first glance , they are , in fact , quantitatively very important as the diffraction coefficient enters the dyson series [ eq .
( [ dysmat ] ) ] to all orders .
note , however , that the gtd - utd would fail in the limit @xmath64 [ as indicated by the kink at @xmath101 in fig .
[ fig : diff_qm ] ( c ) ] .
this deficiency is of no concern for the present applications as the probability for diffractive scattering tends to zero in this limit .
the maximum of the diffractive reflection probability is clearly around the specular value @xmath102 .
however , it is important to note that the probability distribution possesses a local maximum at the backscattering angle @xmath103 . a non - negligible part of the electron wave is back - scattered into the direction from where it came from .
this diffractive back - reflection should not be confused with the well - known andreev back - reflection in which the back - reflected particle simultaneously undergoes a particle - hole exchange.@xcite back - reflected paths are responsible for the symmetrization of the distribution of transmitted paths [ fig .
[ fig : class1kink ] ( b ) ] and are crucial for the understanding of the weak - localization dip in the transmission.@xcite + while the kda and the fda have been successfully used in the past to explain certain features of conductance fluctuations,@xcite only the gtd - utd is precise enough to reproduce transport semiclassically on a quantitative level ( see sec .
[ sec : res ] ) . in sec .
[ sec : wl ] we compare the results for transport properties obtained by implementing the gtd - utd and the fda to full quantum mechanical calculations .
the information on paths governing quantum transport can be reliably extracted from the two - dimensional fourier transforms of the quantum mechanical @xmath12-matrix elements,@xcite @xmath104 .
the @xmath12-matrix elements display a strongly fluctuating pattern as a function of @xmath38 and @xmath39 ( see fig . [
fig : t22 kb ] for @xmath105 ) .
since the canonically conjugate variables to the wavenumber @xmath38 and the magnetic field @xmath39 are the length @xmath106 and the directed area @xmath107 , respectively , the fourier transform @xmath108 allows the unambiguous identification of quantum paths contributing to quantum transport via their length and enclosed area [ fig .
[ fig : laspec ] ( a)-(d ) ] .
no apriori - assumption as to the existence of classical paths @xmath109 with pathlength @xmath106 and area @xmath107 enters eq .
( [ eq : fouriert ] ) .
the two - dimensional pathlength - area spectrum allows to identify both classical as well as non - classical contributions to the full quantum spectra .
[ fig : laspec ] displays examples of pathlength - area spectra @xmath110 . [
note that fig .
[ fig : laspec ] ( c ) is the absolute square of the fourier transform of the transmission amplitude whose absolute square is plotted in fig .
[ fig : t22 kb ] ] . obviously the strong fluctuations of conductance in quantum transport ( fig .
[ fig : t22 kb ] ) are the result of the interference of clearly identifiable ( quantum ) paths ( fig .
[ fig : laspec ] ) . + the quantum mechanical @xmath12-matrix elements
@xmath104 are determined with the help of the modular recursive green s function method@xcite and then numerically fourier transformed .
finite discretized intervals must be used when performing the fourier transform numerically .
the integration intervals are denoted by @xmath111 ( @xmath112 ) and the numerical grid spacings by @xmath113 ( @xmath114 ) .
accordingly , the resolvable length is @xmath115 = @xmath116 and the resolvable area interval is @xmath117
. the magnitude of the @xmath12-matrix elements decreases with increasing length , which is an obvious consequence of open systems : the probability to stay within the cavity decreases with increasing length .
the parameter @xmath113 must be chosen sufficiently small such that the maximum resolvable length ( @xmath115 ) lies already in the region of strongly reduced amplitudes .
otherwise , the fourier spectrum is visibly back - folded onto the fundamental interval .
the magnetic field interval is further restricted by the requirement that the curvature of the paths is negligible , i.e. , the cyclotron radius @xmath118 should be much larger than the circle radius @xmath49 .
we chose the interval @xmath119 and @xmath120 as well as the number of interval points such that a maximum length of @xmath121 is resolved [ in fig .
[ fig : laspec ] ( a)-(d ) only the contributions with @xmath122 are shown , the insets contain the entire spectrum ] . except for @xmath123 ,
the absolute square of the @xmath12-matrix elements is considerably damped at a length of @xmath124 .
thus , the graphs represent essentially the entire length - area spectrum . only for @xmath123 ,
contributions with @xmath125 are non - negligible which leads to a back - folding near @xmath126 [ the lowest branch structure near @xmath127 in fig .
[ fig : laspec ] ( b ) ] .
+ the quantum length - area spectra provide detailed information on the paths contributing to transport through a specific system .
they represent the paths entering the _ full _ feynman path integral .
the following general trends can be observed : long paths are more prevalent in @xmath12-matrix elements connecting low mode numbers ( in the present case they contribute more strongly to @xmath12-matrix elements @xmath128 than to @xmath129 ) .
lower modes favor smaller entrance and exit angles that are associated with longer paths with a larger number of bounces off the cavity boundary .
the most important observation is the remarkably close correspondence of the quantum mechanical length - area spectrum to its classical counterpart .
important contributions are located near classical paths
. moreover the branch structure of classical paths is reproduced [ fig .
[ fig : laspec ] ( a)-(d ) ] . on the other hand , there are distinct structures which do not correspond to classical paths and which can be identified using the distributions in fig .
[ fig : class1kink ] ( a ) and ( b ) .
the quantum mechanical length - area spectra confirm the existence and substantial role of non - classical paths , the pseudo - paths : these are those paths that are one or several times diffractively reflected off the lead before exiting the cavity .
two examples are given in the following : for @xmath130 we identify an ensemble of diffractive paths which , among others , contribute to a symmetrization of the spectrum [ non - classical branches in fig .
[ fig : laspec ] ( c ) , two of them together with the classical partner branches are marked by arrows ] .
@xmath123 reveals the importance of paths that are geometrically reflected off the open lead ( e.g. , the periodic contributions near @xmath66 along the length axis belong to horizontal paths bouncing increasingly many times back and forth ) .
+ the importance of a given class of paths to quantum transport can be delineated by inverting this decomposition process .
deleting a selected class of paths ( classical or non - classical ) from @xmath131 and performing the inverse fourier transform gives rise to truncated @xmath12-matrix elements @xmath132 from which certain path contributions have been removed in a controlled manner .
this is the key to detailed quantitative tests of semiclassical theories . since summation of the ( p)sca over arbitrarily long paths
is prohibitively complicated we can compare truncated quantum and semiclassical @xmath12-matrix elements where both quantum and classical paths only up to a maximum pathlength @xmath133 are included .
the length - area spectra also allow sensitive tests for the proper diffractive weight of a given class of paths in a semiclassical theory . to this end
, we first calculate the @xmath12-matrix elements within the psca and then perform the fourier transform [ eq .
( [ eq : fouriert ] ) ] in analogy to quantum calculations . to analyze the role of a proper diffraction coefficient we use either the gtd - utd ( which gives good agreement with quantum mechanics , see fig .
[ fig : diff_qm ] ) or the fda ( with poor agreement with quantum mechanics , see likewise fig .
[ fig : diff_qm ] ) .
the fact that back - reflection into the cavity is poorly described within the fda is mirrored in the semiclassical length - area spectrum where important non - classical ( diffractive ) contributions have a far too low weight [ fig .
[ fig : laspec_comp ] ( c ) ] . in particular , the diffractive change of the rotational direction
is insufficiently described ( see fig .
[ fig : diff_qm ] for @xmath134 ) .
a clear indication for the essential role of the corresponding paths is the improvement within the gtd - utd .
the length - area spectrum within the gtd - utd remarkably reproduces even fine details of the quantum mechanical spectrum [ compare fig .
[ fig : laspec_comp ] ( a ) and ( b ) ] . the non - classical path sets in fig .
[ fig : laspec_comp ] can be identified using the path distributions within the psca ( fig .
[ fig : class1kink ] ) .
the length - area spectra do not leave any ambiguity as to which paths contribute by which phase and weight . by taking into account pseudo - paths and weighting them with the appropriate diffraction coefficients
the quantum mechanical path spectrum is reproduced on a quantitative level . in the following section
we demonstrate how the accurate representation of the length - area spectra within psca directly translates into quantitative reproduction of transport properties .
phase coherent ballistic transport governed by quantum interference influences the conductance in several important ways : the conductance strongly fluctuates as a function of the fermi energy , the magnetic field or the cavity geometry ( conductance fluctuations , cf ) .
the conductance is , on average , suppressed compared to the classical prediction and increases with external magnetic field ( weak localization , wl ) .
the noise carries the signatures of quantum mechanical uncertainty ( shot noise ) .
+ path interference has been the key to the understanding of phase coherent ballistic transport.@xcite for chaotic systems and large mode numbers quantum transport properties have been attributed @xcite to the interference of classically allowed paths ( sca ) .
we demonstrate in the following for the circular billiard at low mode numbers ( accessible to experiments ) that cf , wl , and shot noise can not be explained by the interference of classical paths alone .
in this regime , the sca overestimates the cf , does not reproduce the weak - localization dip and also shows poor quantitative agreement with the quantum mechanical prediction for shot noise .
these difficulties can be overcome with the psca when a high - level diffraction approximation , the gtd - utd , is employed . for technical reasons , we perform in both the sca and the psca the summation over paths only up to a maximum cut - off length @xmath135 .
correspondingly , we truncate the full quantum scattering matrix elements by setting all elements @xmath136 with lengths exceeding @xmath135 to 0 and carry out the inverse fourier transform . this allows a quantitative comparison between semiclassical and quantum calculations unaffected by the ( inevitable ) unitarity deficiency of a truncated semiclassical path sum .
+ conductance fluctuations ( cf ) have been identified as a direct manifestation of phase coherent transport ( see e.g. refs . and references therein ) .
the strong fluctuations of the conductance as a function of , e.g. , energy originate from path interference and thus give evidence for the wave nature of electrons in quantum dots .
the cf offer one of the most stringent testing grounds for a semiclassical theory , when good agreement on the level of each individual @xmath12-matrix element is required .
+ in the following we compare the results for cf within the psca with the gtd - utd , the sca and the quantum mechanical calculations as a function of @xmath38 at vanishing magnetic fields @xmath30 .
results for @xmath137 averaged over @xmath38 can be found in sec .
[ sec : wl ] .
+ the semiclassical and quantum mechanical results both display strong fluctuations of the conductance [ i.e. , the total transmission @xmath138 . their amplitude is , however , extremely sensitive to any deficiencies in the semiclassical approximations ( inappropriate weighting of paths , missing paths ) .
we emphasize that the comparison of fig .
[ fig : kdep ] is on a fully differential level .
no energy or ensemble average is involved .
unsurprisingly , the agreement between the sca and the quantum data is poor and on a level of qualitative agreement at best [ fig .
[ fig : kdep ] ( a ) and ( b ) ] . the functional dependence of @xmath139 and @xmath140 seems only weakly related to the quantum mechanical prediction . by ensemble averaging ( e.g. , over a suitable @xmath38 interval ) these discrepancies would be ( partially ) removed ( or masked ) . to correctly reproduce the fluctuations in @xmath141 ,
all individual mode - to - mode amplitudes @xmath142 have to be accurate .
obviously , the contribution of pseudo - paths included in the psca but missing in the sca significantly improves the agreement with the quantum conductance fluctuations [ fig .
[ fig : kdep ] ( c ) and ( d ) ] .
+ even for averaged quantities , such as the averaged conductance @xmath143 and the variance of the conductance fluctuations @xmath144 , the deficiencies of the sca are still visible , in particular at larger cut - off lengths @xmath135 .
both @xmath143 and @xmath145 are overestimated ( fig .
[ fig : mean_cond_var ] ) as compared to quantum mechanics which can be attributed , in part , to the lack of correlation among purely classical paths .
+ in contrast , the psca shows very good agreement with quantum mechanics for both the averaged conductance as well as for the variance . for averaged quantities ( fig .
[ fig : mean_cond_var ] ) the inclusion of up to 4@xmath146 order diffractive scattering for @xmath135 = 40 is sufficient while for fully differential quantities ( fig .
[ fig : kdep ] ) @xmath147 order corrections for @xmath135 = 40 still improve the agreement .
+ we point out that for the present system , taking into account paths up to a length of @xmath148 which corresponds to @xmath149 radial traversals through the billiard , one reaches @xmath150 of the unitarity level within both the truncated quantum mechanics and the psca . the essential contributions to cf , wl and shotnoise ( see the following sections ) is thus rooted in this subset of relatively short paths .
+ weak localization ( wl ) is a well - known quantum correction to the classical diagonal approximation ( see e.g. refs . and references therein ) where the later corresponds to the restriction to terms @xmath151 in the double sum over paths when calculating @xmath152 from eq .
( [ fig : laspec ] ) .
off - diagonal terms @xmath153 give rise to quantum interferences which manifest themselves as an increase of the averaged total reflection @xmath154 at @xmath30 in form of a pronounced peak .
correspondingly , the averaged total transmission @xmath155 features a dip which is an immediate consequence of unitarity . for the investigation of the weak localization dip ( peak ) we employ an average over a small window @xmath156\pi / d$ ] of the @xmath38 dependence of the probabilities : @xmath157 , @xmath158 .
+ since semiclassical theories are , by construction , not necessarily unitary at a given level of approximation , the anti - correlated peak - dip structure near @xmath159 provides a sensitive test for semiclassical approximations .
it has been shown that the quantum anti - correlation between reflection and transmission @xmath160 requires a correlation of transmitted and reflected paths.@xcite this correlation is absent in the sca such that no transmission dip is reproduced [ see fig .
[ fig : wllength ] ( b ) ] .
+ the role of diffraction in a quantum billiard manifests itself by a very good agreement of the psca with quantum mechanical results [ see fig .
[ fig : wllength ] ( a ) ] .
tests for wl as a function of the cut - off length @xmath135 of the paths as well as of the order of diffractive scattering included show that convergence toward the ( truncated ) quantum result is reached for @xmath135 = 40 when diffraction up to fourth order is included .
we note an improvement compared to previous third - order calculations@xcite especially due to the inclusion of the ( small ) real part of the diffraction coefficient @xmath91 .
+ it is instructive to analyze the build - up of the weak localization peak in reflection and of the dip in transmission from individual @xmath12-matrix elements . in reflection
only the diagonal elements @xmath161 and @xmath162 show a peak while the off - diagonal elements @xmath163 and @xmath164 exhibit a dip [ fig .
[ fig : wlpsca_gtdutd ] ( c ) , ( d ) ] .
this is due to the fact that time - reversal symmetric paths contributing to @xmath165 with different parity of modes @xmath14 and @xmath16 acquire an additional phase - shift of @xmath166 . as an example consider @xmath164 , entering in mode @xmath167 and exiting in mode @xmath168 produces a phase shift of @xmath166 .
thus the time - reversal symmetric paths interfere destructively .
+ in transmission the major contribution to a dip originates from @xmath169 in the chosen energy window of two open modes [ fig .
[ fig : wlpsca_gtdutd ] ( e ) , ( f ) ] .
note that @xmath170 because of the onsager relation @xmath171 and the symmetry of the circular billiard which ensures that @xmath172 . due to time - reversal symmetry ,
@xmath173 giving rise to a symmetric peak in @xmath174 as a function of @xmath39 ( fig .
[ fig : wlpsca_gtdutd ] ) . for unitary transport , @xmath175 , which implies a symmetric dip in transmission . for the truncated quantum mechanics and semiclassics where long paths are omitted unitarity is not preserved and @xmath176 is not exactly symmetric but features a slight shift of the minimum toward @xmath177 .
+ overall , the agreement between the psca and the full quantum calculation ( truncated at the same pathlength ) is remarkable ( fig .
[ fig : wlpsca_gtdutd ] ) .
the residual minor deviations are mainly ascribed to deficiencies in the diffraction coefficients for which we use an analytical far - field approximation ( sec . [ sec : diff ] and appendix [ appa ] and [ appb ] ) .
as , e.g. , the diffraction coefficients @xmath178 enter @xmath179 , @xmath180 to the fourth power , a small deficiency in @xmath178 can have a sizeable effect on @xmath179 , @xmath180 .
the offset of the total reflection @xmath174 and transmission @xmath181 within psca [ fig .
[ fig : wlpsca_gtdutd ] ( b ) ] compared to the quantum mechanical result [ fig .
[ fig : wlpsca_gtdutd ] ( a ) ] mainly originates from imperfections in @xmath182 and @xmath183 [ compare fig . [
fig : wlpsca_gtdutd ] ( c ) , ( d ) , and ( e ) , ( f ) ] . the imperfection could be cured by including a correction factor of @xmath184 in @xmath185 .
+ to demonstrate how sensitively transport properties depend on the weights of the contributing paths we show the results for weak localization calculated within psca but now using the fda instead of the gtd - utd for the diffraction coefficients ( fig .
[ fig : wlpsca_fda ] ) .
( we have used @xmath186 order of the psca here , since the deviation between @xmath186 and @xmath187 order is much smaller than the errors due to the simpler diffraction theory . )
it is striking that especially both @xmath182 and @xmath183 seem to be underestimated which is due to deficiencies in @xmath188 within the fda .
it is worthwhile pointing out that the fda diffraction could not be `` repaired '' by a correction factor , as the errors in @xmath182 and @xmath183 are different . in other words , transmission is more `` sensitive '' to a correct implementation of pseudo - paths than reflection .
of particular importance are paths which change the rotational direction.@xcite since the fda does not give sufficiently weight to this class of pseudo - paths ( see fig .
[ fig : diff_qm ] ) the transmission dip can not be well reproduced .
another quantity characteristic for quantum transport is the quantum shot noise power of the current ( see e.g. refs . and references therein ) .
at zero temperature ( @xmath189 ) , the time - dependent current noise is due to the granularity of the electron charge and carries information about the wave vs. particle nature of charge transport .
the fano factor @xmath190 measures the amount by which the noise in phase coherent transport is suppressed relative to the poissonian value of uncorrelated classical electrons .
within the landauer - bttiker picture , @xmath190 can be expressed as@xcite @xmath191 with @xmath192 , @xmath193 being the eigenvalues of the hermitian matrices @xmath194 and @xmath195 , respectively .
+ calculating the shot noise power from non - unitary scattering matrices is obviously a delicate matter , as replacement of @xmath196 by @xmath197 leads , unlike for unitary descriptions , to different results .
furthermore , such a replacement may result in negative and thus unphysical values for the shot noise power , as non - unitary scattering matrices allow for the possibility of having @xmath198 such that @xmath199 ( for very high mode numbers as , e.g. , in ref .
such a situation may , however , be unlikely ) . by using eq .
( [ eq : fanoshot ] ) , such difficulties can be avoided as both the transmission and reflection eigenvalues @xmath200 are , by construction , real and positive not only for the truncated quantum calculation but also for the psca and the standard sca .
the standard sca result for @xmath190 strongly deviates from the quantum mechanical data ( see fig . [
fig : fano ] ) . for small cut - off lengths @xmath135 ,
the value of @xmath190 is smaller but increases more rapidly than the quantum mechanical result with @xmath135 .
the psca yields very good agreement with the quantum mechanical result for the shot noise fano factor @xmath190 .
note that for the largest cut - off length @xmath148 the fano factor @xmath190 is already converged to its asymptotic value @xmath201 suggesting that long paths do not play a significant role for @xmath190 .
this result agrees with the finding@xcite that the shot noise power is of similar magnitude for regular and chaotic billiards as differences in the dynamics are most strongly felt by very long paths .
we have presented a semiclassical theory which is able to _ quantitatively _ reproduce full quantum results for scattering through microstructures with a specific geometry , in the present case a circular shaped billiard with leads oriented 90@xmath202 degrees relative to each other .
the present approach does not invoke the limit of large mode numbers , ( where the de broglie wavelength @xmath0 is small relative to the lead width ) but requires @xmath0 to be small only on the scale of the linear dimension of the microstructure ( the circle ) .
this non - asymptotic semiclassical theory allows a direct comparison with quantum calculations as well with experiments on a system - specific level for individual @xmath12-matrix elements avoiding any ensemble averaging or fit parameters .
this level of agreement allows us to perform detailed semiclassical investigations of quantum transport quantities such as the conductance fluctuations , the weak localization , and the shot noise .
our studies show unambiguously that for reproducing these quantities correctly , two major ingredients are , indeed , crucial : ( 1 ) the inclusion of `` pseudo - paths '' in the semiclassical propagator which are diffractively backscattered from the interior side of the cavity openings , ( 2 ) a sufficiently accurate description of the diffraction coefficients for the injection , ejection and back - reflection of particle flux at the cavity openings .
we meet the latter requirement by developing a combined geometric and uniform theory of diffraction ( gtd - utd ) .
pseudo - paths are crucial for reproducing the conductance fluctuations in transport and lead to a reduction of its variance .
also for the weak - localization effect we find that pseudo - paths are crucial , as no signature of weak localization appears in the transmission through the circular cavity without their contribution ( even when the advanced diffraction theory is employed for all truly classical paths ) .
for the shot noise power we showed that a standard semiclassical calculation ( without pseudo - paths ) gives sizeable discrepancies .
the inclusion of pseudo - paths leads to agreement with the quantum mechanical result .
we emphasize that the parameter regime in which we have identified the above effects of pseudo - paths coincides with the typical situation in quantum transport experiments .
the latter usually feature only a few open lead modes @xmath25 .
+ the present results raise several interesting questions for future investigations : the comparison between the semiclassical approximations ( psca and sca ) and full quantum calculations were performed for truncated path sums up to a finite path length @xmath203 .
the primary reason for the truncation was technical , as the number of diffractive pseudo - paths exponentially proliferates with @xmath204 also for classically regular structures and exact path sums become prohibitively difficult to perform .
there is , however , a second conceptual motivation . in the experiment ,
decoherence due to inelastic scattering limits phase - coherent transport to pathlengths @xmath205 , where the phase - decoherence mean free path @xmath206 typically allows only a moderate number of traversals across the cavity .
the latter restriction rules out that very long paths with @xmath207 contribute to quantum interference in the experiment , a feature which is naturally incorporated by way of the cut - off length @xmath135 in our semiclassical theory .
clearly , such long paths can still provide incoherent contributions .
the present approach may thus contribute to a semiclassical unterstanding of decoherence effects in regular cavities.@xcite + in the present treatment of diffractive scattering , both internal diffraction at the open lead mouth giving rise to pseudo - paths as well as the coupling between leads and cavity was performed for sharp edges .
the weight of diffractive contributions can be changed by `` rounding off '' the lead opening .
an investigation of the dependence of the weak localization on the smoothness of the edges is currently underway .
the introduction of rounded corners has , however , another profound effect , apart from changing the weight of diffractive scattering : an open circle with rounded edges of the leads is no longer regular but features a mixed phase space .
this raises the question as to the interplay between diffractive scatterings at the lead opening and chaotic scattering in the interior of the billiard . for generic chaotic systems
a number of alternative semiclassical theories has been proposed to explain universal features of quantum transport ( see e.g. refs . and references therein ) .
these theories typically employ an ensemble average and rely on a @xmath208 limit which makes them complementary to the present system - specific approach for finite @xmath41 .
bridging the gap between these two frameworks would be of great interest .
one key ingredient would be to clarify the interplay between the diffraction - based pseudo - paths and the chaos - based correlated classical path pairs ( richter - sieber orbits@xcite ) .
the relative weight for characteristic quantum transport effects carried by these two classes of paths when both are present ( as in a chaotic billiard with sharp edges ) remains an open question .
we thank piet brouwer , tobias dollinger and klaus richter for helpful discussions .
this work was supported by the austrian fwf under grants no .
fwf - sfb016 `` adlis '' and no . 17359 , and the fwf doctoral program `` complex quantum systems '' .
l.w . acknowledges support by the phc `` amadeus '' of the french ministry of foreign and european affairs .
in a quantum billiard , the electron propagation in the leads is determined by a quantum mechanical wave , while the propagation inside the ballistic cavity can be described semiclassically , i.e. , by propagation along classical trajectories with a quantum phase .
the strong scattering effects that occur especially in the low mode regime at the orifices have been described in past works on transport through open billiards by the kirchhoff diffraction approximation ( kda ) @xcite or on the level of the fraunhofer diffraction approximation ( fda).@xcite in the high - mode regime diffraction has been neglected altogether.@xcite both the kda and the fda perform about equally well ( see , e.g. , fig . [
fig : diff_qm ] ) . for the identification of pseudo - paths in the fourier spectra of the conductance fluctuations , the kda and
the fda have been sufficiently precise .
however , in order to recover unitarity of the semiclassical @xmath12-matrix and weak - localization in transmission , a more precise diffraction theory must be implemented . for multiple - scattering paths , higher order products of the diffraction weights occur and even small errors are rapidly amplified .
we have therefore implemented a combination of the geometric theory of diffraction ( gtd ) @xcite and the uniform theory of diffraction ( utd ) @xcite for the diffraction coefficients in open billiards referred to in the following as the gtd - utd .
both theories have been previously applied separately to the calculation of higher - order scattering corrections to gutzwiller s trace formula in closed quantum - billiards.@xcite we discuss in section [ appa ] diffractive backscattering at a semi - infinite half - plane with an orifice and in section [ appb ] diffraction during propagation from a lead into a semi - infinite plane . + and @xmath95 in a semi - infinite plane in the presence of an open ( lead ) .
contributions from the direct path ( black dotted line ) , from specular reflection ( blue solid line ) , and from scattering at the orifice ( red dashed line ) are depicted.,width=188 ] the propagation between two points in a semi - infinite plane with a connected lead is depicted in fig .
[ fig : half_inf ] . in the absence of a lead ,
the propagation between two points @xmath94 and @xmath95 in a semi - infinite plane is described by a sum of two green s functions , @xmath209 .
the first contribution corresponds to the direct path from @xmath94 to @xmath95 ( black dotted line ) and is in the semiclassical approximation [ eq .
( [ eq : gfree ] ) ] given by @xmath210 where @xmath211 ) is the distance between the two points and @xmath212 is the maslov index ( @xmath213 in this case ) .
the second term corresponds to a propagation via a classical , specularly reflected path ( blue solid line ) : @xmath214- i3\pi/4-i\pi } } { \sqrt{2\pi k \left[l(\vec{r}\,',\vec{a})+l(\vec{a},\vec{r}\,)\right]}}.\ ] ] the orifice gives rise to an additional scattered wave for which we assume in the far - field limit ( @xmath215 and @xmath216 ) a cylindrical wave emanating from the center of the orifice ( point @xmath217 ) .
invoking far - field approximations is at the heart of semiclassical diffraction theories , the validity of which need testing on a case - by - case basis ( see below ) .
+ the diffraction contribution to the green s function ( red dashed line in fig . [
fig : half_inf ] ) is @xmath218 we refer to this term as the ( first - order ) pseudo - path semiclassical contribution , because it corresponds to a classically forbidden path .
we note the different scaling of @xmath219 and @xmath220 with @xmath38 .
the amplitude of @xmath220 scales as @xmath221 while the diffractive contribution scales as @xmath222 . at large distances ( far - field ) and/or large @xmath38 , the geometric reflection amplitude dominates over the diffraction amplitude , as expected .
+ the diffraction coefficient @xmath91 as a function of the incoming and outgoing angles has been calculated in the past in the kirchhoff ( kda ) and fraunhofer diffraction approximations ( fda ) .
both approaches have in common that the amplitude of the cylindrical diffractive wave emanating from the orifice is determined from an integration over the orifice using as a source the amplitude and the phase of the unperturbed incoming wave in the absence of the boundary .
in other words , they are implementations of huygens principle according to which each point of the lead opening is the source of an outgoing circular wave . for details , we refer the reader to ref .
( for the kda ) and to ref .
( for the fda )
. + keller s geometric theory of diffraction ( gtd)@xcite has a different point of departure : it originates from the far - field approximation of the exact sommerfeld s solution for the wave scattering at a wedge @xcite [ see fig . [
fig : diff_wedge ] ( a ) ] . in the following
we discuss the application of the gtd and its refinement within the framework of the uniform theory of diffraction utd @xcite to the problem of the diffractive scattering at the lead mouth [ figs .
[ fig : diff_wedge ] ( b ) and ( c ) ] .
starting point of our determination of the diffraction coefficient @xmath223 is the decomposition of the orifice into two wedges with an inner angle of @xmath224 and outer angle of @xmath225 [ see fig . [
fig : diff_wedge ] ( a ) ] . in the far - field limit , the incident wave can be regarded asymptotically as a plane - wave .
the gtd describes the scattering of a plane wave at an infinitely sharp wedge @xcite [ fig .
[ fig : diff_wedge ] ( a ) ] .
the total wavefunction is a sum of the incoming plane wave , the reflected plane wave and an outgoing cylindrical diffracted wave , emanating from the edge .
is negative and @xmath93 is positive ) .
the points @xmath94 and @xmath95 are assumed to be in the far field region .
( c ) higher - order corrections enter via paths scattered between the wedges.,width=309 ] there are two discontinuities : at the shadow boundary [ @xmath226 , for the definition of the angles see fig .
[ fig : diff_wedge ] ( a ) ] and the boundary of geometric reflection ( @xmath227 ) . outside a small region around these two angles ,
the diffracted wave can be described by an outgoing cylindrical wave modulated by a smooth diffraction coefficient .
the diffractive part of the green s function between the two points @xmath94 and @xmath95 in fig .
[ fig : diff_wedge ] ( a ) can thus be approximated as : ( atomic units ) in eq .
( [ gpsca ] ) originates from @xmath228 ( si units ) and assures that the green s function is correctly normalized [ such that @xmath229 .
we use the same normalization as in ref . ,
section 7.5.4 . ] @xmath230 with @xmath231 , \label{eq : d}\end{aligned}\ ] ] where @xmath232 is the exterior angle ( in units of @xmath166 ) of a perpendicular wedge and @xmath233 is the position of the corner of the wedge [ fig .
[ fig : diff_wedge ] ( a ) ] .
+ we now consider the lead opening as being composed of two wedges [ fig .
[ fig : diff_wedge ] ( b ) ] .
the obvious conceptual difficulty lies in the fact that the two wedges are , in general , not in the far - field limit @xmath234 of each other . with this caveat in mind , diffraction at the lead
can be considered within the gtd to result from the interference of two paths that are diffracted at the corners of the two wedges limiting the lead .
summing up the diffraction weights and phases from the two paths , we obtain the green s functions of the reference pseudo - path as in eq .
( [ pscadef ] ) but with the gtd reflection coefficient @xmath235 in eq .
( [ eq : vgtd ] ) we have neglected the difference of the incoming angles at the left and at the right wedge , respectively , i.e. , we set @xmath236 [ fig .
[ fig : diff_wedge ] ( c ) ] .
likewise , for the outgoing angles , we set @xmath237
. the phase differences of the right / left pseudo - path with respect to the reference path emanating from the center of the orifice can then be written in linear approximation with respect to the transverse lead coordinate as @xmath238 .
the coefficients in eq .
( [ eq : vgtd ] ) are defined as @xmath239 we note that both @xmath240 and @xmath241 have a singularity at the reflection boundary for @xmath242 .
however , in the sum of the two terms , the singularities cancel out and the resulting backscattering amplitude is perfectly smooth ( see fig . [
fig : diff_convergence ] , red chain dotted line ) .
+ the comparison of the gtd with the exact quantum calculation for the diffraction coefficient , @xmath243 ( fig .
[ fig : diff_convergence ] ) , reveals sizeable deviations .
two deficiencies are noteworthy : the almost complete missing of the back - reflection peak and the failure at grazing angles @xmath244
. the diffraction coefficient should approach zero in this limit but , instead , converges toward a finite value ( fig .
[ fig : diff_convergence ] ) .
this behavior originates from treating the diffraction at the lead as two independent local phenomena of diffraction at two separate wedges .
+ the diffraction at a lead can be treated within the utd by using a double - wedge diffraction coefficient ( see ref . and a more recent paper for arbitrary configurations of the wedges , ref . ) .
the double - wedge diffraction coefficient can not be separated into a sequence of single - wedge diffraction coefficients and contains rather involved mathematical expressions such that the beauty and structural simplicity of a semiclassical approach is lost .
we present in the following an ansatz for double - wedge diffraction which assumes a separation of the diffraction process into a sequence of diffraction events and we verify its validity by comparing with quantum mechanical results from ref . .
we show that the shortcomings of eq .
[ eq : vgtd ] can be ( to a large extent ) remedied by taking diffractive paths of higher order into account , i.e. paths that pass between the two wedges once or several times .
this drastically improves the agreement with the quantum mechanical result .
( we point to the conceptual similarity of our approach to the treatment of double - wedge diffraction illuminated by transition region fields by a sum over higher - order diffracted fields.@xcite ) in a first step , we include paths that scatter once between the two edges [ see green line fig .
[ fig : diff_wedge ] ( c ) ] .
there are two such paths .
one approaches the right wedge at angle @xmath62 .
it is scattered into the angle @xmath224 ( with respect to the surface normal ) , and at the left wedge , it is scattered into the angle @xmath93 .
the other path is scattered from the left wedge to the right one with the same entrance and exit angles .
the weights of this pair of paths can not be determined from the gtd diffraction coefficients [ eq . ( [ eq : d ] ) ] : the gtd diffraction coefficient fails in the limit of @xmath245 and @xmath246 [ definition of angles as in fig . [
fig : diff_wedge ] ( a ) ] , because this is in proximity of the shadow boundary , into which the horizontal paths are scattered .
this problem can be overcome by invoking the uniform theory of diffraction ( utd ) .
+ contrary to the gtd , the utd is also valid on the zone boundaries .
the outgoing cylindrical wave is multiplied by a diffraction coefficient which depends not only on the two angles @xmath247 and @xmath248 but also on the distances @xmath249 and @xmath250 , and on the wavenumber @xmath38 , @xmath251 where @xmath252 and @xmath253 is the integer which most closely satisfies @xmath254 .
the function @xmath190 is defined as a generalized fresnel integral : @xmath255 and has the asymptotic form @xmath256 for @xmath257 , i.e. , for large distances and outside the transition zones , @xmath258 , and @xmath259 reduces to @xmath260 . within the transition zone
, the distance dependence of @xmath261 leads to a deviation of the scattered wave from a purely cylindrical wave .
this is necessary to ensure the continuity of the total wave - function at the zone boundaries .
furthermore , by using the utd diffraction coefficient we partially take into account the fact that the two wedges are not in the far - field region with respect to each other .
+ since the gtd fails for large ( near grazing ) angles , we opt for a piecewise construction : we combine gtd and utd referred to in the following as gtd - utd such that the diffraction of the path with the smaller ( absolute value ) of the angle is treated by the gtd and the path with the larger angle on the level of the utd .
accordingly , to first order the gtd - utd correction to the diffraction coefficient can be written as @xmath262 where [ in analogy to eq .
( [ eq : gtdampl ] ) ] we have defined the utd diffraction coefficients at the left ( l ) and right ( r ) wedge as @xmath263 in eq .
( [ eq : vgtdutd1 ] ) , the diffraction of the incoming path ( with angle @xmath62 ) , is treated on the gtd level and the diffraction of the outgoing path ( angle @xmath93 ) on the utd level .
this construction introduces a first - order discontinuity ( `` kink '' ) at @xmath264 .
this kink is , however , negligible for small angles and visible only at large angles close to @xmath224 , in other words the formula breaks down in the limit @xmath265 , see , e.g. , fig .
[ fig : diff_qm ] ( c ) .
( one could alternatively ignore the fact that the utd is not multiplicative in the near - field and employ for both scattering events utd , thus avoiding the kink .
this ansatz , however , breaks down similarly for grazing incidence : the diffraction coefficient does not approach zero for the outgoing angle @xmath266 . for small and medium angles
this approach behaves equally well as the presented gtd - utd approach . ) at large incident and/or outgoing angles the diffraction coefficient is already strongly suppressed such that the resulting error is small .
+ the inclusion of the first order gtd - utd correction ( @xmath267 , see fig .
[ fig : diff_convergence ] ) already considerably improves the agreement with the quantum diffraction pattern .
higher order diffraction corrections include paths that are scattered several times between the wedges .
this includes paths that are incident and backscattered at an angle @xmath268 at the wedge .
this is exactly on the reflection boundary . in this limit
, reiche has shown @xcite that the diffraction pattern of a plane wave with unit amplitude incident on the wedge with an angle of @xmath268 reduces to a reflected plane wave with amplitude @xmath269 and a cylindrical wave .
the green s function of a higher order path which scatters @xmath17 times between the wedges is , therefore , a product of the gtd diffraction coefficient , the utd diffraction coefficient , and the green s function for free propagation along the distance @xmath270 , acquiring a factor @xmath269 and a phase of @xmath166 for each reflection at a wedge . summing the diffraction corrections up to order @xmath271 we obtain where @xmath273 and @xmath274 fig .
[ fig : diff_convergence ] demonstrates that the diffraction coefficient @xmath275 is converged for @xmath276 .
furthermore , the condition @xmath277 for @xmath265 is fulfilled and the agreement with the fully quantum mechanical backscattering weight is excellent .
the agreement deteriorates for increasing entrance angles @xmath62 but is still satisfying compared to the simple fraunhofer diffraction approximation ( fda ) , ( see fig .
[ fig : diff_qm ] ) .
fortunately , large angles do not play an important role because the overall diffraction weight is very low .
the gtd - utd provides a remarkable compromise between simplicity and accurate representation of quantum mechanical results for diffraction at a lead attached to a semi - infinite half - plane .
+ coupled to a half - infinite plane .
an incoming wave in mode @xmath14 can be separated into two rays @xmath278 which diffractively scatter at the lead wedges .
the reference path is denoted by a red dashed line.,width=226 ] the flux - normalized wave - function for mode @xmath14 in a lead of width @xmath3 oriented parallel to the @xmath279-axis is @xmath280 , \label{decomp}\end{aligned}\ ] ] where @xmath281 is the transverse momentum component and @xmath282 the longitudinal component .
( [ decomp ] ) can be viewed as two rays emanating with angles @xmath283 ( fig .
[ fig : difflead ] ) . in the gtd approximation
, each ray hits an edge of the lead mouth and the two cylindrical waves emanating from the edges cause , in turn , an interference pattern at large distances . the gtd diffraction for scattering from the lead mode @xmath14 into the half - space ( in our application the cavity ) can be written in direct analogy with eqs .
[ eq : vgtd ] and [ eq : gtdampl ] as @xmath284 , \label{eq : dngtd}\end{aligned}\ ] ] with the scattering coefficients at left and right wedge @xmath285 the two paths have a phase difference of @xmath286 at the lead mouth and thus a phase difference @xmath287 relative to the reference path that starts at the center of the orifice ( see fig . [
fig : difflead ] ) . with the flux normalization factor @xmath288 from eq .
( [ eq : tnm ] ) @xmath289 is dimensionless . + we include now higher - order scattering events on the level of the utd in order to improve eq .
( [ eq : dngtd ] ) , where @xmath291 @xmath292 is given in eq .
( [ gkj ] ) and @xmath293 , @xmath294 are given by eq .
( [ eq : utd ] ) .
[ fig : diff_cm ] illustrates that the utd corrections become most important when the value of @xmath38 is close to a threshold , i.e. , when the angle @xmath295 is close to @xmath224 .
|
we investigate electron transport through clean open quantum dots ( `` quantum billiards '' ) .
we present a semiclassical theory that allows to accurately reproduce quantum transport calculations .
quantitative agreement is reached for individual energy and magnetic field dependent elements of the scattering matrix .
two key ingredients are essential : ( i ) inclusion of pseudo - paths which have the topology of linked classical paths resulting from diffraction in addition to classical paths and ( ii ) a high - level approximation to diffractive scattering . within this framework of the pseudo - path semiclassical approximation ( psca )
, typical shortcomings of semiclassical theories such as violation of the anti - correlation between reflection and transmission and the overestimation of conductance fluctuations are overcome . beyond its predictive capabilities
the psca provides deeper insights into the quantum - to - classical crossover .
|
nucleation is a fluctuation - driven process that initiates the decay of a metastable state into a more stable one @xcite .
it is usually involved in first - order phase transitions and along with growth of a new phase @xcite .
many important phenomena in nature , including crystallization @xcite , fractures @xcite , glass formation @xcite , and protein folding @xcite , to list just a few , are associated with nucleation . despite much attention , many aspects of nucleation processes in complex systems are still unclear and deserve more investigation . the ising model is a paradigm for many phenomena in statistical physics .
it has also been widely used to study the nucleation process .
for instance , in two - dimensional lattices , allen _ et al _ discovered that shear can enhance the nucleation rate and the rate peaks at an intermediate shear rate @xcite .
sear found that a single impurity may considerably enhance the nucleation rate @xcite .
page and sear reported that the existence of a pore may lead to two - stage nucleation , and the overall nucleation rate can reach a maximum level at an intermediate pore size @xcite . the nucleation pathway of the ising model in three - dimensional lattices has also been studied by sear and pan @xcite . in addition , the ising model has been frequently used to test the validity of classical nucleation theory ( cnt ) @xcite .
nevertheless , all these studies are limited to regular lattices in euclidean space .
since many real systems can be properly modeled by network - organized structure @xcite , it is thus an interesting topic to explore nucleation process in complex networks .
very recently , our group have studied nucleation dynamics on scale - free ( sf ) networks @xcite and modular networks @xcite . in these two papers
, we mainly focused on the nucleation rate and system size effects .
we found that , for sf networks , the nucleation rate decays exponentially with network size , while the critical nucleus size increases linearly .
for modular networks , as the network modularity worsens the nucleation undergoes a transition from a two - step to one - step process and the nucleation rate shows a nonmonotonic dependence on the modularity .
as we know , network topology could play important role in the system s dynamics , involving not only the stationary state , but also the detailed pathway .
for instance , it was shown that network heterogeneity could drastically influence the path to oscillator synchronization @xcite .
nevertheless , how network topology would influence the nucleation pathway has not been reported yet .
motivated by this , we will study the different roles of network architectures in the formation of nucleating clusters , which can reveal the nucleation pathways of the ising model in the underlying networks . since nucleation is an activated process
, it can be extremely slow .
therefore , direct simulations can take excessive amounts of time . to overcome this difficulty , in the present work , we adopt the forward flux sampling ( ffs ) @xcite approach proposed recently , which is efficient and easy to implement to study rare events .
we employ erds - rnyi ( er ) and sf networks as the paradigm of homogeneous and heterogeneous networks respectively . by using ffs
, we obtain lots of configurations at each interface along the nucleation pathway . from these configurations we scrutinize and compare the nucleating clusters in er and sf networks .
it is found that the processes of forming the critical nucleus are qualitatively different between the two cases of networks .
for the former , a dominant cluster arise firstly , and groups smaller clusters gradually , while for the latter , many small clusters emerge at first and then abruptly turn into the critical nucleus .
interestingly , both the cluster size distributions follow power - law distributions and the slopes are nearly the same at early nucleation stage . the paper is organized as follows .
section [ sec2 ] presents the details of our simulation model and the numerical methods we employ to sampling the nucleation pathway .
the numerical results are compared for sf networks and er ones in sec .
a brief summary is given in sec .
we consider the ising model on complex networks consisting of @xmath0 nodes .
each node is endowed with a spin variable @xmath1 that can be either @xmath2 ( up ) or @xmath3 ( down ) .
the hamiltonian of the system is given by @xmath4 where @xmath5 is the coupling constant and @xmath6 is the external magnetic field . for convenience , we set @xmath7 in the following discussions .
the elements of the adjacency matrix of the network take @xmath8 if nodes @xmath9 and @xmath10 are connected and @xmath11 otherwise .
the degree , that is the number of neighboring nodes , of node @xmath9 is defined as @xmath12 the system evolves in time according to single - spin - flip dynamics with metropolis acceptance probabilities @xcite , in which we attempt to flip each spin once , on average , during each monte carlo ( mc ) cycle . in each attempt ,
a randomly chosen spin is flipped with the probability @xmath13 , where @xmath14 with @xmath15 being the boltzmann constant and @xmath16 the temperature , and @xmath17 is the energy change due to the flipping process . in the absence of an external magnetic field ,
the system undergoes an order - disorder phase transition at the critical temperature . above the critical temperature , the system is disordered where up- and down - pointing spins are roughly equally abundant . below the critical temperature ,
the system prefers to be in either of the two states : one state with predominantly up spins , and the other with almost down spins . in the presence of an external field ,
one of these two states becomes metastable , and if initiated predominantly in this metastable state , the system will remain for a significantly long time before it undergoes a nucleation transition to the thermodynamically stable state .
we are interested in the pathways for this transition .
the ffs method has been successfully used to calculate rate constants and transition paths for rare events in equilibrium and nonequilibrium systems @xcite . for clarity , we describe the method again here , together with some relevant details with our work .
this method uses a series of interfaces in phase space between the initial and final states to force the system from the initial state @xmath18 to the final state @xmath19 in a ratchetlike manner .
before the simulation begins , an reaction coordinate @xmath20 is first defined , such that the system is in state @xmath18 if @xmath21 and it is in state @xmath19 if @xmath22 .
a series of nonintersecting interfaces @xmath23 ( @xmath24 ) lie between states @xmath18 and @xmath19 , such that any path from @xmath18 to @xmath19 must cross each interface without reaching @xmath25 before @xmath23 .
the algorithm first runs a long - time simulation which gives an estimate of the flux escaping from the basin of @xmath18 and generates a collection of configurations corresponding to crossings of interface @xmath26 . the next step is to choose a configuration from this collection at random and use it to initiate a trial run which is continued until it either reaches @xmath27 or returns to @xmath26 .
if @xmath27 is reached , the configuration of the end point of the trial run is stored .
this process is repeated , step by step , until @xmath28 is reached . for more detailed descriptions of the ffs method
please see ref .
@xcite . in this work
, we will use ffs to study nucleation pathways of the equilibrium phase from the metastable spin phase . specifically , we let @xmath29 and start from an initial state with @xmath30 for most of the spins .
we define the order parameter @xmath20 as the total number of up spins in the network .
the spacing between adjacent interfaces is fixed at 3 up spins .
we perform @xmath31 trials per interface for each ffs sampling , from which at least @xmath32 configurations are saved in order to investigate the statistical properties along the nucleation pathway .
the results are obtained by averaging over @xmath33 independent ffs samplings and @xmath34 different network realizations .
in what follows , we employ a barabsi - albert sf network , whose degree distribution follows a power law @xmath35 with the scaling exponent @xmath36 @xcite , and the well - known er random network . in fig .
[ fig1 ] , we present schematically the evolution of local nucleating clusters in er and sf networks at different stages ( for clarity only show 100 nodes ) . here
, a nucleating cluster is defined as the component of connected nodes with up spins .
qualitatively , it shows distinct features along nucleation stages . in the er case
, there always exists a dominant cluster , which groups smaller ones gradually . while for sf networks , no dominant cluster appears at the early stage , but then a giant cluster emerges suddenly .
this demonstrates that nucleation follows different pathways on er and sf networks , indicating that heterogeneity of the network topology may play an important role .
for the two different topologies studied ( er and sf ) .
these shown networks contain 100 nodes , in order to have a sizeable picture of the system .
up spins and down spins are indicated by green circles and black circles , respectively .
[ fig1 ] ] to further elucidate detailed characteristics along the nucleation pathway , we use ffs to generate configurations and perform detailed analysis on the nucleating clusters , including the largest cluster size , average degree of the cluster nodes , the number of clusters and cluster size distribution . according to cnt
, there exists a critical nucleus size @xmath37 of the new phase , above which the system grows rapidly to the new phase .
herein , we mainly focus on the nucleation stage where @xmath38 . in our simulation
, we determine @xmath37 by computation of the committor probability @xmath39 , which is the probability of reaching the thermodynamic stable state before returning to the metastable state .
as commonly reported in the literature @xcite , the critical nucleus appears at @xmath40 .
since @xmath37 are different for different networks , we thus introduce @xmath41 as the control parameter .
for consistent comparison , we introduce @xmath42 as the ratio of the size of the largest nucleating cluster to over the total number of up spins , and plot @xmath42 ( averaged over the ensemble at each interface ) as functions of @xmath43 in fig.2 .
clearly , one can see that @xmath42 for er networks is always larger than that for sf ones .
specifically , at @xmath44 , @xmath42 is already more than @xmath45 for er networks , while it is only about @xmath46 for sf ones , as shown by the dashed gray lines in fig.2 .
but when @xmath47 they almost tend to @xmath48 together . of the largest cluster as functions of @xmath43 .
parameters are @xmath49 , the average network degree @xmath50 , @xmath51 , @xmath52 , @xmath53 and @xmath54 .
[ fig2 ] ] to show our results more explicitly , we investigate the average degree @xmath55 of the nodes inside the nucleating clusters , and plot @xmath55 as functions of @xmath56 in fig .
as shown , @xmath55 increases monotonically with @xmath43 for both er and sf networks , which means the new phase tends to grow from those nodes with smaller degrees . for er networks
, @xmath55 grows fast at the very beginning following by a relatively slow increasing .
for sf networks , @xmath55 increases slowly at first and jumps sharply when approaching the critical nucleus .
such a scenario is consistent with fig.[fig1 ] and fig.[fig2 ] . of new phase nodes as functions of @xmath43 .
other parameters are the same as in fig.2 .
[ fig3 ] ] to better understand the aforementioned differences , we present the number @xmath57 of the nucleating clusters as functions of @xmath58 in fig .
[ fig4](a ) .
we observe that @xmath57 non - monotonically depends on @xmath43 and the numbers of clusters in sf networks are always more than that in er ones .
on the other hand , @xmath57 for both networks approach the same magnitude near the formation of critical nucleus , but it decreases much more sharply in sf networks which is also consistent with the picture shown in fig.[fig1 ] to [ fig3 ] . in fig .
[ fig4](b ) , the cluster size distributions @xmath59 for @xmath60 and @xmath61 are shown .
interestingly , @xmath59 follow apparent power - law distributions in the small size range for both types of networks , and in addition , the exponents are nearly the same for fixed @xmath41 .
the power law breaks in the large size range , where large clusters dominate . of nucleating clusters as functions of @xmath43 .
( b ) size distribution @xmath59 of nucleating clusters , on a log - log scale , @xmath62 denote the size of nucleating clusters .
other parameters are the same as in fig.2 .
[ fig4 ] ] the above results can be qualitatively understood in terms of cnt .
cnt assumes that the formation of a nucleus lies in two competing factors : the bulk energy gain of creating a new up spin which favors the growth of the nucleus , and the interfacial energy cost , an opposing factor , which is due to the creation of new boundary links between up and down spins .
that is , the change of free energy @xmath63 may be written as @xmath64 , where @xmath65 denotes the effective interfacial free energy , which mainly depends on the average number of boundary links that an up - spin node has . obviously , a node with more boundary links is more difficult to change its spin state .
for sf networks , it is thus always easier for the leaf nodes with small degrees to change state than the hubs with large degrees . since the degree distribution follows power - law ,
there exist a lot of hubs with intermediate degrees , as well as a few hubs with very large degrees .
usually , many leaf nodes are connected to relatively small hubs , which are further connected to large hubs .
therefore , only small nucleating clusters , consisted of leaf nodes and small hubs , can form at the early stage of the nucleation process .
these small clusters are either away from each other on the network or separated by those crucial hubs with very large degrees . in the final stage of the nucleation ,
once the crucial hubs connecting these small clusters change their states , a giant nucleation cluster will emerge abruptly .
this picture is consistent with those results shown in the above figures . for er networks , however , the degree distribution follows poisson distribution and no crucial hub exists , such that those new - formed clusters are usually connected together and one would not expect a sharp increase in the cluster size , which is observed in sf networks .
in summary , we have studied nucleation pathways of the ising model with metropolis spin - flip dynamics in er and sf networks using the ffs method .
concerning the former , there always exists a dominant cluster which groups small clusters gradually until the critical nucleus is formed ; while concerning the latter , many isolated small clusters grow separately which suddenly merge together into the critical nucleus .
we have performed detailed analysis involving the nucleating clusters along the nucleation pathway , including the cluster size as well as its distribution , the mean degree inside the cluster , and so on , to further demonstrate the above scenario .
the distinct nucleation pathways between er and sf networks further emphasize the very role of network topology .
our study may provide a valuable understanding of how first - order phase transitions take place on complex networks , which could be of great importance not only for physical systems , but also for social and biological networks .
this work was supported by the national natural science foundation of china ( grant nos.21125313 , 20933006 and 91027012 ) .
was also supported by the key research foundation of higher education of anhui province of china ( grant no.kj2012a189 ) .
|
identifying nucleation pathway is important for understanding the kinetics of first - order phase transitions in natural systems . in the present work , we study nucleation pathway of the ising model in homogeneous and heterogeneous networks using the forward flux sampling method , and find that the nucleation processes represent distinct features along pathways for different network topologies . for homogeneous networks
, there always exists a dominant nucleating cluster to which relatively small clusters are attached gradually to form the critical nucleus . for heterogeneous ones ,
many small isolated nucleating clusters emerge at the early stage of the nucleation process , until suddenly they form the critical nucleus through a sharp merging process . by analyzing the properties of the nucleating clusters along the pathway ,
we show that the main reason behind the different routes is the heterogeneous character of the underlying networks .
|
it is well known that there is an intimate connection between the shell structure in quantum systems like nuclei and metallic clusters and the properties of classical trajectories within these systems ( see e.g. , p. 579 ) . in particular , for nuclei it has been argued that closed orbits of triangular shape might lead to an instability of the spherical shape against octupole - type deformations in the region beyond @xmath3pb , p.560 . here
we would like to investigate in detail this possibility by using a semiclassical theory of nuclear response based on the linearized vlasov equation @xcite .
instabilities are expected to show up as some kind of pathological behaviour in response functions ( vanishing eigenfrequencies and diverging response ) and the semiclassical theory of @xcite and @xcite , that has already been shown to give good qualitative results for lower multipolarities @xcite , is an ideal tool to study the role played by classical trajectories in determining the response of large quantum systems . of course
quantum corrections are expected to modify the results of this theory @xcite , especially at very low energy , however having a clear picture of what should be expected already at the classical level might help in making progress .
we study the isoscalar octupole response of a sample `` nucleus '' of @xmath4 nucleons contained in a square - well potential of radius @xmath5 , with @xmath6 .
this is done with different degrees of approximation : in a first approximation , discussed in sect .
ii , the interaction between nucleons is neglected , while in sect .
iii this interaction is taken into account in a separable approximation . finally in sect .
iv the effect of surface vibrations on the octupole response function is included .
because of an interesting scaling property of the zero - order semiclassical response , the @xmath7-dependence can be factorized and becomes trivial .
thus , although we perform explicit calculations only for @xmath4 , our results on stability apply also to the region beyond @xmath3pb .
following ref . , the connection between shell structure and classical trajectories is most easily illustrated by considering the quantum energy levels of a particle in a spherical potential ( no spin - orbit interaction ) . in this case
the energy levels depend on the radial and angular momentum quantum numbers @xmath8 and @xmath9 only . for sufficiently large values of these quantum numbers
, the energy difference between two neighbouring levels can be approximated as [ edi ] ( n_r,l)-(n_r , l)(n_r-n_r ) + ( l-l ) .
now the important point is that the derivatives @xmath10 and @xmath11 are essentially the derivatives of the classical hamiltonian with respect to the corresponding action variable , hence they can be recognized as the radial @xmath12 and angular @xmath13 frequencies of the classical orbits respectively .
thus [ edi2 ] ( n_r,l)-(n_r , l)(n_r-n_r ) _ r+ ( l-l ) _ , ( we use units such that @xmath14 ) .
this combination of frequencies is exactly that appearing in the denominator of the zero - order response function obtained in @xcite from the solution of the linearized vlasov equation with fixed - surface boundary conditions .
moreover , from a comparison of the zero - order vlasov response function with the analogous quantum propagator @xcite , it can be seen that in the vlasov propagator the quantum matrix elements of the excitation operator are replaced by appropriate fourier coefficients that can be evaluated as integrals along the classical trajectories .
since here we are interested in the octupole response , we report the propagator describing the response of a nucleus to an external field of the type [ exf ] q(*r*)=r^3y_3m(*r * ) in the single - particle approximation .
this propagator is given by @xcite [ fsr ] r_l=3 ^ 0(s)= _ n=-^+ _ n=1,3c^2_3n _ 0 ^ 1 dx x^2 s_nn(x)(q^(3)_nn(x)/r^3)^2 .
instead of the frequency @xmath15 , as independent variable we have used the dimensionless quantity @xmath16 , as a consequence the @xmath7-dependence of this propagator is factorized as @xmath17 .
the eigenfrequencies ( [ edi2 ] ) are accordingly replaced by the functions ( for a square - well potential ) s_nn(x)=. the variable @xmath18 is related to the classical angular momentum @xmath19 of a nucleon .
the relation is @xmath20 , where @xmath21 is the angle spanned by the radial vector when the particle moves from the inner to the outer turning point .
clearly , for a square - well potential one has @xmath22 , where @xmath23 is the maximum particle angular momentum @xmath24 , @xmath25 is the fermi momentum , while @xmath26 and @xmath27 are the corresponding velocity and energy .
the sum over discrete angular momentum values of the quantum propagator is replaced here by the integration over @xmath18 , thus the particle angular momentum is treated as a continuous variable .
the quantities @xmath28 in eq .
( [ fsr ] ) are classical limits of the clebsh - gordan coefficients coming from the angular integration of the quantum matrix elements @xcite .
their explicit value is @xmath29 and @xmath30 . in principle
the integer @xmath31 takes values between @xmath32 and @xmath33 , however only the coefficients @xmath34 where @xmath31 has the same parity as @xmath33 are nonvanishing .
the coefficients @xmath35 appearing in the numerator of eq . ( [ fsr ] ) have been defined in ref .
@xcite , they are essentially the classical limit of the radial matrix elements of the octupole operator @xmath36 and can be evaluated explicitly as [ fou3 ] q^(3)_nn(x)=(-)^nr^31+n -+4(|n|-1 ) . for @xmath37
this expression coincides with that appearing in the ( compression ) dipole response ( cf .
eq.(a.5 ) of ref .
@xcite ) , however for the octupole response we also need terms with @xmath38 .
these new modes have an interesting property since the associated eigenfrequencies @xmath39 can vanish in the interval @xmath40 [ the equation @xmath41 has a solution for @xmath42 , corresponding to closed triangular orbits ] . in ref .
it has been pointed out that the vanishing of this eigenfrequency might give rise to a possible instability against octupole - type deformations in nuclei heavier than @xmath3pb .
although at first sight it might seem that the coefficients ( [ fou3 ] ) would diverge when @xmath43 , it is actually possible to check that [ lim ] _ xq^(3)_13(x)=-r^3 .
the very fact that this limit is finite is important for our discussion about the role of triangular nucleon orbits .
the linear response function in the single - particle approximation is well - behaved because [ imro ] _
s0(im r_3 ^ 0(s))=0 , and [ rero ] _
s0(re r_3 ^ 0(s ) ) = -a^3 , a result that we shall need later . since the zero - order octupole strength function is proportional to the imaginary part of the response function ( [ fsr ] ) , we can see that the contribution of triangular trajectories to the octupole strength function does not give any pathology , at least at the single - particle level .
of course instabilities could still arise in the collective response , so we study also this response by using the same model , however , before doing this we look more in detail to the single - particle octupole strength given by the vlasov theory
. the zero - order octupole strength function @xmath44 , given by ( @xmath45 ) s_l=3(e)=- , is shown in fig.1 .
it can be seen that the single - particle octupole strength is concentrated in two regions around 8 and 24 mev . as pointed out already in @xcite , in this respect
our semiclassical response is strikingly similar to the quantum response , which is concentrated in the @xmath0 and @xmath1 regions .
the modes that contribute most are those with @xmath46 for the @xmath0 region and with @xmath47 for the @xmath1 region .
this concentration of strength is quite remarkable because our static distribution , which is taken to be of the thomas - fermi type @xmath48 , does not include any shell effect , however , because of the close connection between shell structure and classical trajectories expressed by eqs .
( [ edi2 ] ) and ( [ fou3 ] ) , we still obtain a stregth distribution that is very similar to the one usually attributed to shell effects .
clearly the integers @xmath49 and @xmath31 correspond to the difference of radial and angular momentum quantum numbers in eq .
( [ edi2 ] ) .
the zero - order response function ( [ fsr ] ) gives only a first approximation to the nuclear response .
when the residual interaction between nucleons is taken into account , a collective response function can be obtained by solving the vlasov equation with appropriate boundary conditions @xcite . if the interaction is assumed to be of the octupole - octupole type , [ vres ] v(*r*_1,*r*_2)=_3 r_1 ^ 3 r_2 ^ 3 _ my^*_3m(_1)y_3m(_2 ) the collective fixed - surface octupole response function is given by @xcite [ collfix ]
r_3(s)=. the parameter @xmath50 specifies the strength of the residual interaction , its value can be estimated in a self - consistent way , giving ( , p.557 ) , _ bm = - -1 .
10 ^ -5 mev / fm^6 for the octupole case .
the parameter @xmath51 is given by @xmath52 . since this estimate
is based on a harmonic oscillator mean field and we are assuming a square - well potential instead , we expect some differences .
hence we shall determine the parameter @xmath50 phenomenologically , by requiring that the peak of the high - energy octupole resonance agrees with experiment .
this requirement implies @xmath53 , which is in agreement with the prescription obtained in the quadrupole case @xcite . in fig.2
we report the collective octupole strength function given by eq .
( [ collfix ] ) , with this value of @xmath50 .
we can clearly see the effects of collectivity that result in a shift and concentration of the strength into two sharp peaks around 20 mev and 6 - 7 mev .
the experimentally observed @xcite concentration of isoscalar octupole strength in the two regions usually denoted by heor ( high energy octupole resonance ) and leor ( low energy octupole resonance ) is qualitatively reproduced , however the considerable strength experimentally observed at lower energy ( low - lying collective states ) is absent from our fixed - surface response function .
like for the quadrupole response @xcite , we need to consider a different solution of the linearized vlasov equation ( in which the nuclear surface is allowed to vibrate @xcite ) in order to account for this feature of the response .
the poles of the collective response function are determined by the vanishing of the denominator in eq .
( [ collfix ] ) .
a solution of this equation at zero or purely imaginary frequency could be interpreted as an instability of the spherical shape against an octupole - type deformation of the ground state , however we can see that this could happen only for interactions stronger than _ 3cr=--3.14 10 ^ -5 mev / fm^6 .
our value of @xmath50 is smaller ( in absolute value ) than this , so in the present model the spherical shape is stable .
this is valid for any value of @xmath7 , as long as we take @xmath54 , since the product @xmath55 is @xmath7-independent .
the triangular orbits do not enter directly into the discussion of this collective response , but the fact that they do not generate pathologies at the level of the single - particle response @xmath56 simplifies the discussion of the stability in the collective response also .
the collective fixed - surface response function ( [ collfix ] ) does reproduce two important features of the octupole response : the heor and the leor , however it misses the experimentally observed low - lying collective @xmath2 states . in order to account for these low - lying states ,
a different solution of the linearized vlasov equation has been proposed in ref .
@xcite . in that approach the nuclear surface
is allowed to vibrate according to the usual liquid - drop model expression [ rot ] r(,,t)=r+_lmr_lm(t ) y_lm ( , ) .
a self - consistency condition involving the surface tension is then used to determine the time - dependence of the additional collective variables @xmath57(t ) .
then , always for a separable residual interaction of the kind ( [ vres ] ) , the collective fixed - surface response function ( [ collfix ] ) is replaced by the following moving - surface response function @xcite [ rtil ] _
3(s)= r_3(s)+ s_3(s ) , with @xmath58 still given by eq .
( [ collfix ] ) , while @xmath59 gives the moving - surface contribution . for the separable interaction ( [ vres ] )
the function @xmath59 can be evaluated explicitly as @xcite [ surfresp ] s_3(s)=- , with @xmath60 ( @xmath61 is the surface tension parameter obtained from the mass formula , @xmath62 gives the coulomb contribution to the restoring force ) and @xmath63 the equilibrium density .
the functions @xmath64 and @xmath65 are defined as [ chiok ] ^0_3(s ) = _ n n c_3n^2_0 ^ 1 dx x^2 s_nn(x)(-)^n ( q^(3)_nn(x)/r^3 ) , and [ chik ] _
3(s)=-_f(s+i ) _ n n c_3n^2 _ 0 ^ 1 dx x^2 , their structure is similar to that of the zero - order propagator ( [ fsr ] ) . in fig.3
we report the octupole strength function given by the moving - surface response function ( [ rtil ] ) and compare it to the collective fixed - surface response given by eq .
( [ collfix ] ) .
the most relevant change induced by the moving surface is the large double hump appearing at low energy .
this feature is in qualitative agreement both with experiment @xcite and with the result of rpa - type calculations @xcite , moreover is rather similar to that found in ref .
@xcite within the same model , but with different excitation operator and residual interaction
. we interpret this low - energy double hump as a superposition of surface vibrations and leor . in order to explain why we do not obtain one or more sharp @xmath2 states at low energy
, we have to analyze our moving - surface response function in some detail .
since the explicit expression ( [ surfresp ] ) looks rather involved , we consider the limit of non - interacting nucleons .
if we let @xmath66 , the function ( [ surfresp ] ) becomes [ zsurf ] s^0_3(s)=-r^6 , and the full moving - surface response function ( [ rtil ] ) [ rtilz ] ^0_3(s)= r^0_3(s)+ s^0_3(s ) . in fig.4
we show the octupole response function evaluated both for @xmath54 ( solid curve ) and for @xmath67 ( dashed curve ) .
we can see from this figure that , while the residual interaction changes drastically the response in the giant resonance region , at low energy the octupole response is affected only very slightly by the interaction ( [ vres ] ) .
thus the low - energy octupole response can be analyzed by using the simpler formula ( [ zsurf ] ) , rather than eq . ( [ surfresp ] ) .
the eigenfrequencies of low - energy collective modes are approximately determined by the vanishing of the denominator in eq .
( [ zsurf ] ) , moreover at low frequency the function @xmath68 can be expanded as @xcite [ lfe ] _
3()=i_3+d_3 ^ 2 + , implying that the parameters @xmath69 , that describe octupole surface vibrations in eq .
( [ rot ] ) , approximately satisfy an equation of motion of the damped oscillator kind : [ odos ] d_3_3m(t ) + _ 3 _ 3m(t)+c_3r_3m(t)=0 .
it has already been pointed out in @xcite that for @xmath4 the numerical values of parameters in this equation are such that this oscillator is actually overdamped , moreover , always in @xcite , the coefficients @xmath70 have been evaluated analyticallty in the low - frequency limit , giving ( for a generic @xmath33 ) [ gamma ] _
l = _ wf 2(4)^2_n=1^l 1n y_ln(2,2)^2 _ n=1^+ _
nn^3 _ nn ( 2-_nn ) , with @xmath71 and @xmath72 .
the angles @xmath73 are related to the nucleon trajectories as discussed in sect .
ii . in our case
the coefficient @xmath74 gets a contribution only from the term with @xmath75 and @xmath76 , thus we see that only nucleons moving along closed triangular trajectories can contribute to the damping of octupole surface vibrations . in order to check that indeed the closed triangular orbits are the main source of landau damping in the low - energy octupole response , we study also the response that is obtained when the contribution of these orbits is excluded .
this exclusion can be made simply by avoiding a small interval of length @xmath77 about the value @xmath78 in the @xmath18-integration in eqs .
( [ fsr ] ) , ( [ chiok ] ) and([chik ] ) . in fig.5
we show the response obtained with a value of @xmath79 .
the shape of the low - energy hunp changes dramatically because of the lack of damping due to the missing triangular trajectories .
a very sharp peak is now developed at low energy .
the position of this peak is influenced by the value of the mass parameter @xmath80 , omitting the triangular orbits changes also the numerical value of this parameter .
once again we see that the closed triangular orbits , rather than generating a shape instability in the octupole channel , are the main source of landau damping in this response function .
of course quantum effects can play an essential role at this level since in quantum mechanics the angular momentum is quantized and there could be no value of angular momentum corresponding to triangular trajectories .
we would also like to point out that our moving - surface response function can display a shape instability in the octupole channel , which can arise if the restoring force parameter @xmath81 vanishes .
this could happen if the repulsive coulomb term @xmath62 exactly balances the attractive surface - tension part of the restoring force . in this case
our moving - surface response function developes a pole at the origin , as shown in fig.6 . for odd multipolarities
the instability condition in our model is the same as for the liquid - drop model @xcite . before concluding , in fig.7 we compare our moving surface response function with the response function of the overdamped oscillator ( [ odos ] ) .
this figure supports our interpretation of the lower - energy hump as due to overdamped surface vibrations .
we have analyzed the octupole response function of a hypothetical heavy nucleus containing @xmath4 nucleons in a square - well potential by employing a semiclassical model that relates the response to features of the classical trajectories of nucleons .
triangular trajectories are particularly relevant for the octupole response because , at the single - particle level , a simple combination of their characteristic frequencies , appearing in this response function , can vanish .
this fact had already been noticed long ago and some speculations had been made on the possible connection between these classical orbits and the onset of a shape instability in heavy spherical nuclei against octupole - type deformations .
our detailed calculations performed in a semiclassical approach show that the vanishing of the eigenfrequencies associated with triangular trajectories has no consequences on the stability of the spherical shape in heavy nuclei , rather , triangular orbits are essential in providing a damping of low - energy octupole excitations . 99 a. bohr , b.r .
mottelson , nuclear structure , vol.ii , ( w.a.benjamin inc . ,
reading , mass . , 1975 )
brink , a. dellafiore , m. di toro , nucl .
a456 ( 1986 ) 205 v. abrosimov , m. di toro , v. strutinsky , nucl .
a562 ( 1993 ) 41 v. abrosimov , a.dellafiore , f. matera , nucl .
a653 ( 1999 ) 115 v.i .
abrosimov , a.dellafiore , f. matera , nucl .
a697 ( 2002 ) 748 v.i .
abrosimov , a.dellafiore , f. matera , nucl .
a717 ( 2003 ) 44 h. kohl , p. schuck , s. stringari , nucl .
a459 ( 1986 ) 265 a. dellafiore , f. matera , d.m .
brink , phys .
a 51 ( 1995 ) 914 g.f .
burgio , m. di toro , nucl.phys .
a476 ( 1988 ) 189 a. van der woude , progr .
18 ( 1987 ) 217 k.f .
liu , g.e .
brown , nucl .
phys . a265 ( 1976 ) 385 k.f .
liu , h. luo , z. ma , q. shen s.a .
moszkowski , nucl .
a534(1991)1 v.i .
abrosimov , o.i .
davidovskaya , v.m .
kolomietz , s. shlomo , phys .
c 57 ( 1998 ) 2342 fig.1 semiclassical octupole strength function analogous to quantum single - particle strength function . note the strength concentration in the @xmath0 and @xmath1 regions .
calculations are for @xmath4 nucleons in a square well potential of radius @xmath82 .
fig.3 the solid curve shows the octupole strength function evaluated by using moving - surface boundary conditions in the vlasov equation , the dashed curve instead , corresponds to fixed - surface boundary conditions .
fig.7 collective moving - surface strength function ( solid ) compared to overdamped oscillator strength function ( dashed ) with appropriate parameters .
the two functions practically coincide for @xmath85 .
|
the isoscalar octupole response of a heavy spherical nucleus is analyzed in a semiclassical model based on the linearized vlasov equation .
the octupole strength function is evaluated with different degrees of approximation .
the zero - order fixed - surface response displays a remarkable concentration of strength in the @xmath0 and @xmath1 regions , in excellent agreement with the quantum single - particle response .
the collective fixed - surface response reproduces both the high- and low - energy octupole rsonances , but not the low - lying @xmath2 collective states , while the moving - surface response function gives a good qualitative description of all the main features of the octupole response in heavy nuclei .
the role of triangular nucleon orbits , that have been related to a possible instability of the spherical shape with respect to octupole - type deformations , is discussed within this model .
it is found that , rather than creating instability , the triangular trajectories are the only classical orbits contributing to the damping of low - energy octupole excitations . #
10= -.025em0 - 0 .05em0 - 0 -.025em.0433em0 @xcite pacs : 24.10.cn , 24.30.cz keywords : vlasov equation , isoscalar octupole resonances , octupole surface modes
|
the top quark , with a mass of @xmath1 gev @xcite , is the most massive of the elementary particles of the standard model ( sm ) . in proton - antiproton collisions ,
top quarks can be produced singly through electroweak interactions and this process provides a unique opportunity to test the standard model and search for non - sm physics . in the sm , the single - top - quark production cross section is predicted to be proportional to the square of the magnitude of the quark - mixing cabibbo - kobayashi - maskawa matrix @xcite element @xmath2 .
consequently , measurements of the single - top - quark production rate could be sensitive to contributions from a fourth generation of quarks @xcite .
the two dominant production modes of the single - top - quark process are shown in fig .
[ fig : feynman_diagrams ] and are sensitive to different classes of sm extensions @xcite .
the @xmath0-channel process @xcite , in which an intermediate @xmath3 boson decays into a top ( antitop ) quark and an antibottom ( bottom ) quark , is sensitive to contributions from additional heavy bosons @xcite ; the @xmath4-channel process @xcite , in which a bottom quark transforms into a top quark by exchanging a @xmath3 boson with another quark , is more sensitive to flavor - changing neutral currents @xcite . independently studying the production rate of these channels provides more restrictive constraints on sm extensions than just studying the combined production rate .
-channel and ( b ) @xmath4-channel single - top - quark production at the tevatron.,scaledwidth=48.0% ] single - top - quark production was first observed independently by the cdf and d0 experiments in refs .
@xcite and @xcite , respectively .
these results were based on an inclusive search targeting the combined contribution of @xmath0-channel and @xmath4-channel production .
several differences in @xmath0- and @xmath4-channel event properties can be used to distinguish them from one another .
for example , events originating from @xmath4-channel production typically contain one light - flavor jet in the forward detector region ( at large pseudorapidity ) .
the direction of this jet ( whether it is along the proton or anti - proton direction ) is also correlated with the charge of the lepton .
these properties are very powerful for distinguishing them from events associated with @xmath0-channel production and other sm background processes .
moreover , events from the @xmath0-channel process are more likely to contain two jets which can be identified as originating from @xmath5 quarks ( @xmath5 jets ) . hence , single - top - like events with two identified @xmath5 jets are more likely to have originated from @xmath0-channel production .
the @xmath4-channel single - top quark production was first observed in 2011 by the d0 experiment @xcite , and later confirmed in 2012 by the atlas and cms experiments at the lhc @xcite . the @xmath0-channel process has not yet been observed . because of the smaller production cross section and larger backgrounds , it is more difficult to isolate it compared to the @xmath4-channel process in proton - antiproton collisions .
it is even more difficult at the lhc , as proton - proton collisions yield a significantly smaller signal - to - background ratio compared to the tevatron .
in fact , to date the lhc experiments have only reported unpublished upper limits on the cross section .
since the magnitude of the @xmath3-top - bottom quark coupling is much larger than the @xmath3-top - down and @xmath3-top - strange quark couplings , each top quark is assumed to decay exclusively into a @xmath3 boson and a @xmath5 quark .
we seek events in which the @xmath3 boson decays leptonically ( @xmath6 ) .
online event selection is based on identifying an isolated high-@xmath7 lepton or a large imbalance of transverse energy .
the offline event selection is split into two distinct topologies , both designed to select single - top - quark events in which the @xmath3 boson decays leptonically .
one final - state topology ( ) , analyzed by both the cdf and d0 collaborations , contains single - top - quark events in which the charged lepton from the @xmath3 boson decay is identified .
events are selected with an identified electron or muon , two or more jets , and .
at least one of the jets must be identified as a @xmath5 jet @xcite .
additional selection criteria are applied to exclude kinematic regions that are difficult to model , and to minimize the quantum chromodynamics ( qcd ) multijet background .
the other final - state topology , analyzed by the cdf collaboration , involves and jets , the event selection is similar to , but no reconstructed isolated charged leptons are allowed in the event(+jets ) .
this additional sample increases the acceptance for @xmath0-channel signal events by encompassing those in which the @xmath3-boson decay produces a muon or electron that is either not reconstructed or not isolated , or a hadronically decaying tau lepton that is reconstructed as a third jet . in order to reduce the large multijet background in this channel , a neural - network event selection is optimized to preferentially select signal - like events .
events passing the and + jets selections are further separated into independent analysis channels based on the number of reconstructed jets as well as the number and quality of @xmath5-tagged jets .
each of the analyzed channels has a different background composition and signal ( @xmath0 ) to background ( @xmath5 ) ratio . analyzing them
separately enhances the sensitivity to single - top - quark production .
both collaborations use monte carlo ( mc ) generators to simulate the kinematic properties of signal and background events , except in the case of multijet production , for which the model is derived from data .
single - top - quark signal events are modeled with next - to - leading - order ( nlo ) accuracy in the strong coupling constant @xmath8 @xcite .
kinematic properties of background events associated with the @xmath3+jets and @xmath9+jets processes are simulated using a leading - order mc generator @xcite , and those of diboson processes ( @xmath10 , @xmath11 and @xmath12 ) and are modeled using mc @xcite . in all cases is used to model proton remnants and simulate the hadronization of all generated partons .
all mc events are processed through -based detector simulations @xcite .
predictions for the normalization of simulated background - process contributions are estimated using both simulation and data .
data are used to normalize the @xmath3 plus light - flavor and heavy - flavor jet contributions using enriched @xmath3+jets data samples that have negligible signal .
all other simulated background samples are normalized to the theoretical cross sections at nlo combined with next - to - next - to - leading log ( nnll ) resummation @xcite for @xmath4-channel single - top - quark production , at next - to - nlo @xcite for , at nlo @xcite for @xmath9+jets and diboson production , and including all relevant higher - order qcd and electroweak corrections for higgs - boson production @xcite . differences observed between simulated events and data in lepton and jet reconstruction efficiencies , resolutions , jet - energy scale ( jes ) , and @xmath5-tagging efficiencies are adjusted in the simulation to match the data , through correction functions obtained from measurements in independent data samples .
d0 has optimized the single top analysis from ref .
@xcite to enhance the sensitivity to the @xmath0-channel in the @xmath13 final state , and has improved the selection by employing a newer more efficient @xmath5-tagging algorithm . in the new analysis @xcite the events
are split into four independent channels depending on the jet multiplicity and the number of @xmath5-tagged jets .
three multivariate methods are applied to the selected data : boosted decision trees ( bdt ) , bayesian neural networks ( bnn ) , and matrix element probability calculations ( me ) .
they are optimized to measure the @xmath0 and the @xmath4 channels separately in all four analysis channels .
the bdt method uses up to 30 kinematic variables , the bnn uses the objects four momenta with a few other variables such as the @xmath5-tag output of jets , @xmath14(light jet ) , and the invariant @xmath3 boson mass , and the me uses the objects four vectors .
the three methods are not fully correlated ( @xmath15 ) and their discriminant outputs are therefore combined by a second bnn which gives a better expected sensitivity than any individual method .
one of the most important improvements in this analysis is the development of a new method to measure the @xmath0- and the @xmath4-channels independently , without assuming the sm @xmath16 ratio when calculating the @xmath17 cross section .
this flexible analysis is used for all of the d0 results discussed below .
the cdf results discussed below are based on four different analyses . in an analysis from a few years ago @xcite ,
a nn is used to make single top quark measurements in the @xmath18 sample .
events with two or three jets and one or two @xmath5 tags are considered . in the category with two tags and two @xmath5 jets ,
the nn is optimized for @xmath0 channel , while in the other three categories the nn is optimized for @xmath4 channel events .
more recently , an analysis has been optimized for @xmath0 channel in all categories @xcite of the lepton plus two jet sample .
the analysis is based on the sm higgs boson search @xcite , which shares the same final state : @xmath19 .
there are three lepton categories : central tight electrons , central tight muons , and extended muons ( loose muons and isolated track lepton - candidates ) .
the identification of @xmath5 jets has also improved with respect to previous searches @xcite .
the selection requires exactly two jets , at least one of which is required to be @xmath5-tagged . the analysis is then split in four channels according to the different @xmath5-tagged properties of the jets .
one operational point ( denoted t for tight ) has 42% @xmath5-jet efficiency and 0.9% mistag rate , and a second operational point ( denoted l for loose ) has 70% @xmath5-jet efficiency and 9% misidentification rate per jet .
events are therefore split according to the four orthogonal categories : tt , tl , t , and ll .
this improves the sensitivity of the analysis by separating regions of the phase space with different signal - to - background ratios .
cdf also has two analyses based on the + jets topology : one for the combined @xmath17 analysis @xcite , and one that is optimized for the @xmath0 channel @xcite .
+ @xmath4 channel).,scaledwidth=68.0% ] for each analysis multivariate discriminants are trained to separate the single top signal from the background processes .
a binned likelihood technique using the multivariate discriminant shapes is employed in order to extract the most likely value for the cross section measurement .
the most recent measurements of the combined @xmath0-channel plus @xmath4-channel cross section measurements from the tevatron are summarized in fig .
[ fig : s_plus_t_summary ] .
from the combined @xmath17 cross section , d0 calculates the value of @xmath20 without any assumption on the number of generations or the unitarity of the ckm matrix , and the result is : @xmath21 , pr @xmath22 at @xmath23 c.l . , assuming a flat prior within @xmath24 .
this is currently the most stringent lower limit on @xmath20 from the single top measurements at the tevatron . a past cdf and d0 combination measured @xmath25 @xcite .
a combination is also planned for the updated analyses .
in addition to the inclusive cross section and the @xmath2 extraction , both cdf and d0 perform a two - dimensional fit for the @xmath0 and @xmath4 channel cross sections . as an example , the two - dimensional posterior probability density as a function of the @xmath0-channel and the @xmath4-channel cross sections from the d0 analysis is shown in figure [ fig:2d_d0 ] .
the predictions from several exotic models are shown on the figure .
cdf and d0 are planning to produce a combined 2-d fit employing the full data set of both experiments .
-channel ( @xmath26 ) and the @xmath4-channel ( @xmath27 ) cross sections .
several models beyond the sm are shown for reference @xcite .
, scaledwidth=68.0% ]
in the summer of 2013 , d0 reported the first evidence for the the @xmath0-channel process with a significance greater than 3 sd : the expected and observed significances are 3.7 sd @xcite . later that summer ,
cdf also reported evidence for the @xmath0 channel with an observed significance of 3.8 sd @xcite .
recently , the collaborations have combined their measurements @xcite .
figure [ fig : s_summary ] summarizes the most recent @xmath0-channel cross section measurements from each experiment and the tevatron combination .
the combined value of the @xmath0-channel single - top - quark production cross section is measured to be @xmath28 pb , in agreement with the sm expectation of @xmath29 pb ( @xmath30 gev ) @xcite .
-channel single top cross section measurements and the tevatron combination.,scaledwidth=68.0% ] the statistical significance of the combined result was quantified by an asymptotic log - likelihood ratio approach ( llr ) @xcite , including systematic uncertainties ( see fig . [
fig : pvalue ] ) .
the probability to measure an @xmath0-channel cross section of at least the observed value in the absence of signal is @xmath31 , corresponding to a significance of 6.3 standard deviations , well beyond the standard to claim discovery .
-channel single top quark production.,scaledwidth=68.0% ]
single - top - quark production was first observed at the tevatron in 2009 independently by the cdf and d0 experiments . soon after , the d0 collaboration reported a significance of greater than 5 standard deviations for the @xmath4-channel process .
both experiments have produced constraints on @xmath2 .
these constraints are still competitive with those from the lhc , and an effort to combine the cdf and d0 constrains is underway .
the lhc experiments have confirmed the @xmath4-channel process , but measurements of the @xmath0-channel process are much more difficult in the lhc environment than at the tevatron .
these proceedings document the first conference announcement ( la thuile 2014 ) that the cdf and d0 collaborations combined their datasets and observed the @xmath0-channel process with a significance of over 6 standard deviations .
e. e. boos , v. e. bunichev , l. v. dudko , v. i. savrin , and v. v. sherstnev , phys .
nucl . * 69 * , 1317 ( 2006 ) .
we use singletop version 4.2p1 .
m. l. mangano , m. moretti , f. piccinini , r. pittau , and a.d .
polosa , j. high energy phys .
07 ( 2003 ) 001 .
we use version 2.11 .
|
first observation of single - top - quark production in the @xmath0 channel is reported .
the result is based on the combination of the cdf and d0 measurements of the cross section in proton - antiproton collisions at a center - of - mass energy of 1.96 tev .
a summary of other recent single - top - quark results are also included .
|
a partition @xmath6 is a nonincreasing sequence of non - negative integers : @xmath46 .
partitions are often identified with the young diagrams .
take a box @xmath47 and let @xmath48 be the hook .
@xmath49 the arm length , the leg length and the hook length are @xmath50 where @xmath51 is the dual partition .
the hook @xmath48 is called @xmath52-hook if @xmath53 . fix a positive integer @xmath8 .
a partition @xmath6 is called @xmath8-core if it does not contain any @xmath8-hook .
otherwise , another partition @xmath54 is obtained from @xmath6 by removing one of the @xmath8-hooks .
the procedure can be continued until we obtain a @xmath8-core .
this partition turns out to be independent from the way @xmath8-hooks are removed and is denoted by @xmath7 .
the set of removed @xmath8-hooks constitutes @xmath8-tuple of partitions @xmath55 .
these partitions are obtained uniquely from @xmath6 and are called @xmath8-quotients . in the case of @xmath56 , we can read the @xmath8-quotients @xmath55 from the following relation : @xmath57 we thank to a. hanany for useful discussions and encouraging us .
y. n. is supported in part by jsps research fellowships for young scientists .
t. n. is supported in part by grant - in - aid for scientific research 15540273 .
f. fucito , j. f. morales and r. poghossian , `` multi instanton calculus on ale spaces , '' nucl .
b * 703 * , 518 ( 2004 ) [ arxiv : hep - th/0406243 ] .
s. fujii and s. minabe , `` a combinatorial study on quiver varieties , '' arxiv : math.ag/0510455 .
l. griguolo , d. seminara , r. j. szabo and a. tanzini , `` black holes , instanton counting on toric singularities and q - deformed two - dimensional yang - mills theory , '' arxiv : hep - th/0610155 .
f. fucito , j. f. morales and r. poghossian , `` instanton on toric singularities and black hole countings , '' arxiv : hep - th/0610154 .
|
we present an expression of a deformed partition function for @xmath0 @xmath1 gauge theory on @xmath2 by using plethystic exponentials . `
hep - th/0611324 ` + ` ou - het 570 ` + november 2006 * plethystics and instantons on ale spaces * yui noma , toshio nakatsu and takeshi tamakoshi + _ department of physics , graduate school of science , osaka university , + toyonaka , osaka 560 - 0043 , japan + _ [ [ section ] ] recently @xmath0 instantons on ale spaces were studied in @xcite and applied to black hole physics in @xcite . in this short note
we point out an expression of a deformed partition function for @xmath0 @xmath1 gauge theory on @xmath2 by using plethystic exponentials .
plethystic exponentials have been used in @xcite and play a central role in counting the bps operators .
let us start with describing the partition function for @xmath0 @xmath1 gauge theory on @xmath2 .
the partition function was computed in @xcite and becomes as follows : @xmath3 where @xmath4 are parameters of the standard torus action on @xmath2 , and @xmath5 denotes a scale parameter for the gauge theory .
the sum in eq.([eq;partition for 4d ] ) is a summation over partitions ( the young diagrams ) .
we remark that any partition @xmath6 can be decomposed into a single partition @xmath7 and @xmath8-tuple partitions @xmath9 @xcite .
they are called respectively @xmath8-core and @xmath8-quotients .
this arises from a division algorithm for the young diagrams analogous to that for integers .
see appendix [ sec;appendix ] for details .
we may write @xmath10 to emphasize the decomposition .
the sum in eq.([eq;partition for 4d ] ) is restricted so that only partitions whose @xmath8-cores are empty contribute to the partition function .
@xmath11 denotes the empty partition .
the symbol @xmath12 is the number of boxes in the young diagram : @xmath13 . for a box @xmath14 in the young diagram , @xmath15 , @xmath16 and @xmath17
are respectively , as given in ( [ eq;arm and leg ] ) , the hook length , the arm length and the leg length . in eq.([eq;partition for 4d ] ) , the products are taken over boxes whose hook length are multiples of @xmath8 . we will deform the partition function ( [ eq;partition for 4d ] ) by using a parameter @xmath18 in the following manner : @xmath19 where @xmath20 the parameter @xmath18 is identified in the gauge theory with a circumference of a circle in the fifth direction . for the case of @xmath21 , the deformed partition function is calculated in @xcite from the view point of refined topological vertices .
it is clear that eq.([eq;q - deformed partition function ] ) reduces to eq.([eq;partition for 4d ] ) at the limit where @xmath18 goes to zero : @xmath22 plethystic exponentials appeared in the study of counting the bps operators @xcite . in particular , by using it , counting functions of holomorphic functions ( the bps operators ) over the symmetric products @xmath23 , where the operators are counted weighted by their @xmath1 @xmath18 charges ( charges of the torus action ) , are reproduced from the counting function for @xmath24 , that is , @xmath25 . for a given function @xmath26 , plethystic exponential @xmath27 is defined by @xmath28 see @xcite for details of plethystic exponentials .
we point out the following expression of the deformed partition function by means of plethystic exponentials : + the rhs of ( [ eq;main claim ] ) can be simply written as @xmath29 the function @xmath30 is a counting function of holomorphic functions on @xmath2 .
the ring of the holomorphic functions is generated by @xmath31 and @xmath32 subject to the relation @xmath33 .
the @xmath1 charges are read as @xmath34 for @xmath35 , @xmath36 for @xmath37 and @xmath38 for @xmath32 . the equality ( [ eq;main claim ] ) for the case of @xmath21 is due to the identities related with macdonald s function @xcite .
it is also understood as one parameter deformation of that made by fujii and minabe @xcite . by letting @xmath39 , ( [ eq;main claim ] )
reduces to their result : @xmath40 we are convinced of the equality ( [ eq;main claim ] ) by comparing the first few coefficients in the taylor expansions with respect to @xmath41 of the both hand sides . as an example , let us see the case of @xmath42 .
we will compare the coefficients of @xmath43 .
the young diagrams which contribute to the coefficient of @xmath43 in ( [ eq;q - deformed partition function ] ) are those of six boxes as shown in figure [ fig;partitions_w_circle ] . in each young diagram , only the boxes filled with black circles contribute to the products in ( [ eq;q - deformed partition function ] ) .
we then see @xmath44 in eq.([eq;q - deformed partition function ] ) for the case of @xmath42 .
the boxes filled with black circles contribute to the products in eq.([eq;q - deformed partition function ] ) .
@xmath45-tuple partitions attached below each young diagram are the corresponding @xmath45-quotients . _ ]
|
numerical relativistic hydrodynamics ( rhd ) has experienced an important step forward in recent years when modern high resolution shock capturing ( hrsc ) techniques began to be applied to solve the equations of rhd in conservation form .
prior to the advent of hrsc techniques the field was dominated for more than one decade by wilson ( 1979 ) s approach to relativistic hydrodynamics .
this approach relies on the use of artificial viscosity in order to handle the discontinuities ( shocks , contact discontinuities , etc . ) that may appear in the flow numerically .
however , techniques based on artificial viscosity are prone to severe numerical difficulties when simulating ultrarelativistic flows ( see , e.g. , centrella & wilson 1984 ) . using modern hrsc techniques instead ,
these difficulties are overcome ( see , e.g. , donat et al .
1998 ) allowing one to simulate challenging relativistic astrophysical phenomena like , e.g. , relativistic jets or gamma - ray bursts ( grb hereafter ) . in astrophysical jets
flow velocities as large as @xmath2 of the speed of light ( lorentz factors @xmath3 ) are required according to the nowadays accepted standard model to explain the apparent superluminal motion observed at parsec scales in many jets of extragalactic radio sources associated to active galactic nuclei .
similar arguments applied to the galactic superluminal sources grs1915 + 105 ( mirabel & rodriguez 1994 ) and groj165540 ( tingay et al . 1995 ) allow one to infer intrinsic velocities of @xmath4 in the jets of these sources
. further independent indication of highly relativistic speeds can be inferred from the intraday variability occurring in more than a quarter of all compact extragalactic radio sources ( krichbaum , quirrenbach & witzel 1992 ) . if the observed intraday radio variability is intrinsec and results from incoherent synchrotron radiation ( according to begelman , rees & sikora 1994 ) , the associated jets must have bulk lorentz factors in the range @xmath5 30 100 . with exception of the remarkable work of van putten ( 1993 , 1996 ) , who used pseudo
spectral techniques to solve the equations of relativistic magnetohydrodynamics , numerical simulations of relativistic jets started soon after the first multidimensional relativistic hrsc codes had been developed ( mart , mller & ibez 1994 , duncan & hughes 1994 , mart et al . 1995 ) . since then many different aspects of relativistic jets have been investigated ( see , e.g. , mart 1997 for a recent review ) .
the morphology , dynamics and propagation properties of relativistic jets have been analysed in mart et al .
komissarov & falle ( 1997 ) investigated the long term evolution of relativistic jets .
first simulations of superluminal sources combining relativistic hydrodynamics and synchrotron radiation transfer at parsec scales have been performed by gmez et al .
( 1995 , 1997 ) and by mioduszewski , hughes & duncan ( 1997 ) and komissarov & falle ( 1997 ) . from their simulations these authors inferred that the observations of such sources can be explained in terms of travelling perturbations in steady relativistic jets . in the last two years
further progress was achieved by simulating relativistic jets in three spatial dimensions and by incorporating magnetic fields ( koide , nishikawa & mutel 1996 ; nishikawa et al .
1997 ; koide 1997 ) .
however , instead of using fluxes obtained by solving riemann problems at zone interfaces , koide and collaborators code rely on the addition of nonlinear dissipation terms to their lax
wendroff scheme to stabilise the code across discontinuities .
this stabilisation method was originally proposed by davis ( 1984 ) , who applied it successfully to the equations of classical hydrodynamics .
the method is robust and simple as no detailed characteristic information is needed .
koide and collaborators did simulate the evolution of the jet only for a very brief period of time .
this fact and the coarse grid zoning used in their simulations , however , prevented them from studying genuine 3d effects in relativistic jets in any detail . on the other hand , the relative smallness of the beam flow lorentz factor ( 4.56 , beam speed @xmath6 ) assumed in their simulations does not allow for a comparison with riemann - solver - based hrsc methods in the ultrarelativistic limit .
an astrophysical phenomenon which also involves flows with velocities very close to the speed of light are gamma - ray bursts ( grb ) .
although known observationally since over 30 years , their nature and their distance ( `` local '' or `` cosmological '' ) is still a matter of controversial debate ( fishman & meegan 1995 ; mzros 1995 ; piran 1997 ) . in order to explain the energies released in a grb various catastrophic collapse events
have been proposed including neutron - star / neutron - star mergers ( pacyski 1986 ; goodman 1986 ; eichler 1989 ) , neutron - star / black - hole mergers ( mochkovitch 1993 ) collapsars ( woosley 1993 ) and hypernovae ( pacyski 1998 ) .
these models all rely on a common engine , namely a stellar mass black hole which accretes several solar masses of matter from a disk ( formed during a merger or by a non - spherical collapse ) at a rate of @xmath7 ( popham , woosley & fryer 1998 ) .
a fraction of the gravitational binding energy released by accretion is converted into neutrino and anti - neutrino pairs , which in turn annihilate into electron - positron pairs .
this creates a pair fireball , which will also include baryons present in the environment surrounding the black hole .
provided the baryon load of the fireball is not too large , the baryons are accelerated together with the e@xmath8e@xmath9 pairs to ultrarelativistic speeds with lorentz factors @xmath10 ( cavallo & rees 1978 ; piran , shemi & narayan 1993 ) .
the bulk kinetic energy of the fireball then is thought to be converted into gamma - rays via cyclotron radiation and/or inverse compton processes ( see , e.g. , mszros 1995 ) . in the following we describe the main features of a special relativistic 3d hydrodynamic code , which is based on explicit hrsc methods , and which is a considerably extended version of the special relativistic 2d hydrodynamic code developed by mart , mller & ibez ( 1994 ) and by mart et al .
the code has been designed modularly which allows one to use different reconstruction algorithms and riemann solvers .
as it is the final goal of our work to simulate relativistic jets and grbs in three spatial dimensions , the code has successfully been subjected to an intensive testing in the ultrarelativistic regime ( see section [ ctest ] ) . in particular , genesis has successfully passed the spherical shock reflection test ( simulated in 3d cartesian coordinates ) involving flow lorentz factors larger than 700 ( see [ essr ] ) .
the paper is organised as follows . in section [ eqs ]
, we introduce the 3d equations of rhd in cartesian coordinates in differential and discretized forms .
the latter have been implemented into our 3d rhd code genesis .
detailed information about the structure and the main features of the code is given in section [ struc ] .
several 1d , 2d and 3d relativistic test problems computed with genesis are described in section [ ctest ] .
the performance of genesis on scalar and multi - processor computers is analysed in section [ cperf ] , and a realistic simulation of a 3d relativistic astrophysical jet is presented in section [ jet ] .
a summary of the paper containing our main conclusions and a discussion of present and future applications of the code in different astrophysical areas can be found in section [ conc ] . in appendix
a , we give the spectral decomposition of the three dimensional system of rhd equations with explicit expressions for the eigenvalues and the right- and left - eigenvectors .
appendix b contains the explicit formulae for the numerical viscosity for marquina s ( donat & marquina 1996 ) riemann solver , and appendix c describes the explicit algorithm to recover the primitive variables form the conserved ones .
the evolution of a relativistic perfect fluid is described by five conserved quantities : rest mass density , @xmath11 , momentum density , @xmath12 , and energy density , @xmath13 ( all of them measured in the laboratory frame and in natural units , i.e. , the speed of light @xmath14 ) , @xmath15 where the lorentz factor @xmath16 and @xmath17 ( the einstein summation convention is used here , and @xmath18 is the kronecker symbol ) .
furthermore , @xmath19 is the rest mass density ,
@xmath20 the pressure and @xmath21 the specific enthalpy given by @xmath22 with @xmath23 being the specific internal energy .
the components of the vector of variables @xmath24 are called _ primitive _ or physical variables .
the relativistic euler equations form a system of conservation laws ( see , e.g. , font et al .
1994 ) which can be written in cartesian coordinates as @xmath25 or , equivalently , as @xmath26 where the vector of unknowns @xmath27 ( i.e. , the conserved variables ) is given by @xmath28 and the fluxes are defined by @xmath29 the system [ f ] of partial differential equations is closed with an equation of state @xmath30 .
anile ( 1989 ) has shown that system [ f ] is hyperbolic for causal equations of state , i.e. , for those where the local sound speed , @xmath31 , defined by @xmath32 satisfies @xmath33 . the structure of the characteristic fields corresponding to the nonlinear system of conservation laws [ f ] has explicitly been derived in donat et al .
( 1998 ) and is summarised in appendix a. in order to evolve system [ f ] numerically , one has to discretize the state vector @xmath27 within computational cells .
the temporal evolution of the state vector is determined by the flux balance across the zone interfaces of each cell and the contribution of source terms . using a method of lines ( see , e.g. , leveque 1991 )
, our discretization reads @xmath34 where latin subscripts @xmath35 , @xmath36 and @xmath37 refer to the @xmath38 , @xmath39 and @xmath40 coordinate direction , respectively .
@xmath41 and @xmath42 are the mean values of the state and source vector ( if non zero ) in the corresponding three - dimensional cell , while @xmath43 , @xmath44 and @xmath45 are the numerical fluxes at the respective cell interface . finally , @xmath46 is a short hand notation of the spatial operator in our method . at this stage ,
our system of conservation laws is a system of ordinary differential equations which can be integrated with a large number of algorithms .
we have chosen a multistep runge
kutta ( rk ) method developed by shu & osher ( 1988 ) which can provide second ( rk2 ) and third ( rk3 ) order in time .
the explicit form of the algorithms is ( subindexes @xmath47 are ommited to clarify the notation ) :
1 . prediction step ( common for both rk2 and rk3 ) : @xmath48 2 .
depending on the order do : * rk2 : @xmath49 being @xmath50 and @xmath51 .
* rk3 : @xmath52 in this case , @xmath53 and @xmath54 . * *
the special relativistic multidimensional hydrodynamic code genesis described in detail in the following is a 3d extension of the 2d hrsc hydrodynamic code developed by some of the authors .
the 2d code has been successfully used for the simulation of relativistic jets ( mart et al .
1994 , 1995 , 1997 ; gmez et al .
1995 , 1997 ) .
the main structural features of mart et al.s code has been kept , but there are important changes in the computational part . besides the addition of the third spatial dimension , a large effort has been made to _
minimise memory requirements _ and to _ optimise the performance _ of the code as well as to enhance its _
portability_. like its predecessor , genesis evolves the equations of rhd in conservation form using a finite volume approach in cartesian coordinates . in accordance with the method of lines , we split the discretization process in two parts .
first , we only discretize the differential equations in space , i.e. the problem remains continuous in time .
this leads to a system of ordinary differential equations ( odes ) in time [ rkec ] .
the numerical fluxes between adjacent cells required for the time integration are obtained by solving the appropriate 1d riemann problems along the coordinate directions ( spatial sweeps ) .
high - order spatial accuracy is achieved by applying a high - order interpolation procedure in space , while high - order accuracy in time is obtained by using high - order ode solvers .
genesis integrates the 3d rhd equations on uniform grids in each spatial direction . in order to have a flexible code genesis
is programmed to allow for different boundary conditions , spatial reconstruction algorithms , riemann solvers , ode solvers for the time integration and external forces .
the user selects these options at the preprocessor level , which reduces the number of _
if_clauses inside the nested 3d loops to a minimum , and thereby maximising the code s efficiency . making the selection at the preprocessing stage has allowed us to obtain a code , which is independent of a specific ( shared memory ) machine architecture .
hence , it runs on different types of machines and processors . up to now , we have tested genesis on sgi platforms ( indy workstations , power challenge and cray - origin 2000 arrays ) , on hp machines ( 712 workstations and j280 computers ) , and on a cray - jedi multiprocessor system . as a next step
we plan to port genesis on a cray - t3e massively parallel computer .
the flow diagram of genesis is shown in figure [ flowdiag ] .
details of the major components of genesis are discussed in the following subsections .
the current version of genesis , which is written in fortran 90 has the capability of allocating memory _
dynamically _ ,
i.e. the number of computational cells can be chosen at run time .
reducing the ram requirements of a 3d hydrodynamic code is obviously crucial . in genesis
multidimensional variables are responsible for about @xmath55 of the code s memory requirement .
thus , the number of these 3d arrays has to be kept at the absolute minimum possible . in its present version ,
genesis only requires three sets of five 3d arrays each , consisting of one set of conserved variables at the begining of each time level ( @xmath56 ) , another set of primitive variables and a third set of scratch variables ( @xmath57 ) .
the time integration scheme ( eqs .
[ rka][rkb ] ) which results in the updated values of the conserved variables at the next time level ( @xmath58 ) then reads : 1 .
prediction step ( common for rk2 and rk3 ) : @xmath59 2 .
depending on the order of accuracy of the time integration scheme do : * @xmath60 + with @xmath50 and @xmath51 , or * @xmath61 + with @xmath53 and @xmath54 .
quantities like entropy , internal energy , sound speed or lorentz factor are implemented as fortran scalars .
consequently , genesis needs about 1 gbyte of ram memory to handle a grid of @xmath62 ( in double precision ) .
the technique of domain decomposition is used to optimise the parallelization of the code and to guarantee its performance in real applications , too .
it is also the first step towards the development of a parallel version of genesis which runs efficiently on parallel computers with distributed memory .
the physical domain is split along _
one _ arbitrary spatial direction ( z , in the present version ) in a set of subdomains ( i.e. _ slices _ , see fig .
[ domdec]a ) of similar computational load .
the subdomains are then distributed across processors .
numerical fluxes at subdomain boundaries are calculated by providing the appropriate internal and external boundary conditions ( see fig .
[ domdec]b , c , respectively , and [ bound ] ) .
the computational grid is extended in each coordinate in positive and negative direction by four so - called _ ghost _ zones , which provide a convenient way to implement different types of boundary conditions .
these boundary conditions have to be provided in each spatial sweep for all primitive variables . in genesis
several types of boundary conditions are available including reflecting , inflow , outflow , time - dependent and analytically prescribed boundary conditions .
flow conditions at subdomain boundaries must be provided , too , in order to calculate numerical fluxes at subdomain interfaces .
hence , subdomains are also enlarged by four ghost zones in each coordinate direction .
note that these ghost zones do overlap with adjacent subdomains ( see fig .
[ domdec ] ) .
the internal boundary conditions in these overlapping regions are defined by copying the corresponding values of the respective adjacent subdomain . for @xmath63 subdomains and @xmath64 computational zones
the number of overlapping cells is @xmath65 , i.e. the fraction of overlapping cells is @xmath66 .
hence , for @xmath67 and @xmath68 ( typical of a jet simulation ) the fraction of overlapping cells is about 12% . in order to improve the spatial accuracy of the code , we interpolate the values of the pressure , the proper rest mass density and the spatial components of the four velocity ( @xmath69 ) within computational cells . these reconstructed variables are afterwards used to compute the numerical fluxes . because of the monotonicity of the reconstruction procedures ( see below ) used in genesis , the occurrence of unphysical ( i.e. , negative ) values in the reconstructed profiles of pressure and density are avoided .
in addition , reconstructing the spatial components of the four - velocity with monotonic schemes , also prevents the occurrence of unphysical values of the flow velocity , i.e. , the flow velocity always remains smaller than the speed of light even in multidimensional calculations .
genesis provides , at the preprocessing level , four different types of reconstruction schemes : piecewise constant , linear using the minmod function of van leer ( 1979 ) , parabolic using the piecewise parabolic method , ppm , of colella & woodward ( 1984 ; see also mart & mller 1996 ) or hyperbolic using the piecewise hyperbolic method , phm , of marquina ( 1994 ) .
gravity , local radiative processes , etc . , are coupled with hydrodynamics through terms on the right hand side of the rhd equations ( i.e. , via the source terms , @xmath70 , in eq .
[ rkec ] ) .
genesis integrates such terms assuming piecewise constant profiles for the source functions . in this paper
we use a variant of marquina s flux formula ( see donat & marquina 1996 ) which has already been shown to work properly in the simulation of relativistic jets in 2d ( mart et al .
1997 ) .
the approach followed by donat & marquina ( 1996 ) relies on the extension of the entropy satisfying scalar numerical flux of shu & osher ( 1989 ) to hyperbolic systems of conservation laws .
given the spectral decomposition of the rhd equations ( see appendix a ) , the implementation of marquina s scheme is straightforward .
the original marquina s algorithm computes the contribution to the numerical viscosity of each characteristic field in a different way depending on whether the corresponding eigenvalue ( characteristic speed ) does change its sign between the left and right states or whether it does not .
however , instead of using the original algorithm , we only consider that part which corresponds to characteristic speeds changing their signs between the left and right states of every numerical interface .
the modified algorithm has a larger numerical viscosity , but it is more stable and does not involve any _ if_-clause .
hence , it can easily be vectorized . in the 2d version used in mart et al .
( 1997 ) the left eigenvectors of the jacobians are calculated numerically by inverting the matrix of right eigenvectors . in genesis
we use the analytical expressions for the left eigenvectors , which allow one to simplify the computation of the numerical viscosity terms .
the explicit expressions for the numerical fluxes ( @xmath71 , @xmath72 in eq .
[ rkec ] ) as a function of the local ( reconstructed ) primitive and conserved variables are given in appendix b. besides its influence on the efficiency of the code , the use of explicit expressions for the left eigenvectors also leads to analytical cancellations in the computation of the numerical viscosity causing a damping of the growth of round - off errors and an improvement of the overall accuracy of the code .
previous versions of genesis , in which numerical fluxes were calculated without the use of analytical expressions , suffered from a growth of round - off errors due to the large number of operations involved and due to the finite precision of floating point arithmetics .
this growth of errors manifests itself in a gradual loss of symmetry in initially perfectly symmetric problems .
our experience shows that the analytical manipulation of the expressions of the numerical flux together with their appropriate symmetrization ( i.e. , using commutating formulas for the components of the velocity parallel to cell interfaces ) allows one to achieve a perfect numerical symmetry ( see [ symmetry ] and [ rssr ] ) .
time integration is carried out by two different total variation diminishing rk methods developed in shu & osher ( 1988 ) .
the user can choose , at preprocessing level , between the rk2 and rk3 algorithm ( see eqs .
[ rka][rkb ] ) .
results of similar quality can be obtained either with the rk3 algorithm or with rk2 using smaller time steps .
nevertheless , for a given time step , the computational cost of rk3 is about a factor 1.5 larger than that of rk2 . as in any explicit hydrodynamic code ,
time steps are limited for stability reasons by the courant - friedrichs - levy ( cfl ) condition , which is computed using the characteristic speeds . at the end of each time
step the size of the new time step is determined as the minimum of the time steps of all subdomains .
this requires a global operation across all subdomains .
experience has shown that acceptable cfl numbers lie in the interval @xmath73 $ ] .
cfl numbers larger than @xmath74 can lead to post shock oscillations .
the solution of the riemann problem requires knowledge of the value of the pressure and its thermodynamic derivatives .
given the functional dependence between conserved and primitive variables ( see eq .
[ s ] ) , the recovering procedure can not be formulated in closed form .
instead a kind of iterative method must be used , which is very time consuming .
hence , usage of the recovering procedure should be reduced to the absolute minimum .
therefore , primitive variables are consistently updated from the mean values of the conserved variables after each runge - kutta step and their values are stored in a set of 3d arrays .
our approach is the same as that of mart , ibez & miralles ( 1991 ) and that of mart et al .
its explicit form can be found in appendix c ( see also mart & mller 1996 ) .
the iterative recovering procedure is based on a second order accurate newton - raphson method to solve an implicit equation for the pressure . in zones where the flow conditions change
smoothly the typical number of iterations ranges from 1 to 3 when a relative accuracy of @xmath75 is requested .
there exist zones , however , inside shocks or near strong gradients , where the number of iterations required is larger depending on the strength of the shock or the steepness of the gradient .
for example , in the shock reflection test in 3d , the shock zone needs about 4 to 8 iterations .
we have taken special care in designing a numerical code that accurately preserves any symmetries present in the initial data .
this is an important point for a code aimed to study , for example , the stability and long term evolution of initially axisymmetric jets .
there exist two potential sources of numerical _ asymmetries _ in our code , both of them are related to the fact that floating point arithmetics is not associative .
one cause of asymmetries is due to the computation of numerical fluxes in spatial sweeps , which violates what we call henceforth _ sweep level symmetry _ ( sls ) . in order to guarantee sls
the expressions by which the numerical fluxes are evaluated have been symmetrized ( see [ ars ] ) .
a second source of ( numerically caused ) asymmetry arises specifically in 3d codes using directional splitting .
it can only be avoided , if the code has a property which we call _ sweep coherence symmetry _
it refers to the symmetry of the integration algorithm with respect to the order in which the 1d - sweeps are performed .
this symmetry property of the algorithm becomes crucial if an initially spherically symmetric state is considered .
we found that its initial symmetry is lost unless special care is taken in the calculation of the lorentz factor ( in the numerical flux routine ) , which involves the summation of the squares of the three velocity components . to guarantee a perfect sweep coherence symmetry of the algorithm the addition of the vector components has to be performed in a cyclic manner , i.e. in the x sweep the components are summed up in @xmath76 order , in the y sweep in @xmath77 order , and finally in the z sweep in @xmath78 order .
due to the stochastic nature of round - off errors , a violation of the sweep coherence symmetry manifests itself only in the last few significant digits of the state variables , if the number of time steps is not too large ( less than about 3000 ; see section [ essr ] ) .
given that round off errors grow sufficiently slow and that they do not interact with the truncation errors due to the finite difference scheme ( which can render the scheme _ unstable _ ) , genesis does keep the symmetry of an initial state at an acceptable level .
we have also tried to develop a version of genesis with a perfect 3d symmetry ( limited by the cartesian discretization ) . for this purpose
, we applied the _ extended partial precision _ technique in the computation of expressions in which the associative property should be satisfied .
the procedure was successful , but increased the total computational costs by more than 30@xmath79 .
all the results presented in the following have been obtained without making use of such a technique .
the capabilities of genesis to solve problems in special relativistic hydrodynamics are checked by means of three tests calculations that involve strong shocks and a wide range of flow lorentz factors . in these test
runs an ideal gas equation of state with an adiabatic exponent @xmath80 has been used .
all results presented in this section have been obtained with the ppm reconstruction procedure and the relativistic riemann solver based on marquina s flux formula ( see previous section for details ) . in the first test
we consider the time evolution of an initial discontinuous state of a fluid at rest .
the initial state is given by @xmath81 , @xmath82 , @xmath83 , @xmath84 , @xmath85 , @xmath86 , @xmath87 and @xmath88 , where the subscript @xmath89 ( @xmath90 ) denotes the state to the left ( right ) of the initial discontinuity .
this test problem has been considered by several authors in the past ( in 1d by hawley , smarr & wilson 1984 , schneider et al .
1993 , mart & mller 1996 , wen , panaitescu & laguna 1997 ; in 2d by mart et al . 1997 ) .
it involves the formation of an intermediate state bounded by a shock wave propagating to the right and a transonic rarefaction propagating to the left .
the fluid in the intermediate state moves at a mildly relativistic speed ( @xmath91 ) to the right .
flow particles accumulate in a dense shell behind the shock wave compressing the fluid by a factor of 5 and heating it up to values of the internal energy much larger than the rest - mass energy .
hence , the fluid is extremely relativistic from a thermodynamical point of view , but only mildly relativistic dynamically . to change this
intrinsically one dimensional test problem into a multidimensional one we have rotated the initial discontinuity ( normal to the x - axis ) by an angle of @xmath92 around the y - axis , and then again by an angle of @xmath92 around the z - axis .
gas states @xmath89 and @xmath90 are placed within a cube of major diagonal equal to 1 that constitutes the 3d numerical grid . the analytical solution to this test problem
can be found in mart & mller ( 1994 ) .
our analysis is restricted to the flow conditions along the major diagonal of the numerical grid , which is normal to the initial discontinuity .
figure [ fig_gsch ] shows the solution along the major diagonal at time @xmath93 .
the shock is captured in two to three zones in accordance with the capabilities of hrsc methods .
the transonic rarefaction has a smooth profile across the sonic point located at @xmath94 , and exhibits sharp corners .
the contact discontinuity is spread out over roughly three zones .
the absolute global errors ( in @xmath95 norm given by @xmath96 , where @xmath97 and @xmath98 are the numerical and exact solution , respectively ) of pressure , density and velocity are given in table [ tab_err1.schn ] for different grid resolutions at @xmath93 . table [ tab_err1.schn ] implies a convergence rate of slightly less than 1 when comparing the errors obtained on the coarsest ( @xmath99 ) and the largest ( @xmath100 ) grid .
this behaviour is expected for multidimensional problems involving discontinuities ( see , e.g. leveque 1991 ) .
this 1d test problem involves the propagation of a strong shock wave generated when two cold gases , moving at relativistic speeds in opposite directions , collide .
the problem has been considered as a test for almost any new relativistic hydrodynamic code ( centrella & wilson 1984 ; hawley , smarr & wilson 1984 ; mart & mller 1994 ; eulderink & mellema 1994 ; falle & komissarov 1996 ) .
after the collision of the two gases , two shock waves are created in the plane of symmetry of the physical domain propagating in opposite directions .
the inflowing gas is heated in the shocks and comes to a rest .
the exact solution of this riemann problem was obtained by blandford & mckee ( 1976 ) .
the initial data are @xmath101 , @xmath102 , @xmath103 , @xmath104 , @xmath105 and @xmath106 , where @xmath107 is the inflow velocity of the colliding gas .
figure [ fig_rssr ] shows the numerical solution at @xmath108 on the left half of a grid having a total of 401 zones .
the results obtained in the right half of the grid are strictly symmetric with respect to the collision point ( @xmath109 ) , i.e. , the sweep level symmetry ( sls ; see section [ symmetry ] ) is exactly fulfilled .
near @xmath109 , the numerical solution shows small errors ( of the same order as the mean error in the post - shock state , @xmath110 ) which are due to the _ wall heating _ phenomenon ( noh 1987 ) characterised by an overshooting of the internal specific energy and an undershooting of the proper rest - mass density . in table
[ tab_err1.rssr ] we give the global absolute errors ( @xmath95 norm ) of the primitive variables for different grids at @xmath108 and for an inflow velocity @xmath111 .
we find a convergence rate about equal to one ( see columns 5 - 7 ) for all variables .
we can use this test problem to check the robustness of genesis in the ultrarelativistic regime . to simplify notation
, we define the quantity @xmath112 , which tends to zero when @xmath107 tends to one .
table [ tab_err4.rssr ] contains the relative global errors of the primitive variables at @xmath108 for a set of calculations performed on a grid of 401 zones , where we have varied @xmath113 from @xmath114 to @xmath115 .
the latter value corresponds to a lorentz factor @xmath116 .
the relative error of the primitive variables shows a weak dependence on the inflow velocity .
it never exceeds @xmath117 and for @xmath118 it is smaller than @xmath119 . the ppm parameters ( see colella & woodward 1984 ) have been tuned to minimize the number of zones within the shock without introducing unacceptable numerical post shock oscillations .
[ fig_rssr_m ] demonstrates that there are no numerical post - shock oscillations for @xmath120 when the shock is captured by 2 to 3 zones .
the initial setup consists of a spherical inflow at speed @xmath107 ( which might be ultrarelativistic ) colliding at the centre of symmetry of a sphere of radius unity . for a hydrodynamic code in cartesian
coordinates this is a 3d test problem , which allows one to evaluate the directional splitting technique as well as the symmetry properties of the algorithm .
figure [ fig_essr ] shows the numerical results for @xmath121 on a grid of @xmath122 zones at @xmath123 .
the shock capturing properties of genesis , which we have already demonstrated in 1d , are retained in this genuine multidimensional case .
two or three zones are required to handle the shock wave .
the pressure and proper rest
mass density have global relative errors of about @xmath124 and @xmath125 respectively .
ultrarelativistic flows have been explored by increasing the inflow lorentz factor .
table [ tab_err2.essr ] gives the growth of the relative global errors ( @xmath126 ) on a fixed grid size of @xmath127 zones for @xmath107 in the range @xmath128 to @xmath129 ( the latter inflow velocity corresponding to a lorentz factor @xmath130 ) .
the relative global errors are acceptable ( considering the inherent difficulty of the test and the resolution of the experiments ) and do not grow dramatically with the lorentz factor .
the observed growth can be explained by the fact that the errors are dominated by the shock region and that the shock strength increases with the lorentz factor .
the cfl factors used in the last two tests of this series are unusually small ( @xmath131 and @xmath132 ) which is due to the strength of the shock and the relatively small grid resolution ( compared with the 1d case ) .
it is noticeable that for @xmath133 the errors are considerably larger ( last entry in table [ tab_err2.essr ] ) .
this has two reasons .
firstly , the global relative errors decrease with time in the rssr test problem .
secondly , we could not continue the run with @xmath134 beyond 1.5 time units , because interaction with the grid boundaries became severe causing the code to crash .
hence , @xmath134 must be considered as the maximum inflow velocity in the rssr test problem , which the present code can handle properly ( for the resolution used ) .
the symmetry properties of the rssr solution are very well maintained by genesis , even though the number of timesteps was very large ( @xmath135 ) in the last two test runs .
the absolute global errors ( @xmath95 norm ) and the convergence rates of the primitive variables at @xmath108 are displayed in table [ tab_err1.essr ] .
obviously , the errors are much larger in the 3d test than in the corresponding 1d one .
this can be explained considering that ( i ) the grids are coarser than in 1d , and that ( ii ) the jumps in pressure and density across the shock are nearly a factor of 30 larger in the 3d test than in the planar case .
the preservation of the sweep - level symmetry ( sls ; see section
[ symmetry ] ) is reflected in the symmetry of the one dimensional profiles in fig .
[ fig_essr ] . moreover , a comparison of the profiles in x and y direction in fig .
[ fig_essr ] shows the capability of the code to maintain the sweep - coherence symmetry ( scs ) , too .
we have parallelized genesis in order to run on multiprocessor computers with shared memory .
appart from the initial setup of variables , the grid generation and the output , the rest of the program is organised in a 4-level nested loop .
the outermost loop runs from one to the total number of sub - domains , assigning one sub - domain to each processor .
this procedure allows an almost complete parallelization of the code employing the corresponding parallelization directives ( see fig .
[ flowdiag ] ) .
the mrrp and rssr tests have been run for different grids on a sgi cray - origin 2000 computer .
tables [ tab_perf.gsch ] and [ tab_perf.essr ] show the total execution time for every run as a function of the number of cpus used .
we also give the _ speed up _ factor , defined as the cpu ratio between a one processor run and one using several processors in parallel .
this factor is a measure of the degree of parallelization of the code and should ideally be equal to the number of cpus used .
the tables also contain the execution time per cell and time iteration ( tci ) .
the tci for a given number of processors is nearly independent of the number of computational cells , and can be used as a time unit to estimate the total execution time needed in a particular simulation . according to the data shown in tables [ tab_perf.gsch ] and [ tab_perf.essr ] the tci is about @xmath136 , @xmath137 and @xmath138 seconds for 1 , 4 and 8 processors , respectively .
a significant drop of the performance is noticeable for a grid of @xmath139 zones due to the phenomenon of _ cache trashing _ , because in this case the dimensions of the 3d matrices are multiples of the size of cache lines .
hence , different 3d matrices are mapped into the same set of cache lines , and every time the program needs to reference a new 3d matrix all cache lines are updated . concerning the speed up factor ,
it is noticeable from tables [ tab_perf.gsch ] and [ tab_perf.essr ] that it increases with the number of grid points , because the 3d nested loops consume a larger percentage of the total cpu time when the number of grid zones is larger .
the maximum speed up factors are 3.7 and 6.5 for 4 and 8 cpus , respectively .
we also notice a _ super linear _
behaviour , for the largest grid , for the mrrp test problem . as typical 3d simulations
are performed with zone numbers larger than the ones used in the test runs , we expect to reach even larger speed up factors in these applications .
the number of mflops ( millions of floating point operations per second ) achieved by the code is about 60 on one processor ( r10000 ) of a sgi cray
origin 2000 computer .
the theoretical peak speed of such a processor is 400 mflops . for comparison , pen ( 1998 )
reports a performance of 48 mflops for his 3d adaptive moving mesh classical hydrodynamic code using a sgi power challenge machine with r8000 processors ( 300 mflops theoretical peak speed ) .
finally , we compare the performance of genesis achieved on the pa8000 processor of hewlett packard with that obtained on the r10000 processor of silicon graphics . for the comparison we used a hpj280 workstation equipped with a pa8000 processor with a 180 mhz clock and a cache memory of 512kbytes and a sgi cray - origin 2000 equipped with a r10000 processor with a 195 mhz clock and 4 mbytes of cache memory .
the test problem selected for the comparison was the relativistic spherical shock reflection test ( rssr ) with an inflow velocity of @xmath128 .
test runs were done with four different grids .
the resulting execution times per zone and time step ( tci ) are given in table [ hpvssgi ] .
we find that @xmath140 . from table
[ hpvssgi ] we can infer a general trend .
the tcis obtained on both machines tend to become similar when the number of zones increases .
furthermore , the tci for the hp machine is nearly independent of the number of zones , while the tci for the sgi machine increases with that number .
this behaviour may result from the fact that the problem size always leads to an overflow of the cache memory on the hp workstation , while this does not generally happen for the larger cache memory of the sgi machine .
next we discuss an astrophysical application computed with genesis , namely the 3d simulation of a relativistic jet propagating through an homogeneous atmosphere .
the properties of the jet are those of model c2 in mart et al .
( 1997 ) . the beam flow velocity , @xmath141 , the beam mach number , @xmath142 , and the ratio of the rest - mass density of the beam and the ambient medium @xmath143 .
the ambient medium is assumed to fill a cartesian domain ( x , y , z ) with a size of @xmath144 , where @xmath145 is the beam radius .
the jet is injected at @xmath146 in the direction of the positive @xmath40-axis through a circular nozzle defined by @xmath147 , and is in pressure equilibrium with the ambient medium .
an ideal gas equation of state with an adiabatic exponent @xmath148 is assumed to describe both the jet matter and the ambient gas .
two different spatial resolutions with 4 and 8 zones per beam radius were used in our calculations ( figs .
[ fig_jet4],[fig_jet ] ) . in mart
1997 ) the simulation was performed in cylindrical coordinates assuming axial symmetry .
the spatial resolution was 20 zones per beam radius both in the axial and radial directions .
it is well known that the propagation of a supersonic jet is governed by the interaction of jet matter with the ambient medium , which produces a bow shock in the ambient medium and an envelope surrounding the central beam ( the _ cocoon _ , in the blandford & rees 1974 model ) .
the cocoon contains jet material deflected backward at the head of the jet . in the case of highly supersonic jets , discussed in mart
( 1997 ) , extensive , overpressured cocoons are formed with large vortices of jet matter propagating down the cocoon / ambient medium interface .
the vortices are the result of kelvin helmholtz instabilities at the interface between the jet and the shocked ambient medium .
the interaction of these vortices with the central beam causes internal shocks inside the beam .
these , in turn , affect the advance speed of the jet making it highly non stationary .
the propagation speed of the jet can be estimated from the momentum transfer between the jet and the ambient medium assuming a one dimensional flow .
for model c2 one obtains an advance speed equal to @xmath149 , whereas the 2d hydrodynamic simulation presented in mart et al .
( 1997 ) gives a mean jet advance speed of @xmath150 .
the four panels in figs . [ fig_jet4],[fig_jet ] display , from top to bottom , the logarithm of the proper rest - mass density , pressure and specific internal energy and flow lorentz factor in the plane @xmath109 at @xmath151 , when the jet has propagated about 75@xmath145 .
the analysis of cross sections of the grid perpendicular to the the jet s direction of propagation ( not shown here ) reveals acceptable symmetry of the numerical simulation , i.e both the sls and the scs properties are maintained ( see [ symmetry ] ) .
the gross morphological and dynamical properties of highly supersonic relativistic jets as inferred from our 3d simulations are qualitatively similar to those established in earlier 2d simulations . an extensive , overpressured cocoon with pressure about 20 times that in the beam at the injection point is found surrounding the jet .
the pressure and density at the head of the jet in the model with 8 zones@xmath152 are a factor of 2 larger and 1.3 smaller , respectively , than in the 2d calculation .
for the model with 4 zones@xmath152 these factors are 1 and 1.3 , respectively .
in contrast with the model with 4 zones@xmath152 , in which the propagation speed coincides with the 1d estimate , the larger pressure at the head of the jet in the model with 8 zones@xmath152 causes it to propagate through the ambient medium at a larger speed in the 3d calculation ( @xmath153 instead of @xmath149 for the 1d estimate and @xmath150 for the 2d simulation ) producing a narrower profile of the bow shock near the head . in all the simulations ,
the supersonic beam displays rich internal structure with oblique shocks effectively decelerating the flow in the beam from a lorentz factor equal to 7 at the injection point down to a value of about 4 near the head . whereas gross morphological properties are qualitatively similar in all three simulations , finer jet details ( e.g. , number , size , position and development of turbulent vortices in the cocoon )
do not agree .
however , it has been pointed out before that the fine structure is highly dependent on the numerical grid resolution ( see , e.g. , kssl & mller 1988 ) . in the model with 4 zones/@xmath145 ( see fig .
[ fig_jet4 ] ) , the material deflected at the head of the jet forms a thick , stable overpressured cocoon surrounding the beam up to the nozzle . due to the small resolution
only large vortices develope in the cocoon / external medium surface which grow slowly .
a turbulent cocoon with smaller vortices growing at a faster rate ( much more similar to the one obtained in the 2d cylindrical model ) are obtained by doubling the resolution ( compare , e.g. , the proper rest mass density pannels in figs . [ fig_jet4],[fig_jet ] ) .
we have described the main features of a novel three dimensional , high resolution special relativistic hydrodynamic code genesis based on relativistic riemann solvers .
we have discussed several test problems involving strong shocks in three dimensions which genesis has passed successfully . the performance of genesis on single and multiprocessor machines ( hp j280 and sgi cray origin 2000 ) has been investigated .
typical simulations ( in double precission ) with up to @xmath154 computational cells can be performed with 1gbyte of ram memory with a performance of @xmath155s of cpu time per zone and time step ( on a sci cray - origin 2000 with a r10000 processor ) .
currently we are working on a version of genesis suited for massively parallel computers with distributed memory ( like , e.g. , cray t3e ) .
genesis has been designed to handle highly relativistic flows .
hence , it is well suited for three dimensional simulations of relativistic jets .
first results will be presented in a separate paper ( aloy et al .
further applications envisaged are the simulation of relativistic outflows from merging compact objects ( see , e.g. , ruffert et.al .
1997 ) , from hypernovae ( paczynski 1998 ) , or collapsars ( macfadyen & woosley 1998 ) . in all these models
ultra - relativistic outflow is thought to occur and to play a crucial role in the generation of gamma - ray bursts .
this work has been supported in part by the spanish dgicyt ( grant pb94 - 0973 and accin integrada
hispano - alemana ha1996 - 0154 ) .
maa expresses his gratitude to the conselleria deducaci i cincia de la generalitat valenciana for a fellowship .
the authors gratefully acknowledge the collaboration of the user s support service of the centre europeu de paral@xmath156lelisme de barcelona ( cepba ) .
the calculations were carried out on a hp j280 and on two sgi origin 2000 , at cepba and at the centre de informtica de la universitat de valncia .
analytical expressions for the spectral decomposition of the three @xmath157 ( in 3d ) jacobian matrices @xmath158 associated to the fluxes @xmath159 of system ( [ f ] ) , @xmath160 have been given by donat et al .
( 1998 ) . in this appendix
, we explicitly show the eigenvalues and the right and left eigenvectors coresponding to matrix @xmath161 , whereas the cases @xmath39 and @xmath40 easily follows from symmetry .
the eigenvalues are : @xmath162 } \right\ } \label{lambdapm}\end{aligned}\ ] ] @xmath163 the following expressions define auxiliary quantities : @xmath164 a complete set of _ right eigenvectors _ is , @xmath165 @xmath166 @xmath167 @xmath168 the corresponding complete set of _ left
eigenvectors _ is @xmath169 @xmath170 @xmath171 @xmath172\ ] ] where @xmath173 is the determinant of the matrix of right - eigenvectors .
@xmath174 for an ideal gas equation of state it can be proven that @xmath175 is always greater than one ( in fact @xmath175 = @xmath21 ) , and @xmath173 is different from zero ( @xmath176 ) .
given a cell interface , the numerical fluxes accross such interface are given , in our modified version of marquina s flux formula , by @xmath177 in the above expression , @xmath178 and @xmath179 are the fluxes at the left and right side of the interface and @xmath180 is a five vector containing the numerical viscosity terms which is calculated according to @xmath181 .
quantities @xmath182 ( @xmath183 ) are written as a sum involving the projectors onto each eigenspace ( i.e. , direct products of left and right eigenvectors ) and the eigenvalues of the corresponding jacobian matrix @xmath184 quantities @xmath185 ( @xmath186 ) are the maximum of the modulus of the two corresponding eigenvalues at the left and right side of the interface , whereas @xmath187 ( @xmath188 ) with @xmath189 , refer to the @xmath35 ( @xmath36 ) component of the right ( left ) eigenvector @xmath36 ( @xmath35 ) . in this expression , subscripts @xmath190 correspond to @xmath191 as defined in appendix a. @xmath192 are the components of the vector of unknowns .
the superscript @xmath193 indicates that the various qunatities are calculated at each side of the interface in terms of the reconstructed variables . omitting the superscript @xmath193 ( the expressions are identical for the left and right side of the interface ) one
derives the following analytical formulas for @xmath194 : @xmath195 + p(\tilde{\lambda}_{+}\tilde{l}_{+}-\tilde{\lambda}_{-}\tilde{l}_{- } ) \right\ } + \\ & & \tilde{\lambda}_{0}p\frac{w}{h}\left\ { \frac{{\cal k}}{{\cal k } - 1 } + \frac{v_t^2+v_{tt}^2 } { 1-v_n^2 } \right \ } \\ q_2 & = & \frac{h^2w}{\delta } \left\ { m{\cal a}_{+}{\cal a}_{- } \left [ { \omega}_{+}\lambda_{+}-{\omega}_{-}\lambda_{- } \right ] + p(\tilde{\lambda}_{+}\lambda_{+}{\cal a}_{+}\tilde{l}_{+}- \tilde{\lambda}_{-}\lambda_{-}{\cal a}_{-}\tilde{l}_{- } ) \right\ } + \\ & & \tilde{\lambda}_{0}pw^2v_n \left\ { \frac{1}{{\cal k } - 1 } + 2\frac{v_t^2+v_{tt}^2}{1-v_n^2 } \right \ } \\ q_3 & = & \frac{h^2w}{\delta}v_t \left\ { m \left [ { \omega}_{+}{\cal a}_{- } - { \omega}_{-}{\cal a}_{+ } \right ] + p(\tilde{\lambda}_{+}\tilde{l}_{+}-\tilde{\lambda}_{-}\tilde{l}_{- } ) \right\ } + \\ & & \tilde{\lambda}_{0}p \left\ { \frac{w^2}{{\cal k } - 1 } + \frac{1 + 2w^2(v_t^2+v_{tt}^2)}{1-v_n^2 } \right \ } \\ q_4 & = & \frac{h^2w}{\delta}v_{tt } \left\ { m \left [ { \omega}_{+}{\cal a}_{-}- { \omega}_{-}{\cal a}_{+ } \right ] + p(\tilde{\lambda}_{+}\tilde{l}_{+}-\tilde{\lambda}_{-}\tilde{l}_{- } ) \right\ } + \\ & & \tilde{\lambda}_{0}p \left\ { \frac{w^2}{{\cal k } - 1 } + \frac{1 + 2w^2(v_t^2+v_{tt}^2)}{1-v_n^2 } \right \ } \\ q_5 & = & \frac{h}{\delta } \left\ { m \left [ { \cal a}_{- } { \omega}_{+}{\cal d}_{+}- { \cal a}_{+ } { \omega}_{-}{\cal d}_{- } \right ] + p[\tilde{\lambda}_{+}\tilde{l}_{+}{\cal d}_{+ } - \tilde{\lambda}_{-}\tilde{l}_{-}{\cal d}_{- } ] \right\ } + \\ & & \tilde{\lambda}_{0}p\frac{w}{h}\left\ { \frac{hw-{\cal k}}{{\cal k } - 1 } + \frac{(2hw-1)(v_t^2+v_{tt}^2)}{1-v_n^2 } \right \ } , \label{viscos}\end{aligned}\ ] ] with @xmath196 , @xmath197 , @xmath198 , @xmath199 , @xmath200 and @xmath201 .
quantities @xmath202 , @xmath203 y @xmath204 denote the velocities normal and parallel to the interface at which the numerical flux is to be computed , and @xmath205 are quantities proportional to the fifth component of the left eigenvectors @xmath206 and @xmath207 given by : @xmath208
in any rhd code evolving the conserved quantities eq .
( [ u ] ) in time , the variables @xmath209 have to be computed from the conserved quantities at least once per time step . in genesis
this is achieved using eqs .
( [ d])([tau ] ) and the equation of state . for an ideal gas equation of state with constant @xmath80
, this implies to find the root of the function @xmath210 with @xmath211 and @xmath212 given by @xmath213 and @xmath214 where @xmath215 and @xmath216 the zero of @xmath217 in the physically allowed domain @xmath218p_{\rm min } , \infty [ $ ] determines the pressure .
the monotonicity of @xmath217 in that domain ensures the uniqueness of the solution .
the lower bound of the physically allowed domain , @xmath219 , defined by @xmath220 is obtained from eq .
( [ rmv ] ) taking into account that ( in our units ) @xmath221 . knowing @xmath20 , eq .
( [ rmv ] ) then directly gives @xmath222 , while the remaining state quantities are straightforwardly computed from eqs .
( [ d])([tau ] ) and the definition of the lorentz factor . in genesis , the solution of @xmath223 is obtained by means of a newton raphson iteration in which the derivative of @xmath224 , @xmath225 , is approximated by @xmath226 where @xmath227 is the sound speed given by @xmath228 this approximation tends to the exact derivative when the solution is approached . on the other hand , it easily allows one to extend the present algorithm to general equations of state .
@xmath99 & 8.0(2.0)e2 & 1.1(0.3)e1 & 0.9(0.4)e2 @xmath229 & 5.2(0.4)e2 & 9.8(0.8)e2 & 1.1(0.3)e2 @xmath230 & 4.5(0.2)e2 & 9.2(0.5)e2 & 1.1(0.1)e2 @xmath231 & 3.7(0.4)e2 & 7.0(0.9)e2 & 7.0(2.0)e3 @xmath100 & 2.5(0.2)e2 & 4.8(0.7)e2 & 5.0(2.0)e3 ccccccc @xmath232 & 19.3(0.3)e+0 & 290.8(0.4)e2 & 2.4(0.1)e2 & & & @xmath233 & 10.8(0.2)e+0 & 147.2(0.7)e2 & 10.1(0.4)e3 & 0.99 & 0.84 & 1.26 @xmath234 & 49.2(0.7)e1 & 85.0(1.0)e2 & 92.8(0.8)e4 & 0.80 & 1.14 & 0.14 @xmath235 & 25.2(0.2)e1 & 37.3(0.1)e2 & 3.4(0.1)e3 & 1.19 & 0.97 & 1.44 @xmath236 & 13.8(0.1)e1 & 187.4(0.7)e3 & 17.3(0.4)e4 & 0.99 & 0.87 & 0.98 cccc @xmath114 & 90.7(0.5)e4 & 96.6(0.5)e4 & 80.3(0.5)e4 @xmath237 & 58.0(0.8)e4 & 72.0(0.8)e4 & 12.6(0.1)e3 @xmath238 & 100.3(0.5)e5 & 79.3(0.5)e4 & 72.0(0.8)e4 @xmath239 & 61.0(0.8)e4 & 93.0(0.1)e4 & 85.6(0.1)e4 @xmath240 & 65.2(0.1)e4 & 103.0(0.1)e4 & 81.3(0.5)e4 @xmath115 & 141.0(0.1)e5 & 340.1(0.1)e4 & 325.7(0.5)e5 ccccccc @xmath241 & 11.8(0.2)e+0 & 30.3(0.4)e+0 & 80.0(3.0)e3 & & & @xmath242 & 76.5(0.7)e1 & 20.1(0.2)e+0 & 55.8(0.6)e3 & 1.09 & 1.03 & 0.91 @xmath127 & 57.5(0.8)e1 & 15.5(0.2)e+0 & 41.0(0.8)e3 & 1.01 & 0.92 & 1.09 @xmath122 & 45.2(0.8)e1 & 12.5(0.1)e+0 & 32.4(0.5)e3 & 0.99 & 0.97 & 1.07 cccccccc @xmath244 & 1 & 3.91e2 & & 86 & 5.34e5 & 64.73/ & 4 & 1.13e2 & 3.48 & & 1.54e5 & 59.03/236.11 & 8 & 6.02e1 & 6.50 & & 8.22e6 & 58.58/468.65 @xmath139 & 1 & 3.85e3 & & 118 & 1.24e4 & 30.05/ & 4 & 1.84e3 & 2.09 & & 5.94e5 & 16.25/65.00 & 8 & 1.49e3 & 2.59 & & 4.81e5 & 10.46/83.70 @xmath245 & 1 & 5.56e3 & & 150 & 6.26e5 & 62.04/ & 4 & 1.56e3 & 3.57 & & 1.75e5 & 56.77/227.08 & 8 & 8.46e2 & 6.58 & & 9.52e6 & 53.99/431.88 @xmath246 & 1 & 1.31e4 & & 183 & 6.35e5 & 62.72/ & 4 & 3.66e3 & 3.57 & & 1.78e5 & 57.02/228.08 & 8 & 2.36e3 & 5.54 & & 1.15e5 & 45.47/363.75 @xmath247 & 1 & 8.94e4 & & 265 & 9.23e5 & 45.12/ & 4 & 1.84e4 & 4.87 & & 1.90e5 & 54.92/219.68 & 8 & 1.15e4 & 7.80 & & 1.18e5 & 44.74/357.91 & 16 & 7.39e3 & 12.09 & & 7.64e6 & 35.90/574.41 @xmath248 & 1 & 8.73e2 & & 114 & 8.40e5 & 62.53/ & 4 & 2.53e2 & 3.45 & & 2.44e5 & 56.88/225.67 & 8 & 1.51e2 & 5.78 & & 1.45e5 & 50.15/401.21 @xmath249 & 1 & 4.94e3 & & 198 & 9.08e5 & 62.27/ & 4 & 1.50e3 & 3.29 & & 2.76e5 & 52.88/211.54 & 8 & 1.10e3 & 4.49 & & 2.02e5 & 37.54/300.28 @xmath250 & 1 & 2.15e4 & & 369 & 9.49e5 & 61.76/ & 4 & 6.13e3 & 3.51 & & 2.71e5 & 55.40/221.60 & 8 & 3.41e3 & 6.30 & & 1.51e5 & 51.35/410.79 @xmath251 & 1 & 9.92e4 & & 890 & 9.63e5 & 62.09/ & 4 & 2.78e4 & 3.57 & & 2.70e5 & 56.41/225.65 & 8 & 1.57e3 & 6.31 & & 1.53e5 & 48.97/391.79 ccccccc @xmath248 & sgi & 8.73e2 & 114 & 8.40e-5 & hp & 2.11e3 & 101 & 2.29e-4 @xmath249 & sgi & 4.94e3 & 198 & 9.08e-5 & hp & 1.33e4 & 220 & 2.20e-4 @xmath250 & sgi & 2.15e4 & 369 & 9.49e-5 & hp & 4.76e4 & 357 & 2.17e-4 @xmath251 & sgi & 9.92e4 & 890 & 9.63e-5 & hp & 2.16e5 & 879 & 2.12e-4 aloy et al .
1998 , in preparation anile , a.m. 1989 , relativistic fluids and magneto - fluids ( cambridge university press ) blandford , r.d . , & mckee , c.f .
1976 , phys .
fluids , 19 , 1130 blandford , r.d . , & rees , m.j .
1974 , , 169 , 395 begelman , m.c . , rees , m.j .
, & sikora , m. 1994 , , 429 , l57 cavallo , g. & rees , m.j . , 1978 , , 183 , 359 centrella , j. & wilson , j.r . 1984 , , 54 , 229 colella , p. , & woodward , p.r . 1984 , jcp , 54 , 174 davis , s.f .
1984 , nasa contractor rep . 172373 , icase rep .
, no.84 - 20 donat , r. , font , j.a .
, ibez , j.m@xmath252 . , & marquina , a. 1998 , jcp , in press donat , r. , & marquina , a. 1996 , jcp , 125 , 42 duncan , g.c . , & hughes , p.a .
1994 , , 436 , l119 eichler , d. , livio , m. , piran , t. & schramm , d.n . , 1989 , , 340 , 126 eulderink , f. , & g. mellema , g. 1994 , , 284 , 652 fishman , g. & meegan , c. , 1995 , _ ann .
_ , 33 , 415 font , j.a . , ibez , j.m@xmath252 . , marquina , a. , & mart , & j.m@xmath252 .
1994 , , 282 , 304 falle , s.a.e.g . , &
komissarov , s.s .
1996 , , 278 , 586 gmez , j.l . ,
mart , j.m@xmath252 , marscher , a.p . ,
ibez , j.m@xmath252 . , & alberdi , a. 1997 , , 482 , l33 gmez , j.l . ,
mart , j.m@xmath252 , marscher , a.p . ,
ibez , j.m@xmath252 & marcaide , j.m .
1995 , , 449 , l19 goodman , j. , 1986 , , 308 , l47 hawley , j.f . ,
smarr , l.l . , & wilson , j.r .
1984 , , 55 , 221 kirchbaum , t.p . ,
quirrenbach , a. , & witzel 1992 , in variability of blazars , ed .
e. vataoja & m. valtonen ( cambrige : cambridge univ .
press ) , 331 .
kssl , d. , & mller , e. 1988 , , 206 , 204 koide , s. 1997 , , 487 , 66 koide , s. , nishikawa , k .-
& muttel , r.l .
1996 , , 463 , l71 komissarov , s.s . , &
falle , s.a.e.g .
1997 , , 288 , 833 leveque , r.j .
1991 , numerical methods for conservation laws ( birkhuser ) macfadyen , a. & woosley , s.e . , 1998 , ,
submitted ; and astro - ph/980102748 marquina , a. 1994 , siam j. scient .
, 15 , 892 mart , j.m@xmath252 . 1997 , in relativistic jets in agns , ed . by m. ostrowski , m. sikora , g. madegski , m. begelman ( astronomical observatory of the jagiellonian university , cracovia ) , p.90 mart , j.m@xmath252 . ,
ibez , j.m@xmath252 . , &
miralles , j.a .
1991 , , 43 , 3794 mart , j.m@xmath252 . , & mller , e. 1994 , j. fluid mech .
, 258 , 317 mart , j.m@xmath252 . , & mller , e. 1996 , jcp , 123 , 1 mart , j.m@xmath252 . ,
mller , e. , & ibez , j.m@xmath252 .
1994 , , 281 , l9 mart , j.m@xmath252 . ,
mller , e. , font , j.a .
& ibez , j.m@xmath252 .
1995 , , 448 , l105 mart , j.m@xmath252 . ,
mller , e. , font , j.a . ,
ibez , j.m@xmath252 .
& marquina , a. 1997 , , 479 , 151 mszros , p. , 1995
, in _ proc . of the 17th texas symp . on relativistic astrophysics and cosmology _ , h. bhringer , g.e
morfill , & j.e .
trmper ( eds . ) , n. y. acad .
, p.440 mioduszewski , a.j . ,
hughes , p.a . , & duncan , g.c .
1997 , , 476 , 649 mirabel , i.f . , &
rodriguez , l.f .
1994 , , 371 , 46 mochkovitch , r. , hernanz , m. , isern , j. & martin , x. , 1993 , , 361 , 236 nishikawa , k .-
, koide , s. , sakai , j .-
christodoulou , d.m . ,
sol , h. , & mutel , r.l .
1998 , , 498 , 166 noh , w.f .
1987 , jcp , 72 , 1 pacyski , b. , 1986 , , 308 , l43 pacyski , b. , 1998 , , 494 , l45 pen , u .- l .
1998 , , 115 , 19 piran , t. , 1997 , in _ unsolved problems in astrophysics _
bahcall & j.p .
ostriker ( eds . ) , princeton univ . press , p.343 piran , t. , shemi , a. & narayan , r. , 1993 , , 263 , 861 popham , r. , woosley , s.e . & fryer , c. , 1998 , , submitted ; and astro - ph/9807028 ruffert , m. , janka , h .- t . ,
takahashi , k. & schfer , g. 1997 , , 319 , 122 schneider , v. , katscher , v. , rischke , d.h . , waldhauser , b. , marhun , j.a .
, & munz , c .- d .
1993 , jcp , 105 , 92 shu , c.w .
, & osher , s.j . 1989 , jcp , 83 , 32 strang , g. 1968 , siam j. num .
, 5 , 506 tingay , s.j . , et al . 1995 , , 374 , 141 van leer , b. 1979 , jcp , 32 , 101 van putten , m.h.p.m . 1993 , , 408 , l21 van putten , m.h.p.m . 1996 , , 467 , l57 wen , l. , panaitescu , a. , & laguna , p. 1997
, , 486 , 919 wilson , j.r . 1979 , in sources of gravitational radiation , ed . by l.l .
smarr ( cambridge university press ) woosley , s.e . , 1993 , , 405 , 273
|
the main features of a three dimensional , high resolution special relativistic hydro code based on relativistic riemann solvers are described . the capabilities and performance of the code
are discussed . in particular , we present the results of extensive test calculations which demonstrate that the code can accurately and efficiently handle strong shocks in three spatial dimensions .
results of the performance of the code on single and multi - processor machines are given .
simulations ( in double precision ) with @xmath0 computational cells require less than 1 gb of ram memory and @xmath1 cpu seconds per zone and time step ( on a sci cray origin 2000 with a r10000 processor ) .
currently , a version of the numerical code is under development , which is suited for massively parallel computers with distributed memory architecture ( like , e.g. , cray t3e ) .
|
one of the testing grounds for various models of grand unification is calculating the strong coupling constant @xmath6 using , as input , the experimental values of the electromagnetic coupling constant @xmath7 and @xmath8 , where @xmath9 is the weinberg angle .
these calculations have been repeatedly carried out in different models and under different assumptions ( for recent reviews see , _
e.g. _ , ref .
it has been shown , in particular , that the simplest grand unification based on the standard model ( sm ) and @xmath10 gauge group leads to too small a value of the strong coupling constant , @xmath11 @xcite and is , thus , ruled out @xcite .
in contrast , supersymmetric models generally predicted @xmath6 in agreement @xcite with experimental data available at that time .
a straightforward supersymmetrization of sm gives rise to the minimal supersymmetric standard model ( mssm ) @xcite .
actually , to fully specify the model one has to make an additional assumption about the pattern of supersymmetry ( susy ) breaking .
the most popular mechanism is that of soft breaking in which one adds to the lagrangian all possible soft susy breaking terms and treats them as independent parameters .
such terms arise , _
e.g. _ , when the mssm is coupled to supergravity @xcite .
this mechanism of generating soft terms is so deeply rooted that quite often in the current literature no distinction is made between the mssm _ per se _ and the mssm plus the assumptions of the minimal supergravity - based susy breaking .
in fact , an overwhelming majority of papers devoted to even purely phenomenological studies of the mssm assume some ( but typically not all ) relations stemming from minimal supergravity , _
e.g. _ , the relation between the mass parameters of the gauginos of @xmath12 and @xmath13 . encouraged by early studies @xcite , many authors ( see , _ e.g. _ , refs .
@xcite ) then studied unification in the context of the mssm coupled to minimal supergravity .
the set of susy breaking terms generated this way is quite restrictive .
in particular , in the context of minimal @xmath14 supergravity the masses of all gauginos gluinos of @xmath15 , winos of @xmath12 and the bino of @xmath13 turn out to be the same at the planck scale .
similarly , the soft mass parameters of all squarks and sleptons are equal at that scale . in this restrictive model ,
which was called constrained mssm ( cmssm ) @xcite , one assumes universal masses for all the gauginos ( @xmath16 ) and all the scalars ( @xmath17 ) at the gut scale , and often additionally imposes a mechanism of radiative electroweak symmetry breaking ( ewsb ) @xcite . accepting these assumptions
one arrives at quite definite predictions for the spectra of masses of the model at the weak scale and for @xmath6 . for example
, the gluino turns out to be roughly three times heavier than wino @xcite .
furthermore , @xmath6 generally decreases with increasing @xmath16 and @xmath17 .
restricting @xmath16 and @xmath17 ( or alternatively all the masses ) below roughly 1tev leads to @xmath18 @xcite .
for example , an updated analysis of ref .
@xcite quotes @xmath19 .
the theoretical error here is mostly due to uncertainty associated with the so - called threshold corrections at the gut and low ( susy breaking ) scales and higher - dimensional non - renormalizable operators ( nro s ) in the gut scale lagrangian .
the above prediction for @xmath6 was considered as a great success and the strongest evidence in favor of the mssm in light of the fact that , as was believed , the direct measurement of the strong coupling constant at lep and sld yields @xmath20 @xcite .
recently it has been pointed out , however , that qcd can not tolerate such a large value of the coupling constant @xcite .
a wealth of low - energy data indicates that @xmath6 must be very close to 0.11 @xcite , three standard deviations below the alleged lep / sld value .
a method of determining @xmath0 which seems to be clean theoretically is extracting @xmath0 from deep inelastic scattering ( dis ) @xcite .
a similar number is obtained in the lattice qcd @xcite .
another reliable approach is using @xcite ( euclidean ) qcd sum rules .
the observation of ref .
@xcite motivated a new analysis of the @xmath21 sum rules @xcite claiming the record accuracy achieved so far , @xmath22 the apparent clash between the low - energy determinations of the strong coupling constant and those at the @xmath23 peak may be explained @xcite by contributions going beyond sm which were not taken into account in the global fits .
it should be stressed that the two scenarios large @xmath0 versus small @xmath0 can not coexist peacefully , as it is sometimes implied in the current literature .
our starting point is the assumption that the large @xmath0 option @xcite , inconsistent with crucial features of qcd , will eventually evaporate and the value of the strong coupling constant at @xmath24 will stabilize close to 0.11 .
in fact , in ref . @xcite it has been argued that the systematic error usually quoted in the lep number is grossly underestimated , and that at present lep experiments can only claim @xmath25 .
the question arises whether grand unification within the framework of the mssm can accommodate small @xmath2 .
this study addresses this question .
our task is to sort out assumptions ( sometimes implicit ) which are inevitable in analyses of this type and to find out which assumptions of the cmssm absolutely preclude one from descending to small @xmath6 and , therefore , have to be relaxed .
there are several possible ways to reconcile the prediction for @xmath6 in supersymmetric grand unification with @xmath26 .
one is to remain in the context of the cmssm but adopt a heavy susy scenario with the susy mass spectra significantly exceeding 1tev .
this scenario would not only put susy into both theoretical and experimental oblivion but is also , for the most part , inconsistent with our expectations that the lightest supersymmetric particle ( lsp ) should be neutral and/or with the lower bound on the age of the universe of at least some 10 billion years @xcite .
another possibility is to invoke large enough negative corrections due to gut - scale physics .
the issue has been reanalyzed in a very recent publication @xcite . under a natural assumption ( the so - called no - conspiracy assumption )
it was found that @xmath27 . relaxing this assumption
one can , in principle , construct models of the cmssm with large negative contributions coming , say , from nro s which could decrease the value of @xmath6 by @xmath28 @xcite .
( alternatively , one can entertain the possibility of an intermediate scale @xcite around @xmath29 whose existence is motivated by other reasons . in this case , however , many more unknowns affect the running of the gauge couplings and one can not really talk about _ predicting _ @xmath6 . )
none of these possibilities seem particularly appealing to us .
although it may well happen that the gut - scale and nro corrections are abnormally large , the guiding idea of grand unification becomes much less appealing in this case , and the predicitive power is essentially lost . indeed ,
by appropriately complicating gut - scale physics one could , perhaps , achieve gauge coupling unification even in the standard model . below we will discuss an alternative route .
we will adopt a down - to - earth , purely phenomenological attitude , with no assumptions about mechanisms of susy breaking .
we do not assume @xmath14 supergravity , nor any mass relations associated with this scheme , for instance , the equality of the gaugino masses at the gut scale .
if no theoretical scheme for the mass generation of susy partners is specified one is free to consider any values of these masses .
our task is to try to find out what pattern of masses is preferred by phenomenology .
we consider the mssm and limit ourselves to a minimal set " of assumptions : ( i ) all gauge coupling constants are exactly equal to each other at the gut scale ; ( ii ) the breaking of supersymmetry occurs below 1tev .
we will show that by relaxing the cmssm to the mssm one can easily descend to @xmath30 .
the only effect which is actually important in dramatically reducing the minimal value of @xmath6 is untying the gluino and wino masses .
one firm conclusion is a relatively light gluino ( in the ballpark of 100gev , and typically below 200gev ) and a relatively heavy wino ( at least a few hundredgev ) , _ i.e. _ , a relation opposite to the one emerging in the cmssm .
this summarizes our main results .
the procedure for predicting @xmath6 assuming gauge coupling unification has been adequately described in the literature ( see , _
e.g. _ , ref .
@xcite and references therein ) , and we will only summarize it briefly here . the strategy is simple : the coupling constants @xmath31 and @xmath32 ( which are known more accurately than @xmath0 ) are evolved from their experimental values at @xmath24 up to the point where they intersect ( which thus defines the unification scale @xmath33 and the gauge strength @xmath34 ) . at that point
one identifies @xmath0 with @xmath34 and runs it down to @xmath24 , thus predicting the value of @xmath6 as a function of input parameters . one- and two - loop corrections
are taken into account .
the renormalization group equations ( rge s ) for the gauge couplings are given by @xmath35 where @xmath36 , @xmath37 and @xmath38 .
the one - loop coefficients @xmath39 of the @xmath40 functions for the gauge couplings change across each new running mass threshold . in the mssm
they can be parametrized as follows @xcite @xmath41 where @xmath42 . in eqs .
( [ b1:eq])([b3:eq ] ) @xmath43 stands for the ( mass degenerate ) higgsino fields , @xmath44 for the winos , the partners of the @xmath12 gauge bosons ( @xmath45 ) , and @xmath46 stands for the gluino , all taken to be mass eigenstates in this approximation .
also , in this approximation @xmath47 stands for a heavy higgs doublet , as explained in ref .
( the full 2-loop gauge coupling @xmath40-functions for the sm and the mssm which we use in actual calculations can be found , _
e.g. _ , in ref .
@xcite . ) eqs .
( [ b1:eq])([b3:eq ] ) represent so - called leading log approximation and involves some simplifications
. however , as we will argue later , it will be sufficient to present the basic points of our analysis and answer the question how low one could descend in the values of @xmath6 assuming only strict unification of the gauge couplings in the mssm .
the prediction for @xmath6 depends on the adopted values of the input parameters : @xmath7 , @xmath48 , and @xmath49 .
it also receives corrections from : the two - loop gauge and yukawa contributions , scheme dependence ( @xmath50 _ versus _
@xmath51 ) , mass thresholds at the electroweak scale and , finally , the gut - scale mass thresholds and nro contributions .
we will discuss these effects in turn now .
the input values of @xmath31 and @xmath32 at @xmath52 can be extracted from the experimental values of @xmath53 and @xmath48 . for the electromagnetic coupling
we take @xcite @xmath54 recently , three groups have reanalyzed @xmath53 @xcite and obtained basically similar results : @xmath55 ( martin and zeppenfeld ) , @xmath56 ( eidelman and jegerlehner ) , and @xmath57 ( swartz ) .
adopting even the largest ( central ) value of swartz would shift @xmath6 up by only 0.001 @xcite .
the range of input values of @xmath48 is rather critical .
this sensitivity is due to the fact that @xmath58 does not change between @xmath52 and the gut scale @xmath59 as much as the other two couplings .
thus , a small increase in @xmath48 has an enhanced ( and negative ) effect on the resulting value of @xmath6 .
following ref .
@xcite we assume @xcite @xmath60 .
\label{s2winput : eq}\ ] ] moreover , the global analysis of ref .
@xcite implies that in the mssm @xmath61 .
recently , both the cdf and d0 collaborations have reported discovery of the top quark and quoted somewhat higher mass ranges : @xmath62 ( cdf ) @xcite and @xmath63 ( d0 ) @xcite .
such high ( central ) values of @xmath49 would lower @xmath48 and _ increase _ @xmath6 by 0.002 and 0.005 , respectively . including the two - loop terms in the rge s increases @xmath6 by about 10% .
there are two types of contributions to @xmath6 at the two - loop level .
pure gauge term yields @xmath64 if one assumes susy in both one- and two - loop coefficients of the @xmath40 function all the way down to @xmath52 .
this is the most important correction to the one - loop value of @xmath6 .
if , instead , the two - loop coefficients of the pure gauge part are changed to their sm values at @xmath65 , one finds an additional shift @xmath66 .
since this shift is negligibly small , we keep the two - loop coefficients supersymmetric all the way down to @xmath24 .
corrections due to the yukawa - coupling contribution to the rge s are also small , although negative @xcite . in the limit of large top yukawa coupling ( @xmath67 , @xmath68 , as in the small @xmath69 scenario ) one finds @xmath70 while even in the extreme case of the large @xmath71 scenario ( @xmath72 ) @xmath73 , in agreement with ref .
@xcite . above @xmath74tev
we also change from the conventional @xmath50 scheme , that we use throughout this paper , to the fully supersymmetric @xmath51 scheme .
the corresponding shift in @xmath6 is about 0.0002 and is negligible numerically @xcite . before proceeding to discussing in more detail the contribution from one - loop threshold effects ,
a remark is in order on possible corrections from the gut - scale mass thresholds and nro s . since in this paper
we look for an alternative way of lowering @xmath6 , we switch off all corrections from the gut - scale physics whatsoever . as was noted previously @xcite they are gut - model dependent and , in principle , can be sizeable .
for instance , according to refs .
@xcite the corresponding effect in @xmath6 can be as large as @xmath75 ; a factor of 2.5 larger effect is needed , however , to ensure @xmath26 . building a fully elaborated and phenomenologically acceptable model of this type
seems to be a task for the future .
what remains to be done is to explain our treatment of the mass thresholds at the electroweak and susy scales .
we use usual the step - like approximation in the coefficients of the @xmath40-function , eqs .
( [ b1:eq])([b3:eq ] ) . in the one - loop coefficients
the jumps occur at the positions of the masses of the individual particles while in the two - loop coefficients it is sufficient , to our accuracy , to consider one jump at a common susy scale , as explained above . as a matter of fact , with no loss of accuracy , we take this scale in the two - loop coefficients to be lower than @xmath24 so that in our evolution from @xmath33 down to @xmath24 we treat the two - loop coefficients as fully supersymmetric . also , the @xmath76 quark is not frozen at @xmath49 in the two - loop coefficients .
it is well known that the step - like approximation is not absolutely accurate in the problem of the coupling constant evolution ( see , _ e.g. _ , ref .
@xcite for a recent discussion ) , especially if the mass thresholds are rather close to @xmath24 , as is the case with @xmath76 quark .
we find that the other thresholds are far less important , since , as we vary their positions , the effect of the variation mimics the non - logarithmic corrections omitted in the step approximation .
the error in @xmath6 due to the inacuracy of our approximation of the @xmath0 evolution at @xmath49 is less than 1% and is , thus , unimportant .
the question we want to address is whether supersymmetric grand unification necessarily predicts large values of @xmath77 as long as all susy masses are restricted to lie below 1tev .
this is indeed the case in the cmssm with additional assumptions of common gaugino mass and common scalar mass , as described in the introduction . in order to track the role of these mass relations we begin by treating the masses of the different types of states as completely independent parameters .
we choose to remain open - minded and not biased by any additional ( even well - motivated ) assumptions about the parameters involved , other than the basic idea of gauge coupling unification .
thus , we assume no relation between squarks and sleptons , or between the gauginos .
( actually , the structure of supersymmetry alone forces certain relations between sfermion masses and gaugino masses , thus disallowing , for example , very light squarks and very heavy gauginos @xcite .
we will see below that this will not have any substantial effect on our results . )
we also do not impose a mechanism of radiative electroweak symmetry breaking .
we will see _ a posteriori _ that requiring ewsb will not change our conclusions significantly . in fig .
[ allmass : fig ] we show @xmath6 as a function of the mass of each relevant type of state . we assume all other masses to be degenerate and equal to either 100gev or 1tev . generally , we will treat all squarks and all sleptons as mass - degenerate .
the only exception to this rule will be the scalar top states , @xmath78 and @xmath79 .
this is because their masses are typically expected to be significantly different from the other squarks and from each other .
it is obvious from the form of the @xmath40-functions , eqs .
( [ b1:eq])([b3:eq ] ) , that the resulting value of @xmath80 will most sensitively depend on two parameters only : the gluino mass @xmath81 and the soft mass parameter @xmath82 of the wino .
the reasons are twofold : not only are their @xmath40-function coefficients among the largest but also they change only one out of the three @xmath39 s .
[ allmass : fig ] clearly confirms our expectation .
also , table [ als : table ] shows @xmath6 for several choices of relevant parameters .
the first four rows are meant to demonstrate the dependence of @xmath6 on @xmath82 and @xmath81 .
we are interested in the lowest possible values of @xmath6 allowed by ( strict ) grand unification . as it is obvious from fig .
[ allmass : fig ] , minimization of @xmath6 requires minimizing @xmath81 and @xmath83 while simultaneously maximizing the masses of the wino , the sleptons , the higgsino , and of the heavy higgs .
we have also verified that , in order to minimize @xmath6 , one should also set @xmath84 ( @xmath85 ) at its lowest ( largest ) possible value . since the standard " prediction for @xmath6 emerging in the cmssm is quoted above under the assumption that all sparticles are lighter than 1tev we accordingly restrict all the masses to that range . at the lower end , we allow the masses to lie as low as @xmath86 .
( lowering this limit down to @xmath24 would not noticeably change @xmath6 @xcite . ) in the last row of table [ als : table ] we show the lowest value of @xmath6 obtained by varying all the mass parameters between 100gev and 1tev .
experimental bounds on most of those states are still less than @xmath24 . even for @xmath81 and the masses of the squarks there are no inescapable lower bounds , other than roughly @xmath87 from lep @xcite .
( very recently , the d0 collaboration @xcite has published new improved limits : @xmath88 for any @xmath89 and @xmath90 for @xmath91 .
adopting these limits in the last row of table [ als : table ] would increase @xmath92 by only 0.002 and 0.003 , respectively . )
we also display in fig .
[ allmass : fig ] @xmath92 ( thick solid line ) as a function of the mass of each individual state , while setting all the other masses as in the last row of table [ als : table ] .
it is clear that in general one can easily obtain values of @xmath6 small enough to accomodate the range @xmath93 which we favor .
furthermore , @xmath6 shows little dependence on the masses of the states other than the @xmath12 and @xmath15 gauginos .
therefore one actually has considerable freedom in choosing the other masses as desired .
this justifies our approach of assuming all sleptons to be mass - degenerate , and similarly with squarks .
furthermore , relatively weak dependence of @xmath6 on the mass of the higgsino ( which we approximate by the higgs / higgsino mass parameter @xmath94 ) shows that imposing ewsb would probably not lead to any strong increase in the lower bound on @xmath6 .
this is because the conditions of ewsb determine @xmath94 in terms of ( soft ) higgs mass parameters which influence @xmath6 even less .
it is also evident from the gluino window of fig .
[ allmass : fig ] that the mass of the gluino is strongly confined to rather small values in the range of a few hundred gev only .
this is a distinctive feature and a strong prediction of our approach . the exact value of the upper bound on @xmath81 that one allows clearly depends on how large gut- related corrections one assumes and also how large values of @xmath6 one is willing to accept .
= 3.5 in on the other hand , the wino mass parameter @xmath82 should preferably be larger than @xmath81 , contrary to what is commonly expected .
this is clearly shown in fig .
[ winogluino : fig ] where , in the plane ( @xmath95 ) , we plot the lowest allowed values of @xmath6 found by assuming all other mass parameters as in the last row of table [ als : table ] .
it is clear that @xmath93 favors relatively small @xmath81 and large @xmath82 . among perhaps the most commonly assumed , and least questioned , relations are the ones between the mass parameters of the gauginos @xmath96 where the susy breaking parameters @xmath97 , @xmath82 and @xmath81 of the bino , the wino , and the gluino states are evaluated at the electroweak scale .
virtually all phenomenological and experimental studies adopt at least the relation ( [ monemtwo : eq ] ) . strictly speaking , however , both relations are not necessary in the context of the mssm .
they both originate from the assumption that , in minimal @xmath10 @xmath14 supergravity , the kinetic term of the gauge bosons and gauginos is equal to a kronecker delta . clearly , _ a priori _ this assumption is not an indispensible part of the mssm . from our previous analysis
it is evident that any additional assumption relating the masses of the wino and the gluino will have a significant impact on the prediction of @xmath6 . in fig .
[ xgluinogut : fig ] we plot @xmath92 versus @xmath81 for @xmath98 .
we set all the other masses in such a way as to minimize @xmath6 , as in the last row of table [ als : table ] .
we also show the lowest allowed @xmath6 ( thick solid curve ) as a function of @xmath81 only by setting also @xmath99 .
it is clear that the usually assumed ratio @xmath100 forces @xmath6 above @xmath101 . to be consistent with @xmath26
the ratio @xmath102 is required .
this corresponds to @xmath103 at the gut scale .
= 3.5 in furthermore , fig .
[ xgluinogut : fig ] shows that the mass of the gluino must again be rather small , @xmath104 , in the absence of large gut - scale corrections , unless one allows for the wino mass parameter @xmath82 significantly above 1tev .
the above considerations put into doubt also the relation ( [ monemtwo : eq ] ) , which has its root in the same assumption of the equality of all the gaugino masses at the gut scale .
it is true that the mass parameter of the bino @xmath97 does not enter eqs .
( [ b1:eq])([b3:eq ] ) and can not be directly related to @xmath82 and @xmath81 . however , in the cmssm the lightest neutralino almost invariably comes out to be an almost pure bino @xcite and @xmath105 .
it is also an excellent dark matter candidate .
there are also stringent limits on the cosmic abundance of exotic particles with color and electric charges . requiring that the lightest ( bino - like ) neutralino be lighter than the gluino , and thus a likely candidate for the lightest supersymmetric particle ( lsp ) leads to @xmath106 ( or @xmath107 at @xmath33 ) , thus violating the relation ( [ monemtwo : eq ] )
@xcite .
many phenomenological and dark matter properties of the neutralinos depend on the relation ( [ monemtwo : eq ] ) . relaxing
it may bear important consequences for neutralino detection in accelerators @xcite and in dark matter searches @xcite , as well as in placing bounds on other sparticles .
basically , the mass of the ( lightest ) bino - like neutralino is @xmath105 . reducing the ratio @xmath108 leads to lighter neutralinos .
the region of the plane ( @xmath109 ) ( as it is usually presented ) where @xmath110 remains mostly bino - like actually increases somewhat @xcite . also ,
even rather light neutralinos with mass in the range 3gev to a few tens of gev are in principle not excluded and possess excellent dark matter properties ( @xmath111 ) @xcite .
finally , it is worth commenting that , even in the context of @xmath14 supergravity one can relax the assumptions ( [ monemtwo : eq])([mtwomgluino : eq ] ) @xcite .
this can be done by considering a general form of the kinetic term of the gauge and gaugino fields , rather than assuming it to be equal to unity . in this case one
finds that the gauge couplings at @xmath33 need not be equal ( thus making the gut energy scale @xmath33 somewhat ill - defined ) and , in general , relations among gaugino masses become arbitrary .
if , however , one assumes @xmath112 then one finds , at @xmath33 , @xmath113 @xcite . in the limit in which the gauge couplings are only slightly displaced from each other at @xmath33
we find @xmath114 .
one solution is the usual @xmath115 .
but there exist also solutions to this relation which are consistent with small @xmath6 , for example @xmath116 and @xmath117 , in agreement with what we have found above
. thus it may be possible to reconcile @xmath93 with some non - minimal versions on @xmath14 supergravity .
the version of supersymmetric grand unification considered here leads to several distinct implications .
one is the necessary existence of a relatively light gluino below @xmath5 200 gev and preferably large wino mass parameter @xmath82 .
the likely violation of the commonly assumed relations ( [ monemtwo : eq])([mtwomgluino : eq ] ) may lead to many important consequences for placing bounds on various sparticles and to more promissing prospects for neutralino dark matter searches .
below we discuss how the existence of a light gluino affects possible solutions to the long - lasting anomaly of the @xmath118 width .
furthermore , @xmath119 may lead to a significant relaxation of the constraints on @xmath71 from requiring @xmath120@xmath121 mass unification .
we discuss these points below .
if @xmath26 does indeed require the gluino mass to lie in the ballpark of 100gev , as was argued above , the question which immediately comes to one s mind is : what are other phenomenological implications of such a light gluino ? " first and foremost , with this mass , the gluino must be accessible to direct searches at the tevatron .
currently , a gluino mass range up to about 200gev is probed @xcite but no firm assumption - independent bounds can be drawn . on the other hand , with the main injector upgrade , the tevatron experiments will be able to probe @xmath81 in the range up to 300gev .
if the gluino is indeed found below some 240gev and no ( wino - like ) chargino is found at lep - ii up to some 80gev , we will know that the relation ( [ mtwomgluino : eq ] ) does not hold .
second , light gluinos propagating in loops make the corresponding radiative corrections more pronounced .
they can then become important in understanding several facts where hints on disagreement between observations and sm expectations were detected .
the most well - known example of this type is the problem of @xmath0 itself . as was noted in refs .
@xcite the gluino exchange correction to the @xmath122 vertices is positive so that the gluino correction enhances the hadronic width of @xmath23 , imitating in this way a larger value of @xmath0 .
fig . 2 of ref .
@xcite shows that the correction can reach @xmath123 in each quark channel provided that @xmath124 and @xmath125 .
with such a correction the value of @xmath0 measured at the @xmath23 peak slides down by @xmath28 solving the problem in full .
on the other hand , it seems extremely unlikely that the very same mechanism may be responsible for the alleged enhancement in the @xmath126 channel .
indeed , if we take the central value for the experimental @xmath118 width , the excess over the theoretical expectation amounts to @xmath127 mev @xcite , a factor of 5 larger than the excess produced by the gluino correction above .
one would have to descend to unacceptably low squark and gluino masses to get this factor of 5 .
recently , another possible solution of the @xmath128 problem was suggested in ref .
@xcite . in this work
the mass parameters of the mssm were also considered as _ a priori _ unrelated .
it was shown that , in order to induce large enough susy correction to reconcile the measured value of @xmath128 with the sm prediction , a relatively light ( below roughly 80gev ) higgsino - like chargino is required .
the authors also need at least one stop with a significant @xmath79 component in the same mass range . in order to examine what predictions for @xmath6
this scenario leads to we have set the higgsino mass parameter @xmath94 and @xmath83 at @xmath24 , and chosen all other mass parameters in such a way as to minimize @xmath6 , as before .
we find @xmath129 .
another problem where the relatively light gluino can help is the deficit of the semileptonic branching ratio in @xmath130 mesons and the charm multiplicity @xcite .
theoretical calculations of these quantities are at a rather advanced stage now .
both perturbative and non - perturbative effects have been considered .
the most detailed analysis of the non - perturbative effects is carried out in ref .
@xcite , with the conclusion that they can be essentially neglected in the problem at hand . as for perturbative calculations ,
they have been repeatedly discussed in the literature .
e.g. _ , recent papers @xcite and references therein . )
the theoretical prediction turns out to be rather sensitive to the choice of the value of @xmath0 and the normalization scale @xmath94 relevant to the process .
smaller values of @xmath94 and larger values @xmath0 tend to enhance the non - leptonic width and , thus , lower the prediction for the semileptonic branching ratio . on the contrary ,
larger values of @xmath94 and smaller @xmath0 suppress the non - leptonic width and enhance the branching ratio . the theoretical prediction can be made marginally compatible @xcite with the data on the semileptonic branching ratio @xcite provided that @xmath0 is chosen on the high side and @xmath94 on the low side . at the same time , if @xmath26 the prediction for br@xmath131 does not fall lower than 11.5% @xcite , while the corresponding experimental number is @xmath132 .
moreover , no reasonable choice of the parameters above allows one to eliminate a very substantial deficit in the charm multiplicity .
both discrepancies evaporate if the @xmath130 non - leptonic decays receive a contribution from the @xmath133 + gluon transition , at the level of @xmath5 15% of the total width .
then the theoretical prediction for br@xmath131 shifts down to 10.4% ; simultaneously , the charm multiplicity turns out to be within error bars . as was observed in ref .
@xcite , in supersymmetric models such a transition can naturally arise , with the right strength , if the gluino and squark masses lie in the 100gev ballpark .
what is important is that the additional graphs giving rise to @xmath133 + gluon transition do not spoil the @xmath133 + photon transition .
indeed , the ratio of the photon to gluon probabilities is @xmath134 where @xmath135 is the down quark electric charge , and @xmath136 is a numerical factor including , among other effects , an enhancement of the @xmath133 + gluon transition due to the gluon radiative corrections . according to ref .
@xcite @xmath137 to 3 . with @xmath119
the ratio is close to @xmath138 .
this means that the @xmath133 + gluon transition can well contribute at the level of 15% ; the corresponding contribution to the @xmath139 is at the level of @xmath140 , which is quite acceptable phenomenologically @xcite . it has been argued that , in the mssm alone , with no additional mass relations , the requirement of strict @xmath120@xmath121 mass unification can only be achieved in a relatively very narrow region of the ( @xmath141 ) plane for a wide range of @xmath6 @xcite .
however , it was noted in ref .
@xcite that , if @xmath6 is small @xmath142 , the above strong relation between @xmath71 and @xmath49 can be significantly relaxed provided that strict unification condition @xmath143 at the gut scale is reduced somewhat ( @xmath144 ) .
( see figs .
1 and 2 of ref . @xcite . )
gut - scale uncertainties of this size are actually typically present in gut s @xcite .
the observation that the gauge coupling constants , which look so different at the electroweak scale , evolve and converge at a scale somewhat smaller than the planck mass was crucial in the original idea of grand unification @xcite .
later on , with more accurate data and more precise calculations available , it turned out that the gauge couplings do not intersect at one point .
the fact that we are off by only a relatively very small amount is very encouraging and shows that the original idea is viable , and only details must be adjusted .
this first led people from the sm to the mssm .
this work concludes that , if @xmath6 is indeed close to @xmath145 , the gluino must be rather light , @xmath146 , and thus accessible to present direct searches .
it is also gratifying to note that , with the mass of the gluino lying in this ballpark , other problems ( like the @xmath128 excess at lep , a deficit of the semileptonic branching ratio of b - mesons , _ etc . _ ) might find their solutions as well .
finally , many studies of susy , including mass bounds on sparticles and dark matter searches , rely on the mass relations ( [ monemtwo : eq])([mtwomgluino : eq ] ) .
this analysis provides arguments for relaxing them .
this work was supported in part by the u.s .
department of energy under the grant number de - fg02 - 94er40823 .
p. langacker , _ test of the standard model and searches for new physics _ , to be published in _ precision tests of the standard electroweak model _ , ed p. langacker , world scientific , singapore , 1994 [ hep - ph/9412361 ] ; _ grand unification and the standard model _ , invited talk at int .
_ radiative corrections _ , gatlinburg , tennessee , 1994 , preprint upr-0639 t [ hep - ph/9411247 ]
. p. langacker and n. polonsky , phys
* d47 * ( 1993 ) 4028 .
( for an update see n. polonsky , _ unification and low - energy supersymmetry at one and two - loop orders _ , phd thesis , university of pennsylvania , 1994 . )
earlier calculations were phrased in different terms , but the conclusions were the same ; see , _
e.g. _ , u. amaldi , w. de boer and h. frstenau , phys .
b260 * ( 1991 ) 443 ; j. ellis , s. kelley and d.v .
nanopoulos , phys . lett .
* b260 * ( 1991 ) 131 ; p. langacker and m .- x .
luo , phys . rev .
* d44 * ( 1991 ) 817 ; f. anselmo , l. cifarelli , a. peterman , and a. zichichi , nuovo cim .
* 104a * ( 1991 ) 1817 , and nuovo cim .
* 105a * ( 1992 ) 581
. for reviews , see , _ e.g. _ , h .-
nilles , phys . rep . * 110 * ( 1984 ) 1 ; h.e .
haber and g.l .
kane , phys .
* 117 * ( 1985 ) 75 ; l.e .
ibez and g.g .
ross , in _ perspectives in higgs _
, ed . by g.l .
kane ( world scientific , singapore , 1993 ) ; r. mohapatra , _ unification and supersymmetry _
, 2nd edition , springer - verlag , 1992 .
ibez , phys . lett .
* 118b * ( 1982 ) 73 p. nath , r. arnowitt , and a. chamsedine , phys .
* 49 * ( 1982 ) 970 ; j. ellis , d.v .
nanopoulos , and k. tamvakis , phys . lett .
* 121b * ( 1983 ) 123 ; h.p .
nilles , m. srednicki , and d. wyler , phys .
120b * ( 1982 ) 346 ; r. barbieri , s. ferrara , and c. savoy , phys . lett .
* 119b * ( 1982 ) 343 . r.g .
roberts and g.g .
ross , nucl .
b377 * ( 1992 ) 571 .
roberts and l. roszkowski , phys .
b309 * ( 1993 ) 329 . g. kane , c. kolda , l. roszkowski , and j. wells , phys .
* d49 * ( 1994 ) 6173 .
v. barger , m.s .
berger , and p. ohmann , phys . rev .
* d49 * ( 1994 ) 4908 ; r. arnowitt and p. nath , phys . lett . *
b287 * ( 1992 ) 89 ; s. kelley , j.l .
lopez , d.v .
nanopoulos , h. pois , and k. yuan , phys . lett . *
b273 * ( 1991 ) 423 ; d. j. castao , e. j. piard , and p. ramond , phys . rev .
* d49 * ( 1994 ) 4882 ; b. de carlos and a. casas , phys . lett .
* b309 * ( 1993 ) 320 .
ibez and g.g .
ross , phys .
110b * ( 1982 ) 215 ; k. inoue , a. kakuto , h. komatsu , and s. takeshita , progr .
* 68 * ( 1982 ) 927 ; l. alvarez - gaum , j. polchinsky , and m. wise , nucl . phys .
* b221 * ( 1983 ) 495 ; j. ellis , d.v .
nanopoulos , and k. tamvakis , in ref .
@xcite . ref .
@xcite quotes even a higher value , @xmath147 .
the errors are believed to be dominated by theoretical uncertainties .
the most exhaustive theoretical analysis is done for the total hadronic width @xmath148 .
the error quoted above , @xmath149 , is essentially determined by analyzing @xmath148 .
g. altarelli , _ qcd and experiment status of @xmath0 _ , in _ qcd 20 years later _ ,
proceedings of the 1992 aachen workshop , eds .
p. zerwas and h. kastrup [ world scientific , singapore , 1993 ] , vol . 1 , page 172 ; s. bethke , _ summary of @xmath0 measurements _ , to be published in proc .
workshop qcd 94 , montpellier , france , july 1994 , [ preprint pitha 94/30 ] .
an example of the dis data analysis can be found , _
e.g. _ , in m. virchaux and a. milsztajn , phys .
* b274 * ( 1992 ) 221 ; a. martin , w. stirling and r. roberts , _ pinning down the glue in the proton _ , preprint ral-95 - 021 [ hep - ph/9502336 ] . a nice compilation is given in ref .
@xcite .
el - khadra , g. hockney , a. kronfeld and p. mackenzie , phys . rev . lett .
* 69 * ( 1992 ) 729 ; c. davies , k. hornbostel , g.p .
lepage , a. lidsey , j. shigemitsu and j. sloan , _ a precise determination of _
@xmath0 _ from lattice qcd _ , preprint ohstpy - hep - t-94 - 013 [ hep - ph/9408328 ] .
s. eidelman , l. kurdadze and a. vainshtein , phys . lett . * 82b * ( 1979 ) 278 .
m. voloshin , _ precision determination of @xmath0 and @xmath150 from qcd sum rules for @xmath126 _ , preprint tpi - minn-95/1-t [ hep - ph/9502224 ] ( int . j. mod
a , to appear ) .
j. ellis , s. kelley , and d.v .
nanopoulos , nucl . phys .
* b373 * ( 1992 ) 55 and phys . lett .
* b287 * ( 1992 ) 95 ; j. hisano , h. murayama , and t. yanagida , phys .
* 69 * ( 1992 ) 1014 .
v. barger , m.s .
berger , and p. ohmann , phys . rev .
* d47 * ( 1993 ) 1093 .
l. montanet , _ et al . _
( pdg ) , phys . rev .
* d50 * ( 1994 ) 1173 .
a. martin and d. zeppenfeld , _ a determination of the qed coupling at the @xmath23 pole _
, preprint mad - ph-855 [ hep - ph/9411377 ] ; m.l .
swartz , _ reevaluation of the hadronic contribution to @xmath153 _ , preprint slac - pub-6710 [ hep - ph/9411353 ] ; s. eidelman and f. jegerlehner , psi report pr-95 - 1 [ hep - ph/9502298 ] .
let us parenthetically note that some authors investigate a wider range of variations of @xmath48 .
for instance , in ref .
@xcite the highest value considered was @xmath154 . as was explained above inching @xmath48 up one
lowers the minimal value of @xmath6 .
thus , the result of ref .
@xcite in the cmssm with the planck mass correction switched off is @xmath155 .
the question whether the value @xmath154 is admissible is left open ; we stick to eq .
( [ s2winput : eq ] ) .
ibez and c. lpez , nucl . phys .
* b233 * ( 1984 ) 511 ; l.e .
ibez , c. lpez , and c. muoz , nucl . phys .
* b256 * ( 1985 ) 218 .
e.g. _ , l. galtieri , to appear in the proceeding of the _ conference on beyond the standard model iv _ , lake tahoe , december 13 - 18 , 1994 , ed .
j. gunion , t. han , and j. ohnemus ; a. jonckheere ( d0 ) , _ ibid _ ; s. eno ( d0 ) , to appear in the proceeding of the _ rencontres de moriond xxx , electroweak interactions and unified theories _ , ed .
j. tran thanh van , march 1995 ; m. paterno ( d0 ) , to appear in the proceeding of the _ les rencontres de physique de la vallee daoste _ , lathuile , ed .
g. belletini , march 1995 ; f. abe , _ et al . _
( cdf ) , phys .
* 69 * ( 1992 ) 3439 ; s. abachi , _ et al . _
( d0 ) , _ search for squarks and gluinos in @xmath158 collisions at @xmath159 _ , fermilab pub-95/057-e ( march 1995 ) ( to appear in phys .
lett . ) . in a recent work of j. bagger ,
_ et al . _ , ( _ precision corrections to supersymmetric unification _
, preprint jhu - tipac-95001 [ hep - ph/9501277 ] ) the non - leading mass - threshold effects were found to be substantial in the cmssm for @xmath160 a corrected evolution curve significantly deviates from the leading log ( ll ) result .
the reason is quite obvious : a jump in the slope of the ll evolution curve is due to the fact that at @xmath161 the wino becomes lighter than @xmath23 and freezes out .
nothing of the kind happens in our analysis and we generally do not expect the non - leading mass - threshold effects to be large .
k. griest and l. roszkowski , phys . rev .
* d46 * ( 1992 ) 3309 .
a. bartl , h. fraas , w. majerotto , and n. oshima , phys . rev .
* d40 * ( 1989 ) 1594 ; m. drees and x. tata , phys . rev . * d43 * ( 1991 ) 1971 . j. ellis , k. enqvist , d. nanopoulos , and k. tamvakis , phys . lett . * 155b * ( 1985 ) 381 .
m. drees , phys .
lett . * 158b * ( 1985 ) 409 .
k. hagiwara and h. murayama , phys .
b246 * ( 1990 ) 533 . g. bhattacharyya and a. raychaudhuri , phys . rev . * d47 * ( 1993 ) 2014 ; a. djouadi , m. drees and h. konig , phys . rev . *
d48 * ( 1993 ) 308 .
j. wells , g. kane and c. kolda , phys .
* b338 * ( 1994 ) 219 .
i. bigi , b. blok , m. shifman , and a. vainshtein , phys . lett .
* b323 * ( 1994 ) 408 .
e. bagan , p. ball and v. braun , _ charm quark mass corrections to non - leptonic inclusive b decays _ ,
preprint tum - t31 - 67 - 94 [ hep - ph/9408306 ] ; e. bagan , p. ball , v. braun , and p. gosdzinsky , _ theoretical update of the semileptonic branching ratio of b mesons _ , preprint desy-94 - 172 [ hep - ph/9409440 ] .
r. patterson , _ weak and rare decays _ , plenary talk at int .
conference on high energy physics , glasgow , july 1994 ; + p. roudeau , _ heavy quark physics _ , plenary talk at int .
conference on high energy physics , glasgow , july 1994 .
v. barger , m.s .
berger , p. ohmann , and r.j.n .
phillips , phys .
b314 * ( 1993 ) 351 ; m. carena , s. pokorski and c. wagner , nucl . phys .
* b406 * ( 1993 ) 59 .
p. langacker and n. polonsky , phys
. rev . * d49 * ( 1994 ) 1454 .
h. georgi and s.l .
glashow , phys .
* 32 * ( 1974 ) 438 ; h. georgi , h.r .
quinn , and s. weinberg , phys .
lett . * 33 * ( 1974 ) 451 .
|
we argue that supersymmetric grand unification of gauge couplings is not incompatible with small @xmath0 , even without large gut - scale corrections , if one relaxes a usual universal gaugino mass assumption .
a commonly assumed relation @xmath1 is in gross contradiction with @xmath2 .
instead , small @xmath0 favors @xmath3 .
if this is indeed the case our observation casts doubt on another commonly used relation @xmath4 which originates from the same constraint of a common gaugino mass at the gut scale .
one firm prediction emerging within the small @xmath0 scenario with the unconstrained gaugino masses is the existence of a relatively light gluino below @xmath5 200gev .
@=11 caption#1[#2]#3 @=12 # 1 + * university of minnesota * tpi - minn-95/04-t + umn - th-1330 - 95 + hep - ph/9503358 + march 1995 + ( revised version ) l. roszkowski and m. shifman +
|
the quantized hall effect@xcite provides particularly interesting tests of our understanding of electrical transport .
application of a resistance formula which treats all contacts to a two - dimensional electron gas on equal footing@xcite has considerably revised the traditional picture of the quantized hall effect and has led to the successful explanation of many novel experiments@xcite .
it is the purpose of this work to approach the low frequency electrical transport in two - dimensional electron systems ( 2des ) subject to strong magnetic fields from a similar point of view .
in contrast to the dc - transport properties , which have become increasingly well understood , the ac - transport properties have found much less attention .
however , a charge and current conserving theory for the low frequency admittance @xmath0 of a general arrangement of mesoscopic conductors has recently been worked out @xcite .
we apply this theory to hall systems in the integer quantum hall regime at a plateau . a charge and current
conserving theory requires knowledge of the non - equilibrium potential distribution inside the conductor . in the quantum hall regime
the determination of this potential becomes simple due to the formation of edge channels @xcite .
as discussed in detail by chklovskii et al.@xcite and closely related works @xcite , there occurs a decomposition of a 2des in metal - like edge channels and dielectric - like regions .
consequently , the non - equilibrium potential is also determined by the properties of the edge channels . if the edge channels behave like perfect metals they screen any excess charge .
the resulting non - equilibrium potential is determined by the geometry of the edge - channel arrangement alone . on the other hand ,
if the charge in the edge channels is not perfectly screened the non - equilibrium potential depends on the density of states of the edge channels .
the resulting potential distribution is not of geometrical nature alone but contains quantum corrections due to the finite density of states of the edge channels .
it is of particular interest to investigate to what extent such quantum corrections affect the dynamic transport properties of a 2des . + the admittance @xmath1 gives the linear current response @xmath2 at a contact @xmath3 of a device , if at contact @xmath4 a voltage oscillation @xmath5 is applied : @xmath6 the voltage variation @xmath7 is related to the variation of the electro - chemical potential @xmath8 in reservoir @xmath4 by @xmath9 , where @xmath10 is the electron charge . the theory @xcite deals with the dc - conductance , @xmath11 , and the first - order term with respect to frequency , @xmath12 , which is called the _ emittance matrix . the low frequency admittance
can then approximately be written in the form @xmath13 for an array of macroscopic conductors of which each is connected to a single contact , the emittance is just a geometrical capacitance , i.e. @xmath14 .
however , this is not true for _ mesoscopic _ conductors and conductors which connect _ different _ reservoirs @xcite .
firstly , it is not the geometrical capacitance but rather the _ electro - chemical capacitance _ which relates charges at mesoscopic conductors with voltage variations in the reservoirs .
secondly , conductors which connect different reservoirs allow a transmission of charge which leads to inductance - like contributions to the emittance .
+ we shall show that the emittance @xmath15 of a quantized hall sample is the sum over elements of the electro - chemical capacitance matrix , @xmath16 , for edge - channels @xmath17 into which charge is injected at contact @xmath4 and for edge - channels @xmath18 from which charge is emitted into contact @xmath3 .
the electro - chemical capacitance matrix , @xmath16 , is determined by considering each edge channel as a metal strip connected to a single contact .
our expression for the emittance is simple enough in order to discuss arbitrarily complicated edge - channel arrangements without much technical effort , once the electro - chemical capacitance matrix of the edge - channel arrangement is known .
we emphasize here that our theory satisfies charge and current conservation which are due to a perfect screening of electric fields in the reservoirs and in the gates used to form the conductor .
current conservation implies that the admittance satisfies @xmath19 .
+ two simple geometries can be used to illustrate the different behavior of the emittance , namely the hall bar geometry ( fig .
[ fig1].a ) , and the corbino geometry ( fig . [ fig1].b ) . we will show that in a hall bar kinetic charge motion of electrons along the edge channels dominates the coulomb interaction between the reservoirs .
the emittance is a negative electro - chemical capacitance , i.e. @xmath20 with @xmath21 . on the other hand , in the corbino geometry contacts
are located at the inner and the outer perimeter of an annular film@xcite .
hence , edge channels do not connect different reservoirs and will thus not contribute to a dc - current .
moreover , in contrast to the bar geometry in the corbino disc capacitive effects dominate and the emittance is a capacitance , i.e. @xmath22 .
+ the transverse potential profile in a cross section of these conductors is qualitatively shown shown in fig . [ fig2 ] which is to be discussed below . here
we only mention that the similarity of this potential for the two different setups applies only to the bulk of the sample .
we will assume in this work that the capacitances and emittances are dominated by the bulk and that contact capacitances can be neglected .
+ the paper is organized as follows . in sect .
[ equilibrium ] we briefly recall the edge - channel picture of a 2des at equilibrium . in sect .
[ nonequilibrium ] , we discuss the dc - nonequilibrium electric potential in terms of an electro - chemical capacitance matrix , and the expression for the dc - conductance @xmath11 is derived . in sect .
[ emittance ] we outline the theory of emittances and derive an expression for the emittance matrix @xmath23 for quantized hall samples .
the result is applied to various specific examples in sect .
[ examples ] . in sect .
[ incoherent ] we investigate the effect of a voltage probe . as an application
, we calculate in sect .
[ hall ] the longitudinal resistance and the hall resistance of a four - probe quantum hall bar for low frequencies . _
we begin with a brief discussion of important equilibrium properties of a 2des at an ( integer ) hall plateau .
consider the two - terminal quantum hall bar in fig .
[ fig1].a .
the bar is connected on either side via ideal contacts to particle reservoirs @xmath24 at electro - chemical potentials @xmath25 . here ,
@xmath26 and @xmath27 denote the chemical and the electric potential of reservoir @xmath28 , respectively .
the strong magnetic field is assumed to be perpendicular to the plane of the 2des .
translational invariance of the potential @xmath29 in the @xmath30-direction allows one to restrict the considerations to a transverse cross section of the sample .
the single - particle potential as a function of @xmath31 is sketched in fig .
[ fig2].a for the equilibrium case where the reservoirs are kept at equal electro - chemical potential , say @xmath32 where @xmath33 . for the moment
, we assume that the fermi level lies in the region between the extended bulk states of the first and the second ( spin - split ) landau levels . hence , in the bulk the states of a single landau level are completely filled ( black dots in fig . [
fig2].a ) .
at the sample boundary , however , the confinement potential strongly bends up the single - particle potential which , therefore , intersects the fermi energy .
this leads to the existence of extended states at the fermi level ( edge channels ) along the sample boundary .
for non - interacting electrons@xcite the intersection of the single particle energy with the fermi energy is sharp .
the transverse size of an edge channel is of the order of a magnetic length , @xmath34 .
the mean drift velocity of a carrier with coordinate @xmath31 points in @xmath30-direction and is given by @xcite @xmath35 this is just the lorentz drift of the center of a cyclotron orbit in an electric field .
+ in a quantized hall sample a current density exists which is a pure equilibrium phenomenon and can not lead to a current between reservoirs . for a filled landau level , the diamagnetic current density can be written in the form @xcite @xmath36 the total current through a contact is obtained by a transverse spatial integral of @xmath37 .
it vanishes at equilibrium since at both boundaries of integration , @xmath38 holds , where @xmath39 labels the edge channels . of course , this statement is valid independent of the geometrical arrangement of the edge channels as long as the cross section is constructed such that all edge channels of a contact are included . in particular , it is independent of the specific space dependence of the equilibrium potential which can be very complicated .
+ the inclusion of coulomb interaction , even within a mean - field approximation , drastically affects the results of the single - particle approach .
coulomb interactions lead to an electro - static restructuring of the edge @xcite .
the 2des is composed of alternating strips of compressible and incompressible electron liquids with finite widths .
incompressible regions where the filling factor has discrete values behave like dielectrics .
quantitative analytical predictions of the widths of the compressible and incompressible strips have been made by chklovskii et al.@xcite .
these predictions are in good agreement with numerical work by lier and gerhardts@xcite .
edge channels correspond to the compressible regions where single - particle states are partially filled and the electric potential is pinned to the fermi level ( flat parts in fig . [
fig2].a ) .
edge channels have screening properties similar to metallic strips .
the many - particle effects become important if the strength @xmath40 of the ( unscreened ) confinement field is weaker than the characteristic electric field of electron - electron interaction , i.e. if @xmath41 , where @xmath42 is the dielectric constant .
the strength of the confinement field depends on the specific fabrication of the boundaries of the 2des under consideration ( etching , gates , etc . ) .
it has been argued that compressible and incompressible strips can even become comparable in size @xcite .
interaction - dominated edge channels are useful in the theory of the fractional quantum hall effect @xcite . however , the structure of fractional edge channels is much more complicated , and we shall restrict our considerations to the integer quantum hall regime . + in the following considerations , three quantities which characterize the equilibrium state of a sample are important in order to discuss low frequency transport close to equilibrium .
first , the density of states , @xmath43 , of the edge channel @xmath18 at the fermi level gives the change in the number of states if the electro - chemical potential is varied for fixed electrostatic potential ( i.e. fixed band bottom ) .
for non - interacting electrons this density of states is determined by the ( equilibrium ) velocity of carriers @xmath44 along the the path @xmath45 of the edge channel , where @xmath31 is now the transverse coordinate .
it is given by @xmath46 , where the integral over @xmath45 is along the entire path of the edge channel from one sample contact to the other .
for the interacting model this density of states diverges at @xmath47 since the single particle potential is flat . but the density of states is finite for any non - vanishing temperature .
a finite dos of edge channels due to a considerable suppression of screening at small temperatures is indicated by the numerical results presented by lier and gerhardts@xcite .
+ secondly , one can attribute to the arrangement of metal - like edge channels a geometrical capacitance matrix @xmath48 . for given geometry
, this matrix can in principle be derived with the help of poisson s equation .
for metallic screening this capacitance matrix is determined by the width and location of the edge channels .
it is , therefore , also a function of the magnetic field and the electro - chemical potentials applied to the contacts and the gates @xcite .
+ we finally take into account that each edge channel connects reservoirs in a directed way , due to the uni - directional velocity of the carriers .
this connection is determined by the transmission and reflection probability of the contact . in the following
, we shall always regard the just mentioned characteristics of the equilibrium state to be given .
consider the two - terminal bar of fig .
[ fig1].a under nonequilibrium conditions .
a cross section of the single - particle potential in the nonequilibrium case is shown in fig .
[ fig2].b .
a small increase of the voltage @xmath7 at contact @xmath4 , say @xmath49 , implies an electro - chemical voltage shift @xmath50 in channel @xmath18 .
if the transmission probability from the contact into the edge channel is @xmath51 , then the chemical potential shift of that edge channel is the same as that of the reservoir @xmath52 as a consequence , into the edge channel a charge is injected which is proportional to the density of states ( dos ) @xmath43 of the edge channel @xmath18 .
this added charge creates in the whole sample an electric nonequilibrium potential which shifts the band bottom .
this leads , in turn , to the injection of screening charge . the total charge @xmath53 in edge channel @xmath18 is then given by @xmath54 where @xmath55 is the quantum capacitance of edge channel @xmath18 .
the nonequilibrium electric potential @xmath56 of edge channel @xmath18 can be calculated for a given charge distribution by solving the electrostatic boundary - value problem associated with poisson s equation .
this leads to the introduction of the geometrical capacitance matrix @xmath57 of the edge - channel configuration by @xmath58 .
note , that the @xmath59 are calculated for edge channels which are _ disconnected _ from the contacts and where charge is not conserved .
but the relevant ( gauge invariant ) potentials are the electro - chemical potentials and not the electrostatic potentials of disconnected edge channels .
we define thus an electro - chemical capacitance matrix @xmath60 by @xcite @xmath61 considering charge conservation , @xmath62 , which is due to the connection to the reservoirs .
one finds then from eqs .
( [ eq5 ] ) and ( [ eq7 ] ) @xmath63 , where the relative electro - chemical capacitance @xmath64 of the two edge channels is given by @xmath65 this describes the relative geometrical capacitance @xcite , @xmath66 , in series with the quantum capacitances @xmath67 . + as an example , we consider the non - interacting case where the widths @xmath68 of the edge channels @xmath39 of length @xmath69 are very small ( i.e. @xmath70 ) . for the sake of simplicity , we assume them to be equal to each other , @xmath71 .
the distance between the edge channels is denoted by @xmath72 , and the charge is to be uniformly distributed in the edge channels . for line charges ,
the geometrical capacitance becomes @xcite @xmath73 .
the density of states @xmath67 , on the other hand , are given by @xmath74 , where we assumed very steep confinement potentials @xmath75 at the edge channels located at @xmath76 .
the electro - chemical capacitance @xmath64 can then be written as @xmath77 here @xmath78 is the ratio between the confinement field and the interaction field at the edge channel @xmath18 . to be consistent with the non - interacting case
, one must have @xmath79 .
note that @xmath64 depends on the magnetic field via @xmath80 , and via the @xmath81-dependence of the geometry of the edge - channel arrangement . + charge conservation in the sample is reflected by the sum rule @xmath82 which is a well - known property of a set of capacitors where ground is included .
one concludes that two - terminal systems are particularly simple since @xmath83-matrices satisfying eq .
( [ eq10 ] ) are characterized by a single quantity and have thus purely scalar properties .
we will see later on that equations analogous to eq .
( [ eq10 ] ) hold also for the dc - conductance and the emittance @xcite . below it
will be important that the electro - chemical capacitance matrix is symmetric and an even function of the magnetic field , i.e. @xmath84 and @xmath85 , respectively .
these properties are evident from our definition of @xmath86 .
+ to find the dc - conductance in the transmission approach the current can be evaluated in response to a small variation of the chemical potential of the contacts keeping the electro - static potential fixed at its equilibrium value .
the transmission probabilities are a functional of the equilibrium electrostatic potential only . here
we briefly discuss the derivation of the dc - conductance using the actual non - equilibrium current . for a detailed discussion of the various possible definitions of currents and
their physical interpretation we refer the reader to komiyama and hirai@xcite . in order to find the dc - conductance @xmath87 of the bar in fig .
[ fig1].a , we remark that the total nonequilibrium current through a contact consists of two contributions @xcite . at contact @xmath51 , for example , a first part @xmath88 originates from the action of the nonequilibrium electric field on the occupied equilibrium states in the landau level .
this part is obtained from a spatial integration of the current density ( [ eq3 ] ) in the region between the edge channels . a second part @xmath89 is caused by the motion of the added charge density @xmath90 with an equilibrium velocity @xmath91 in edge channel @xmath18 . in the present notation , the relation between the velocity and the dos for quasi one - dimensional conductors reads @xmath92 where the sign is different for opposite edge channels . using this and eq .
( [ eq5 ] ) gives @xmath93 .
it follows immediately that the total current @xmath94 depends only on the electro - chemical potentials of the contacts and is given by @xmath95 with a zero - frequency conductance @xmath96 .
this universal result reflects the integer quantum hall effect @xcite . using @xmath97 and @xmath98 with @xmath64 given by eq .
( [ eq8 ] ) we find for the ratio of the two currents @xmath99 .
this ratio is large for small dos @xmath67 of the edge channels , i.e. for a sufficiently steep slope of the confinement potential , when the chemical contribution to the current predominates . on the other hand ,
if the edge channels are _ macroscopic metallic conductors with complete screening , ( @xmath100 and @xmath101 ) the chemical contribution vanishes and the electrostatic contribution predominates .
_ consider now a more general quantum hall sample with @xmath102 contacts and @xmath103 edge channels .
we assume that the density of states , @xmath67 , and the electro - chemical capacitance matrix , @xmath104 ( @xmath105 ) , are known .
an expression for @xmath106 in terms of the geometrical capacitance , @xmath59 , and the dos of the edge channels , @xmath67 , is derived in an appendix .
( [ eq5 ] ) and ( [ eq7 ] ) are still valid in the present case .
each edge channel @xmath18 is connected to reservoirs @xmath3 and @xmath4 , where @xmath107 is permitted .
reservoir @xmath4 injects carriers into edge channel @xmath18 from which carriers are emitted into reservoir @xmath3 . for simplicity and to be definite , we assume that the contact resistances of this sample are quantized@xcite
. the transmission probability of a carrier in contact @xmath108 to enter edge channel @xmath18 is denoted by @xmath109 and for quantized contact resistances is given by @xmath110 similarly , we introduce the probability of a carrier which approaches contact @xmath108 on an edge channel @xmath18 to enter the contact @xmath111 from the micro - reversibility properties of the transmission probabilities we have @xmath112 .
the transmission probability of the contact plays the role of a topological factor determined by the connectivity of the edge channel to the contacts of the sample . with the help of the contact transmission probability ( [ eq16 ] )
, the variation @xmath113 of the electro - chemical voltage of edge channel @xmath18 can be expressed in terms of the voltages @xmath7 in the reservoirs : @xmath114 the charge @xmath115 in edge channel @xmath18 is then related to voltage variations in the contacts by @xmath116 the total charge in all those channels into which contact @xmath3 injects is @xmath117 .
if there is no transmission ( i.e. if all edge channels are connected to a single contact ) , one has @xmath118 . a capacitance measurement yields then a capacitance matrix @xmath119 below
, we shall see that the assumption of the absence of transmission between different contacts is crucial in order to find a magneto - capacitance according to eq .
( [ eq18 ] ) .
+ the dc - conductance @xmath120 can be calculated following the same lines as above for the two - terminal case with a single landau level .
the current @xmath121 through contact @xmath3 is obtained from a sum over all incoming and outgoing channels @xmath18 with a contribution @xmath122 and over all landau levels with a contribution @xmath123 .
the well - known result @xcite reads in our notation @xmath124 for the derivation of eq .
( [ eq19 ] ) we used that @xmath125 where @xmath126 is the number of edge channels in which contact @xmath4 injects .
the diagonal element , @xmath127 , is @xmath128 times the number of channels which leave contact @xmath4 and which do not return to this contact , and @xmath129 ( @xmath130 ) is @xmath128 times the number of directed channels going from contact @xmath4 to contact @xmath28 .
note that both the current conservation property @xmath131 and the onsager - casimir reciprocity relations , @xmath132 , are satisfied .
first , we recall the theory of the emittance for a general arrangement of mesoscopic conductors by closely following ref . @xcite .
once the electrostatic geometrical capacitance matrix is known , we can formulate our discussion in terms of a discrete set of potentials which we take to be constants along each edge channel .
a general formulation of the theory for such a discrete potential model is the subject of ref.@xcite .
it is well - known that the transmission approach to current transport relates conductances to scattering matrices of the conductors .
a scattering matrix relates incoming and outgoing current amplitudes of the contacts @xmath133 of a sample for each conduction channel @xmath134 .
the dos @xmath43 of channel @xmath18 expressed in terms of the scattering matrix can then be written as a sum of _ partial densities of states , @xmath135 .
the quantity @xmath135 is the dos of channel @xmath18 associated with carriers which are scattered from contact @xmath4 to contact @xmath3 .
a slight variation @xmath7 of the voltage in contact @xmath4 causes the injection of a total charge @xmath136 through contact @xmath3 .
thus , a slowly oscillating voltage implies an additional current @xmath137 at this contact .
now , it follows from the definition in eq .
( [ eq2 ] ) that the emittance @xmath138 can be identified with @xmath139 .
note that there are two contributions to @xmath140 . a first part which neglects screening
is given by a kinetic contribution @xmath141 this part gives the charge which is scattered from the contacts @xmath4 to contact @xmath3 due to the shift of the electro - chemical potentials @xmath7 at fixed electrostatic potentials @xmath56 .
however , the nonequilibrium electrostatic potential which is due to the nonequilibrium charge - distribution in the edge channels is still neglected in eq .
( [ eq20 ] ) . in order to take it into account , we recall that the @xmath56 are shifts of the band bottoms of the edge channels , which cause an induction of additional screening charges .
hence , there is a second contribution @xmath142 given by the part of the screening charge which is eventually scattered to contact @xmath3 .
it can be expressed in the form @xmath143 the quantity in the large bracket , @xmath144 , is a partial dos associated with carriers in channel @xmath18 emitted into contact @xmath3 irrespective of the injecting contact @xmath4 .
the change of the electric potential @xmath56 at channel @xmath18 is determined by the variations of the electro - chemical potentials of the conductors . within linear response theory
we write @xmath145 where the _ characteristic potentials
@xmath146 @xcite give the change of the electrostatic potential of conductor @xmath18 if the voltage is changed in contact @xmath4 by unity .
the sum of the two parts given by eqs .
( [ eq20 ] ) and ( [ eq21 ] ) leads finally to the emittance matrix @xmath147 : @xmath148 the occurrence of the characteristic potentials @xmath149 indicates the necessity of the knowledge of the nonequilibrium state in order get the emittance . since the characteristic potentials are sample specific , one can not expect to obtain a universal result for the ac - admittance .
furthermore , since the kinetic part and the screening part contribute with opposite signs , the emittance elements can have positive or negative sign depending on which part is dominant @xcite . _ _ to apply the result ( [ eq24 ] ) to quantized hall samples , one uses the fact that the partial dos can be expressed in terms of the transmission probabilities @xmath150 , @xmath151 , and the density of states @xmath67 of edge channel @xmath18 : @xmath152 this follows directly from the suppression of backscattering in an edge channel .
the characteristic potentials @xmath153 follow from eqs .
( [ eq5 ] ) , ( [ eq7 ] ) and ( [ eq16a ] ) : @xmath154 by inserting eqs .
( [ eq25 ] ) and ( [ eq26 ] ) in eq .
( [ eq24 ] ) one obtains @xmath155 this is the key result of our work .
the emittance is the sum of all those charges which are _ emitted at contact @xmath3 due to the injection of charge at contact @xmath4 mediated by coulomb interaction between edge channels .
the elementary process which contributes to the emittance is illustrated by the diagram in fig .
+ the emittance has the following properties . from @xmath156 and eq .
( [ eq10 ] ) one concludes that @xmath157 which is a consequence of charge neutrality@xcite . since @xmath158 ,
the onsager - casimir reciprocity relations@xcite @xmath159 based on micro - reversibility are satisfied , too .
in contrast to @xmath160 , the emittance matrix @xmath161 is in general _ not _ symmetric .
a comparison of the eqs .
( [ eq18 ] ) and ( [ eq27 ] ) implies that the emittance is a ( symmetric ) capacitance , i.e. @xmath162 , if each edge channel @xmath18 is connected to a single reservoir , i.e. if @xmath163 holds .
in this section we apply the previous results to various examples of hall devices . the electro - chemical capacitance matrix @xmath164 is always assumed to be known .
the two - terminal devices in figs .
[ fig1 ] and [ fig4].a can be characterized by the scalar admittance @xmath165 . while @xmath166 for the hall bar in fig .
[ fig1].a , the dc - conductance vanishes identically for the corbino disc in fig .
[ fig1].b since there is no dc - current flowing through the contacts . from eq .
( [ eq27 ] ) one finds the emittances @xmath167 and @xmath168 for the bar and the disc , respectively . here
, @xmath64 denotes the relative electro - chemical capacitance between the edge channels .
while the emittance of a corbino disc is an electro - chemical capacitance , the emittance of a quantum hall bar turns out to be a _ negative _ electro - chemical capacitance .
the interchange of the sign can be understood intuitively by remarking that the kinetic part @xmath169 and the part @xmath170 associated with screening are interchanged for the two different topologies .
indeed , for a voltage oscillation at contact @xmath51 of the bar , transmitted charge goes to reservoir @xmath171 and induced charge comes back via edge channel @xmath171 . in the corbino geometry , on the other hand , transmitted charge comes back to contact @xmath51 and screening charge goes to reservoir @xmath171 .
+ in order to obtain an intuition for the signs of emittances , consider fig .
[ fig4].a where a two - terminal quantum hall bar with two pairs of edge channels is shown .
a constriction is assumed to bend back the second pair ( 3 and 4 ) which will thus not contribute to the dc - conductance .
the dc - conductance is @xmath128 as for the case of the bar in fig .
[ fig1].a .
however , the second pair gives a capacitive contribution to a time dependent current . from eq .
( [ eq27 ] ) and using eq .
( [ eq10 ] ) , one finds immediately @xmath172 .
hence , the emittance is a capacitance ( i.e. @xmath173 ) if the coulomb interaction between edge channels @xmath174 and @xmath175 is stronger than the coulomb interaction between edge channels @xmath51 and @xmath171 .
the transmitting edge channels contribute thus with a negative capacitance .
it is very remarkable that only two elements of the full capacitance matrix @xmath176 determine the emittance .
this is a consequence of the quantized contact transmission - probabilities and of the symmetry and current - conservation properties of the capacitance matrix . the direct way in which our approach permits to derive this result demonstrates its usefulness . + the three - terminal device in fig .
[ fig4].b consists of a quantum hall bar with a gate on top of the 2des and close to one sample edge .
the gate is connected to a further contact and couples only capacitively to the edge channels .
this setup has been investigated in ref .
@xcite . clearly , all the @xmath177 and @xmath178 vanish .
the dc - conductance for the contacts @xmath51 and @xmath171 is equal to @xmath87 for the quantum hall bar in fig .
[ fig1].a .
the presence of the gate breaks the symmetry of the quantum hall bar where at equal time the magnetic field @xmath81 is inverted and the reservoirs @xmath51 and @xmath171 are interchanged .
one expects thus that the emittance matrix @xmath179 is an asymmetric function of the magnetic field .
equation ( [ eq27 ] ) yields @xmath180 for instance , by measuring the current at contact @xmath51 for a voltage oscillation at the gate , one finds @xmath181 for one polarity of the field @xmath81 , but @xmath182 for the other field polarity .
this follows directly from the reciprocity relations ( [ eq29 ] ) . because the capacitance @xmath183 between channels @xmath51 and @xmath174 is different from the capacitance @xmath184 between channels @xmath171 and @xmath174 , one observes a completely asymmetric emittance coefficient @xmath185 as a function of the magnetic field .
this prediction is in agreement with the experimental results reported in ref .
the symmetry of the emittance matrix strongly reflects the geometry of the edge - channel arrangement .
+ in fig .
[ fig5 ] , four - terminal samples are shown which are used in order to investigate the quantum hall effect @xcite . in such devices , two contacts serve as current source and sink , whereas the two remaining contacts are used as voltage probes . in fig .
[ fig5].a an ideal bar is shown where edge channels connect subsequent contacts along the sample edge . in the sample of fig .
[ fig5].b , on the other hand , there are certain edge channels leaving one and the same contact but connecting different contacts .
+ let us assume for the ideal four - terminal bar in fig .
[ fig5].a a filling factor between the integers @xmath186 and @xmath187 such that @xmath186 edge channels exist along each sample edge which connect contact @xmath18 with contact @xmath188 .
it is possible to define electro - chemical capacitances @xmath189 between these sets of edge channels which leave contact @xmath18 and of those which leave contact @xmath190 . for each of those sets we plotted a single directed line .
( [ eq19 ] ) and ( [ eq27 ] ) yield @xmath191 for the dc - conductance , and @xmath192 for the emittance , respectively . here
, we defined @xmath193 , and the indices @xmath194 and @xmath175 have to be identified with each other .
+ on the other hand , for the specific non - ideal hall bar plotted in fig .
[ fig5].b the connection between contacts via edge channels is not simply determined by the topology of the sample boundary . in the particular case of fig .
[ fig5].b , the dc - conductance becomes @xmath195 for the emittance coefficients one finds expressions of the form @xmath196 etc .
, where the @xmath176 denote the electro - chemical capacitances between edge channels labelled as shown in fig .
[ fig5].b .
below , we will use this example in order to discuss the effect of voltage probes .
furthermore , we will derive the frequency dependent longitudinal and hall resistances for the ideal hall sample in fig .
[ fig5].a .
in this section we study the crossover from a @xmath102-terminal sample to a @xmath197-terminal sample by using one contact , say contact @xmath198 , as a voltage probe . for the dc - conductance
, this problem has been investigated in ref.@xcite .
we assume that there is at least one edge channel which connects contact @xmath199 with a different contact .
for an ideal voltage probe , there is no possibility for charge to pass through contact @xmath198 such that @xmath200 . by eliminating @xmath201 in eq .
( [ eq1 ] ) , one obtains from @xmath202 a new admittance @xmath203 for the remaining @xmath197 contacts , where @xmath204 a brief calculation confirms that eqs .
( [ eq28 ] ) and ( [ eq29 ] ) remain valid for eqs .
( [ eq30 ] ) and ( [ eq31 ] ) .
the additional terms appearing in eqs .
( [ eq30 ] ) and ( [ eq31 ] ) describe scattering between edge channels at contact @xmath199 ( incoherent terms ) @xcite . now
, the probability of a carrier to go from contact @xmath4 to contact @xmath3 is no longer restricted to the values zero and unity .
the additional terms have the following simple interpretations .
firstly , the second term on the right hand side of eq .
( [ eq30 ] ) describes the equipartition of the current which comes from contact @xmath4 to @xmath199 between the channels which go from contact @xmath199 to @xmath3 .
secondly , the three correction terms on the right hand side of eq .
( [ eq31 ] ) can be associated with processes + 1 ) where carriers go from contact @xmath4 to contact @xmath3 bypassing @xmath199 and obtain a ` self - emittance ' contribution @xmath205 , + 2 ) where carriers induced via the emittance @xmath206 are transmitted from contact @xmath199 to contact @xmath3 , and + 3 ) where carriers which are transmitted from contact @xmath4 to contact @xmath198 interact via the emittance @xmath207 with contact @xmath3 .
+ as an example , we consider the four - terminal sample of fig .
[ fig5].b where contact @xmath174 is to serve as the voltage probe .
( [ eq30 ] ) and ( [ eq31 ] ) yield the following three - terminal dc - conductance for the contacts @xmath51 , @xmath171 , and @xmath175 : @xmath208 for example , carriers from edge channel @xmath171 will be scattered at contact @xmath174 with probability one half to channel @xmath174 and one half to channel @xmath209 , which implies @xmath210 .
similar interpretations can be found for the other elements of the dc - conductance matrix ( [ eq32 ] ) .
+ from eq .
( [ eq31 ] ) one obtains the emittance matrix @xmath211 where the @xmath212-matrix @xmath213 is obtained from the matrix @xmath179 by deleting row @xmath174 and column @xmath174 . the fact that @xmath214 holds can be easily understood from eq .
( [ eq31 ] ) : there are neither edge channels which go from contact @xmath175 to @xmath174 nor from contact @xmath174 to @xmath171 .
simple interpretations exist also for the other emittance coefficients .
for example , consider the additional term @xmath215 of @xmath216 .
a voltage oscillation in contact @xmath171 induces a current in edge channel @xmath171 which leads to contact @xmath174 .
this current is divided into _ two _ parts ( channels @xmath174 and @xmath209 ) at contact @xmath174 .
hence , a contribution @xmath217 with a factor one half occurs .
the integer quantum hall effect corresponds to the quantization of the hall resistance and the vanishing of the longitudinal resistance of the ideal four - probe quantum hall bar of fig .
[ fig5].a at zero frequency @xcite .
two of the contacts serve as voltage probes , whereas the two remaining contacts are used as source and sink for the current .
the discussion of the quantum hall effect in terms of edge channels is provided by ref .
@xcite . with the help of the theory presented in this paper , the results of these references
can now be extended to the low frequency case .
+ if the contacts @xmath174 and @xmath175 in fig .
[ fig5].a are the voltage probes , the longitudinal resistance is defined by @xmath218 . on the other hand ,
the hall resistance is defined by @xmath219 , provided the contacts @xmath171 and @xmath175 are voltage probes . with the help of eq .
( [ eq1 ] ) , @xmath220 and @xmath221 can be expressed in terms of the @xmath222 . after some linear algebra one finds @xcite @xmath223 and @xmath224 , where @xmath225 is the determinant of the @xmath212 matrix @xmath226 restricted to , say , @xmath227 . by using the results of sect .
[ examples].c , one obtains a longitudinal resistance @xmath228 , where @xmath229 with @xmath186 being the number of edge channels along an edge . with @xmath230 one obtains @xmath231 the leading term of the longitudinal resistance is determined by the coulomb coupling between the current circuit and the voltage circuit which are represented by edge channels @xmath51 and @xmath174 , respectively . on the other hand ,
the hall resistance turns out to be @xmath232 this result implies that , in contrast to the longitudinal resistance , for the hall resistance the sign of the first - order term with respect to frequency depends on the specific locations of the contacts . in principle , the capacitances @xmath233 and @xmath234 can be found independently by measuring @xmath220 for appropriate choices of current and voltage probes .
a further measurement of @xmath221 provides then a test for the validity of eqs .
( [ rlong ] ) and ( [ rhall ] ) .
finally , a direct calculation shows that the eqs .
( [ rlong ] ) and ( [ rhall ] ) satisfy the reciprocity relations @xcite @xmath235 ; in particular , @xmath236 holds .
we investigate the low frequency admittance of quantized hall samples by using a simple discrete potential model based on the decomposition of the 2des in ( well - separated ) metallic and dielectric parts and by applying a general theory of the low frequency admittance @xmath237 for mesoscopic conductors .
the main result is an expression for the emittance matrix @xmath15 in terms of electro - chemical capacitance elements which depend on the geometrical configuration and the density of states of the edge channels .
we emphasize that the theory satisfies the important requirement of charge neutrality and current conservation .
the emittance gives the charge emitted through contact @xmath3 mediated by the coulomb interaction of edge channels for a voltage - variation at contact @xmath4 .
if there is no transmission of charge between different reservoirs , the emittance is a capacitance , but in the presence of transmission the emittance can even be a negative capacitance .
this has been exemplified by comparing corbino and bar geometries .
the symmetry of the emittance matrix with respect to the magnetic field depends significantly on the geometry of the edge channels .
the presence of a voltage probe and the resulting inter edge - channel scattering at the voltage probe is investigated .
we finally derive expressions for the frequency dependent longitudinal and hall resistances of an ideal four - probe bar . due to the intuitive expression for the emittance ,
all results have simple interpretations .
+ _ acknowledgement this work was supported by the swiss national science foundation . _
in order to derive the electro - chemical capacitance matrix @xmath238 for a system with many edge channels , we assume that the spatial variation of the electrostatic potential _ inside _ the edge channels can be neglected ( discrete potential model ) . then , eq . ( [ eq5 ] ) remains valid and the definition of a geometrical capacitance @xmath48 is meaningful .
it is convenient to use vector and matrix notation .
let us write the @xmath239-matrix for the geometrical capacitance of the ( disconnected ) edge channels by @xmath240 , and in a similar way for the electro - chemical capacitance @xmath241 , and the dos @xmath242 .
we introduce @xmath103-dimensional vectors @xmath243 , @xmath244 , and @xmath245 for the charges , the electrostatic and the electro - chemical potentials of the edge channels , respectively .
the solution of the poisson equation for a given charge distribution yields an electric potential @xmath246 , where @xmath247 is a vector with all components being unity .
note that a constant potential shift @xmath248 in the whole sample is always a solution of the poisson equation and is determined by charge conservation , @xmath249 .
hence , @xmath250 which defines a matrix @xmath251 such that @xmath252
. the electro - chemical capacitance matrix @xmath253 follows from @xmath254 and can be expressed in the form @xmath255 where @xmath256 denotes the identity matrix . 9 k. von klitzing , g. dorda , and m. pepper , phys .
* 45 * , 494 ( 1980 ) .
m. bttiker , phys .
b * 38 * , 9375 ( 1988 ) . for reviews
see r. j. haug , semicond .
. technol .
* 8 * , 131 , ( 1993 ) ; and _ semiconductors and semimetals _ , vol .
35 , edited by m. reed , academic press , boston , ( 1992 ) .
m. bttiker , h. thomas , and a. pretre , phys .
a * 180 * , 364 ( 1993 ) ; z. phys .
b * 94 * , 133 ( 1994 ) .
m. bttiker , j. phys . :
condensed matter * 5 * , 9361 ( 1993 ) .
b. i. halperin , phys .
b * 25 * , 2185 ( 1982 ) .
d. b. chklovskii , b. i. shklovskii , and l. i. glazman , phys .
b * 46 * , 4026 ( 1992 ) .
n. r. cooper and j. t. chalker phys . rev .
b * 48 * , 4530 ( 1993 ) .
k. lier and r. r. gerhardts , phys .
b * 50 * , 7757 ( 1994 ) . a. l. efros , solid state commun . *
65 * , 1019 ( 1988 ) ; solid state commun . * 70 * , 253 ( 1989 ) .
solid state commun .
* 70 * , 253 ( 1989 ) a. m. chang , solid state commun . *
74 * , 871 ( 1990 ) . c. w. j. beenakker , phys
lett . * 64 * , 216 ( 1990 ) .
b. y. gelfand and b. i. halperin , phys . rev .
b * 49 * , 1862 ( 1994 ) . c. mndez and j. simonin , phys . rev .
b * 49 * , 17199 ( 1994 ) . b. j. van wees phys .
lett . * 66 * , 2033 ( 1991 ) .
b. jeanneret et al . , phys .
b * 51 * , 9752 ( 1995 )
. a. h. mcdonald , t. m. rice , and w. f. brinkman , phys .
b * 28 * , 3648 ( 1983 ) .
d. b. chklovskii , phys .
b * 51 * , 9895 ( 1995 ) . for the fractional quantized hall effect a widely discussed alternative description of edge states
is based on chiral luttinger liquids ( see x. g. wen , phys . rev . * b43 * , 11025 ( 1991 ) ) .
this approach neglects electrostatic restructuring but can be used to discuss the role of contacts ( see c. l. kane and m. p. a. fisher ( unpublished ) ) .
a. o. govorov , phys . rev . * 51 * , 14498 ( 1995 ) .
l. d. landau and e. m. lifschitz , ( pergamon press , oxford , 1980 ) .
viii h. hirai and s. komiyama , phys .
b * 49 * , 14012 ( 1994 ) . w. chen , t. p. smith iii , m. bttiker , and m. shayegan , phys .
* 73 * , 146 ( 1994 ) .
s. komiyama and h. hirai , ( unpublished ) d. j. thouless , phys .
lett . * 71 * , 1879 ( 1993 ) . v. b. shikin , jetp lett . * 61 * , 227 ( 1995 ) .
b. j. van wees et al .
, phys . rev .
* b43 * , 12431 ( 1991 ) . g. mller et al . , phys . rev . * b42 * , 7633 ( 1990 ) .
a. pretre , h. thomas , and m. bttiker , ( unpublished ) .
m. bttiker , ibm j. res
. develop .
* 32 * , 317 ( 1988 ) .
|
we present a current and charge conserving theory for the low frequency admittance of a two - dimensional electron gas connected to ideal metallic contacts and subject to a quantizing magnetic field . in the framework of the edge - channel picture
, we calculate the admittance up to first order with respect to frequency .
the transport coefficients in first order with respect to frequency , which are called _ emittances , determine the charge emitted into a contact of the sample or a gate in response to an oscillating voltage applied to a contact of the sample or a nearby gate .
the emittances depend on the potential distribution inside the sample which is established in response to the oscillation of the potential at a contact .
we show that the emittances can be related to the elements of an electro - chemical capacitance matrix which describes a ( fictitious ) geometry in which each edge channel is coupled to its own reservoir .
the particular relation of the emittance matrix to this electro - chemical capacitance matrix depends strongly on the topology of the edge channels : we show that edge channels which connect different reservoirs contribute with a negative capacitance to the emittance .
for example , while the emittance of a two - terminal corbino disc is a capacitance , the emittance of a two - terminal quantum hall bar is a negative capacitance .
the geometry of the edge - channel arrangement in a many - terminal setup is reflected by symmetry properties of the emittance matrix .
we investigate the effect of voltage probes and calculate the longitudinal and the hall resistances of an ideal four - terminal hall bar for low frequencies . _
|
in order to find a clue on the quantum gravity problem in spacetime for which @xmath0 , a very rich model of different lower dimensional gravity has been developed . in the particular case of @xmath1 , it is well known that the einstein - hilbert action has been used as the gravity sector . however ,
this model is locally trivial because the einstein - hilbert action in @xmath1 is just a topological invariant ( gauss - bonnet theorem ) .
if we want to obtain the dynamical degree of freedom , we need to couple this action with different fields besides the gravitational one . under this perspective
, the dilatonic field has shown a very rich structure and includes black hole solutions .
the dilatonic field naturally arises , for instance , in the compactifications from higher dimensional gravity or from string theory .
two dimensional dilatonic - gravity has black hole solutions that play an important role and reveal various physical aspects such as spacetime geometry , the quantization of gravity , and also the physics related to string theory @xcite .
furthermore , technical simplifications in two dimensions often lead to exact results , and it is hoped that this might help to address some of the conceptual problems posed by quantum gravity in higher dimensions .
the exact solvability of two - dimensional models of gravity have been a useful tool for research in black hole thermodynamics @xcite .
such lines of research are provided to give deeper understanding of some key issues , including the microscopic origin of black hole entropy @xcite , and the final stages of black hole evaporation @xcite . for a review of two dimensional dilaton gravity see @xcite , in the specific subject of black hole physics , these are several studies that have contributed to understanding the scattering and absorption properties of waves in black holes .
because spacetime geometry surrounding a black hole is non - trivial , the hawking radiation emitted at the event horizon is modified by this geometry , and therefore an asymptotic observer measuring the black hole thermal spectrum , will measure a modified spectrum and no longer the well known black body thermal spectrum @xcite .
the factors that modify the emitted spectrum of black holes are known as greybody factors and can be obtained through the classical scattering for fields under the influence of a black hole . because the nature of hawking radiation is quantum , the study of greybody factors allows increases the semiclassical gravity dictionary , and also gives further steps into the quantum nature of black holes , for a review about this topic see @xcite . in the present work we study
the reflection and transmission coefficients , and the greybody factors of massive charged fermions fields on the background of two - dimensional charged dilatonic black holes @xcite , @xcite .
greybody factors for scalar and fermionic field perturbations on the background of black holes have received great attention . in this context , it was shown that for all spherically symmetric black holes , the low energy cross section for massless minimally - coupled scalar fields is always the area of the horizon , where the contribution to the absorption cross section comes from the mode with lowest angular momentum @xcite . however , for asymptotically ads and lifshitz black holes , it was observed that , at the low frequency limit there is a range of modes with highest angular momentum , which contribute to the absorption cross section in addition to the mode with lowest angular momentum @xcite . also , it was observed that the absorption cross section for the three dimensional warped ads black hole is larger than the area , even if the @xmath2-wave limit is considered @xcite .
recently it has been found that the zero - angular - momentum greybody factors for non - minimally coupled scalar fields in four - dimensional schwarzschild - de sitter spacetime tends to zero around the zero - frequency limit @xcite .
otherwise , for fermionic fields , it was shown that the absorption probability for bulk massive dirac fermions in higher - dimensional schwarzschild black hole increases with the dimensionality of the spacetime and decreases as the angular momentum increases .
for this spacetime , it was also revealed that the absorption probability depended on the mass of the emitted field , that is , that the absorption probability decreases or increases depending on the range of energy when the mass of the field increases .
also , it has been observed that the absorption probability increases for higher radii of the event horizon @xcite , see for instance @xcite for the decay of dirac fields in higher dimensional black holes . for further reference ,
massive charged scalar field perturbations of the kerr - newman black hole background were studied in @xcite , the absorption of photons and fermions by black holes in four - dimensions in @xcite , the fermion absorption cross section of a schwarzschild black hole in @xcite , and charged fermionic perturbations in the reissner - nordstrom anti - de sitter black hole background in @xcite . for higher - dimensional black hole background see @xcite . furthermore , fermionic perturbations on the background of two - dimensional dilatonic black holes have been studied in which it was shown that the absorption cross section vanishes at the low and high frequency limits . however , there is a range of frequencies where the absorption cross section is not null @xcite . besides , charged fermionic field perturbations
have been studied in order to obtain the quasinormal modes and to study the stability of these black holes @xcite .
as such this paper is organized as follows . in sec .
[ fp ] , we study massive charged fermionic perturbations in the background of two - dimensional dilatonic black holes , and in sec . [ coeff ] we calculate the reflection and the transmission coefficients , and the absorption cross section . finally , our conclusions are in sec
. [ remarks ] .
let us begin with the effective action of maxwell - gravity coupled to a dilatonic field @xmath3
@xcite @xmath4where @xmath5 is the ricci scalar , @xmath6 is the central charge , and @xmath7 is the electromagnetic strength tensor . if we perform the variation of the metric , gauge , and dilaton field , we obtain the following equations of motions . @xmath8 in order to describe the black hole solution , we considered the following form of the static metric for charged black holes @xmath9 in this expression , @xmath10 , @xmath11 , and @xmath12 .
we used that @xmath13 because the asymptotic flatness condition for the spacetime require .
it is well known that @xmath14 and @xmath15 ( free parameters ) are proportional to the black hole mass and charge , respectively .
the positions of the horizons are given by @xmath16 we can obtain one single horizon solution ( @xmath17 ) if the following condition is fulfilled @xmath18 , from which it is straightforward to see that @xmath19 corresponds an extremal case , where @xmath20 . on the other hand ,
using the coordinate transformation @xmath21 yields @xmath22 , where the spatial infinity is now located at @xmath23 .
we can see , this solution represents a well - known string - theoretic black hole teo:1998kp , mcguigan:1991qp , witten:1991yr .
as it is well known , charged fermionic perturbations on the background of two - dimensional charged dilatonic black hole are governed by the dirac equation @xmath24where @xmath25 denotes the electromagnetic potential , @xmath26 and @xmath27 denote the charge and the mass of the fermionic field @xmath28 respectively , and @xmath29represents the covariant derivative @xmath30 . in this last expression
@xmath31 $ ] are the generators of the lorentz group , and @xmath32 are the gamma matrices in curved spacetime .
these are defined by @xmath33 , where @xmath34 are the gamma matrices in flat spacetime . here ,
we consider the following representation for the @xmath35 gamma matrices @xmath36where @xmath37 are the pauli matrices . now , in order to find the solution to the dirac equation in this background we use the diagonal vielbein given by @xmath38and from the null torsion condition @xmath39 , we obtain the spin connection @xmath40therefore , choosing the following ansatz for the fermionic field @xmath41we obtain the following coupled system of equations @xmath42 now , decoupling the above equations we obtain the following equation for @xmath43 @xmath44 and now performing the transformation @xmath21 , eq .
( radial ) becomes @xmath45 where @xmath46 are the roots of the function @xmath22 , which are given by @xmath47 and the constants @xmath48 , @xmath49 and @xmath50 are defined by the expressions : @xmath51 additionally , we perform the change of variable @xmath52 , and making the substitution @xmath53 in eq .
( [ diff ] ) , we obtain the following equation for @xmath54 @xmath55 where @xmath56 @xmath57 therefore , as eq .
( [ he ] ) corresponds to the hypergeometric equation , its solution is given by @xmath58 which has three regular singular points at @xmath59 , @xmath60 and @xmath61 . here
, @xmath62 denotes the gauss hypergeometric function and @xmath63 , @xmath64 are integration constants and @xmath65 @xmath66 @xmath67 now , imposing boundary conditions at the horizon , i.e. , that there is only ingoing waves , and choosing @xmath68 implies that @xmath69 .
thus , the solution for @xmath70reduces to @xmath71 otherwise , in order to find the solution for @xmath72 , we use the change of variable defined before , i.e. , @xmath52 .
thus , the second equation of the system ( [ system ] ) , can be written as @xmath73 now , by using the integrating factor @xmath74 given by : @xmath75 we integrate eq . ( [ psi2 ] ) and we obtain the solution @xmath76 which can be written as @xmath77 by using the relation @xmath78
the reflection and transmission coefficients depend on the behavior of the radial function , at the horizon and at the asymptotic infinity , and are defined by @xmath79 where @xmath80 is the flux , and is given by @xmath81 where , @xmath82 , @xmath83 , @xmath84 , and @xmath85 , which yields @xmath86 the behavior of the fermionic field @xmath28 at the horizon is given by eq .
( [ rhorizon ] ) for @xmath87 and eq .
( [ psi2horizon ] ) for @xmath88 in the limit @xmath89 .
then , using eq .
( [ flux ] ) , we get the flux at the horizon @xmath90 besides , in order to obtain the asymptotic behavior of @xmath87 and @xmath88 we use the kummer s formula @xcite : @xmath91 in eq .
( [ rhorizon ] ) and eq .
( [ psi2horizon ] ) , and by using eq .
( [ flux ] ) we obtain the flux at the asymptotic region @xmath92 @xmath93 where , @xmath94 where in the last two equations we have used the property @xmath95 .
therefore , the reflection and transmission coefficients are given by @xmath96 @xmath97 and the absorption cross section @xmath98 , reads @xmath99 now , we perform a numerical analysis of the reflection coefficient ( [ coef1 ] ) , transmission coefficient ( [ coef2 ] ) , and absorption cross section ( [ absorptioncrosssection ] ) of two - dimensional charged dilatonic black holes , for charged fermionic fields . in fig . ( [ coefficients ] ) we show the behavior of the reflection and transmission coefficients and the absorption cross section , for charged fermionic fields for @xmath100 , @xmath101 , @xmath102 , @xmath103 , and @xmath104 .
essentially , we found that the reflection coefficient is 1 at the low frequency limit , that is @xmath105 , whereas for the high frequency limit this coefficient is null , the opposite behavior of the transmission coefficient , with @xmath106 . also , we observe that the absorption cross section is null at the low and high - frequency limits , but there is a range of frequencies for which the absorption cross section is not null , and also has a maximum value .
( solid curve ) , the transmission coefficient @xmath107 ( dashed curve ) , @xmath108 ( thick curve ) , and the absorption cross section @xmath98 ( dotted curve ) as a function of @xmath109 , @xmath110 ; for @xmath100 , @xmath101 , @xmath102 , @xmath103 , and @xmath104.,scaledwidth=45.0% ] in addition , in fig .
( [ aq ] ) we show the behavior of the absorption cross section for different ( positive and negative ) values of @xmath15 , where we observe that the absorption cross section is null at the low and high - frequency limit , but there is a range of frequencies for which the absorption cross section is not null , and also has a maximum value .
also , we observe in figs .
( [ aq ] ) , ( [ aqq ] ) and ( [ aq1 ] ) that for @xmath111 the absorption cross section decreases when @xmath112 increases , due to the electric repulsion .
however , for @xmath113 we found that the absorption cross section does not depend on the value of @xmath112 . also , we observe in fig .
( [ amp ] ) that the absorption cross section increases if the mass of the fermionic field increases ; however , beyond a certain value of the frequency , the absorption cross section is constant and null for the high - frequency limit . on the other hand , in fig .
( [ am ] ) , we plot the absorption cross section for different values of @xmath14 and we observe that it does not depend on the mass of the black hole . finally , we observe that the absorption cross section increase if @xmath114 decrease , see fig .
( [ aqm ] ) .
( dotted curve ) as a function of @xmath109 , @xmath110 ; for @xmath115 , @xmath116 , @xmath103 , @xmath104 , and @xmath117 for the left figure , and @xmath118 for the right figure.,title="fig:",scaledwidth=45.0% ] ( dotted curve ) as a function of @xmath109 , @xmath110 ; for @xmath115 , @xmath116 , @xmath103 , @xmath104 , and @xmath117 for the left figure , and @xmath118 for the right figure.,title="fig:",scaledwidth=45.0% ] ( dotted curve ) as a function of @xmath109 , @xmath110 ; for @xmath116 , @xmath103 , @xmath104 , @xmath119 and @xmath120 for the left figure , and @xmath121 for the right figure.,title="fig:",scaledwidth=45.0% ] ( dotted curve ) as a function of @xmath109 , @xmath110 ; for @xmath116 , @xmath103 , @xmath104 , @xmath119 and @xmath120 for the left figure , and @xmath121 for the right figure.,title="fig:",scaledwidth=45.0% ] ( dotted curve ) as a function of @xmath109 , @xmath110 ; for @xmath122 , @xmath116 , @xmath103 , @xmath104 and @xmath123 for the left figure , and @xmath124 for the right figure.,title="fig:",scaledwidth=45.0% ] ( dotted curve ) as a function of @xmath109 , @xmath110 ; for @xmath122 , @xmath116 , @xmath103 , @xmath104 and @xmath123 for the left figure , and @xmath124 for the right figure.,title="fig:",scaledwidth=45.0% ] ( dotted curve ) as a function of @xmath109 ; for @xmath100 , @xmath101 , @xmath102 , @xmath104 , and @xmath125.,scaledwidth=45.0% ] ( dotted curve ) as a function of @xmath109 ; for @xmath100 , @xmath101 , @xmath103 , @xmath104 , and @xmath126.,scaledwidth=45.0% ] ( dotted curve ) as a function of @xmath109 , @xmath110 ; for @xmath100 , @xmath101 , @xmath102 , @xmath103 , and @xmath127.,scaledwidth=45.0% ]
in this work we have studied massive charged fermionic perturbations on the background of two - dimensional charged dilatonic black holes , and we have computed the reflection and transmission coefficients , and the absorption cross section , and we have shown numerically that the absorption cross section vanishes at the low and high frequency limits . therefore , a wave emitted from the horizon , with low or high frequency , does not reach infinity and is totally reflected , since the fraction of particles penetrating the potential barrier vanishes ; however , we have shown that there is a range of frequencies where the absorption cross section is not null .
the reflection coefficient is 1 at the low frequency limit and null for the high frequency limit , demonstrating behavior opposite of the transmission coefficient , with @xmath106 .
it is worth mentioning that these results , greybody factors , are consistent with other geometries of dilatonic black holes @xcite .
also , we have studied the effect of the electric charge of the fermionic field over the absorption cross section , and we have observed different behaviors depending on the sign and the value of the product of the charges @xmath128 . that is , for @xmath111 we have found that the absorption cross section decreases when @xmath112 increases , due to the electric repulsion .
however , for @xmath113 we have found that the absorption cross section does not depend on the value of @xmath112 , and for this case we obtain the same value of the absorption cross section as for the case @xmath129 .
also , we have found that the absorption cross section increases if the mass of the fermionic field increases ; however , beyond a certain value of the frequency , the absorption cross section is constant . also , we have found that the absorption cross section for massive charged fermionic fields in a charged two - dimensional dilatonic black hole does not depend on the mass of the black hole .
this work was funded by comisin nacional de ciencias y tecnologa through fondecyt grants 11140674 ( pag ) , 1110076 ( js ) and 11121148 ( yv ) and by di - pucv grant 123713 ( js ) .
r. b. acknowledges the hospitality of the universidad diego portales where part of this work was undertaken .
s. r. das , g. w. gibbons and s. d. mathur , phys .
* 78 * ( 1997 ) 417 . c. campuzano , p. gonzalez , e. rojas and j. saavedra , jhep * 1006 * ( 2010 ) 103 .
p. gonzalez , e. papantonopoulos and j. saavedra , jhep * 1008 * ( 2010 ) 050 .
p. a. gonzalez and j. saavedra , int .
j. mod .
a * 26 * ( 2011 ) 3997 .
p. a. gonzalez , f. moncada and y. vasquez , eur .
j. c * 72 * ( 2012 ) 2255 .
j. j. oh and w. kim , eur .
j. c * 65 * ( 2010 ) 275 .
l. s c. b. crispino , a. higuchi , e. s. oliveira and j. v. rocha , phys .
d * 87 * ( 2013 ) 10 , 104034 [ arxiv:1304.0467 [ gr - qc ] ] .
m. rogatko and a. szyplowska , phys .
d * 79 * ( 2009 ) 104005 [ arxiv:0904.4544 [ hep - th ] ] .
r. moderski and m. rogatko , phys .
d * 77 * ( 2008 ) 124007 [ arxiv:0805.0665 [ hep - th ] ] . g. w. gibbons , m. rogatko and a. szyplowska , phys .
d * 77 * ( 2008 ) 064024 [ arxiv:0802.3259 [ hep - th ] ] .
r. a. konoplya and a. zhidenko , phys .
d * 88 * ( 2013 ) 024054 [ arxiv:1307.1812 [ gr - qc ] ] .
s. s. gubser , phys .
d * 56 * ( 1997 ) 7854 [ hep - th/9706100 ] . c. doran , a. lasenby , s. dolan and i. hinder , phys .
d * 71 * ( 2005 ) 124020 [ gr - qc/0503019 ] .
r. g. cai , z. y. nie , b. wang and h. q. zhang , arxiv:1005.1233 [ gr - qc ] .
e. jung , s. kim and d. k. park , jhep * 0409 * ( 2004 ) 005 [ hep - th/0406117 ] .
|
we study massive charged fermionic perturbations in the background of a charged two - dimensional dilatonic black hole , and we solve the dirac equation analytically .
then , we compute the reflection and transmission coefficients and the absorption cross section for massive charged fermionic fields , and we show that the absorption cross section vanishes at the low and high frequency limits . however , there is a range of frequencies where the absorption cross section is not null .
furthermore , we study the effect of the mass and electric charge of the fermionic field over the absorption cross section .
|
heavy elements ( or metals ) in the universe are created in stars and supernovae ( sne ) .
these metals enrich the interstellar medium and are recycled into generations of stars .
some part of these should have been transported into intergalactic space via galactic winds or ram pressure stripping .
indeed , the intracluster medium ( icm ) contains an amount of metals comparable to the total amount of metals found in galaxies ( e.g. * ? ? ?
the distribution of these metals can be measured exclusively by x - ray spectroscopy .
this constrains the total number of sne in galaxies and the efficiency of the mass transport from galaxies into intergalactic space .
moreover , the abundance ratios among elements ( e.g. si / fe ratio ) can be used to identify their origins ( e.g. * ? ? ?
_ asca _ and _ beppo - sax _
observations provided the first systematic measurements of si , s and fe abundances and their spatial distributions ( e.g. * ? ? ? * ) .
_ xmm - newton _ has allowed more detailed studies including the o abundance determination ( e.g. * ? ? ? * ) .
recent _ suzaku _ data have further improved measurements of elements such as o and mg ( e.g. * ? ? ? * ) . mostly , observed abundance patterns are explained by some combinations of origins from type ia and type ii sne ( e.g. * ? ? ?
note that the nature of these sne and hence their total metal products have not completely been understood .
in addition , some authors claim that the observed metal abundances require an alternative source such as hypernovae in early generations of stars @xcite .
however , measurements are still limited to abundant elements and even for those elements statistical and systematic errors are large .
see @xcite for a recent review on x - ray spectroscopy of clusters . here
we present _ suzaku _ xis ( x - ray imaging spectrometer ) spectroscopy of the central region of the perseus cluster of galaxies , which is the brightest cluster in x - rays .
this is a calibration target of the xis and hence has been observed regularly .
we used all available data from several exposures .
the deep exposure of this unique object along with the high sensitivity of the xis provides one of the best quality x - ray spectra from clusters .
we have detected x - ray line emission from ionized cr and mn for the first time from an extragalactic source .
furthermore we accurately measured elemental abundances of ne , mg , si , s , ar , ca , fe , and ni . throughout this paper , we assume cosmological parameters as follows ; @xmath2 km s@xmath3mpc@xmath3 , @xmath4 , and @xmath5 . at the cluster redshift of 0.0183
, one arc - minute corresponds to 22.2 kpc at the cluster distance .
we use the 90% confidence level .
as a calibration target , the perseus cluster has been observed twice a year in various operation modes .
here we use the data obtained from february 2006 to february 2009 in the normal window mode .
sequence numbers of the data are 800010010 , 101012010 , 101012020 , 102011010 , 102012010 , 103004010 , 103005010 , and 103004020 .
the central galaxy , ngc 1275 , has been located at the ccd center .
the detector covers a central @xmath6 square region .
the xis 2 sensor was available only until 2006 .
the cumulative exposure time in the spaced - row charge injection _ off _ mode is 300 ks times sensors ( xis 0 , xis 2 , and xis 3 ; fi ) .
those in the spaced - row charge injection _ on _ mode are 580 ks times sensors ( fi ) and 265 ks ( xis 1 ; bi ) .
detailed descriptions of the _ suzaku _ observatory , the xis instrument , and the x - ray telescope are found in @xcite , @xcite , @xcite , respectively .
we started the analysis from archived cleaned event files , which were filtered with the standard selection criteria .
we used the latest calibration file as of may 2009 .
we examined light curves excluding the central bright region events ( @xmath7 ) for stable - background periods .
there was no flaring event in all data .
the instrumental ( non - x - ray ) background was estimated using the night earth observation database and the software xisnxbgen @xcite .
the cosmic x - ray background was estimated to be well below 1% of the source over almost entire enegy band .
we therefore ignore this cosmic background in the following analysis .
we extracted spectra from three annulus regions with boundary radii of @xmath8 , @xmath9 , @xmath10 , and @xmath0 centered on the x - ray maximum .
the central region ( @xmath11 ) is excluded to avoid the contribution from the central galaxy .
we prepared an x - ray telescope response function for each spectrum using the xis arf builder software xissimarfgen @xcite , assuming an icm surface brightness distribution based on an _ xmm - newton _ public image of the cluster as a source image .
we simulated events so that the relative error of the response at each energy bin is smaller than 0.25% ( using a parameter accuracy ) . to made the xis energy response function , we used the software xisrmfgen . to describe the thermal emission from a collisional ionization equilibrium plasma
, we use the apec model @xcite with the solar metal abundances taken from @xcite .
the galactic absorption is described by the photo - electric absorption of wisconsin cross - sections ( wabs model in xspec ) with the reported neutral hydrogen column density of @xmath12@xmath13 toward the cluster .
we fit the data to constrain the thermal nature and metal abundances . to describe the complex thermal structure of the central cooling core , projection effect , and photon mixing due to the point spread function , we introduce two thermal components .
temperatures and iron abundances of the two components are made independently free .
on the contrary , other abundances of ne , mg , si , s , ar , ca , and ni are coupled between the two components , e.g. ne@xmath14 @xmath15 ne@xmath16 . a power - law component with a fixed photon index of 1.8
is also included .
this describes a combination of scattering of the central galaxy emission and the possible distributed hard x - ray emission suggested by @xcite . in some spectra
we found line emission at the position of he - like cr . because there is no line from cr included in the apec model , here ,
we add a gaussian component for the cr line into the fitting model .
a detailed analysis of this line is given in the next subsection . to avoid calibration errors , we ignore energy ranges around the si k edge ( @xmath17 kev ) and the au m edge ( @xmath18 kev ) . to compensate errors of the energy - pi relation , redshifts of the two apec components
are allowed to be different . in @xmath19 regions ,
we use the fi and bi spectra simultaneously . in the @xmath20 region , we use only the fi spectrum , since the bi spectrum has a relatively low signal to background ratio at higher energies , @xmath21 kev . the fitting result is given in table [ tbl:2t - fit ] .
statistical errors of elemental abundances are @xmath22% of the best fit value . obtained abundances are shown in fig .
[ fig : abun](a ) . here
we use the fe abundance averaged over those of the two components by weighting by the emission measure .
examples of the spectral fitting are shown in fig .
[ fig:2t - fit ] .
there are two types of systematic residuals .
the one is those found around strong lines , such as the fe k line at @xmath23 6.5 kev in the @xmath24 spectral fit .
limited accuracies in the energy response could cause these discrepancies .
furthermore , uncertainties in the modeling of the fe k line structure could be larger than the statistics of the current data.% .
a similar level of model uncertainties is expected . ]
the other residual can be seen in the energy range of @xmath25 kev only in the @xmath26 fit .
this could be related to instrumental background .
at this energy band , the background is dominated by line emission such as mn k@xmath27 ( 5.9 kev ) , mn k@xmath28 ( 6.5 kev ) , and ni k@xmath27 ( 7.5 kev ) .
time variations in line fluxes and changes in the line spread function create errors on the background estimation @xcite .
nevertheless , each model describes the data with residuals smaller than five percent in most energy bins .
llllllllll @xmath24 & fi+bi & 1246/471 & @xmath29 & @xmath30 & @xmath31 & @xmath32 & @xmath33 & @xmath34 & @xmath35 + @xmath36 & fi+bi & 1207/471 & @xmath37 & @xmath38 & @xmath39 & @xmath40 & @xmath41 & @xmath42 & @xmath43 + @xmath20 & fi & 485/239 & @xmath44 & @xmath45 & @xmath46 & @xmath47 & @xmath48 & @xmath49 & @xmath50 + to estimate systematic errors associated with the instrumental calibration , we fit the fi and bi spectra separately and compare the results with those obtained from the combined ( fi and bi ) fits above . in these fits , statistical errors on abundances are about @xmath22% and @xmath51% for the fi and bi spectra , respectively . as shown in fig .
[ fig : abun](b ) relative differences between the separate and combined fits are within 20% mostly and within 30% in all cases . by this comparison and the current calibration status of the xis
, we suppose that the possible systematic errors on the abundances are about @xmath52% of the best - fit values . in the core of the perseus cluster , resonant scattering may redistribute some fe line photons spatially from the core to the outer region .
we examined this effect by following @xcite .
the resonant scattering should be relatively significant on the resonance transition of he - like fe ( 1s@xmath53 - 1s2p ) at 6.70 kev .
then , we exclude this line energy band ( @xmath54 kev ) and fit the data again .
this fit gives consistent abundances of fe and ni with the previous one .
this indicates that the scattering is insignificant in our data . using _
xmm - newton _ data , @xcite and @xcite independently reported consistent results and suggested that the scattering effect is reduced because of gas motions in the core . as mentioned above , we detected x - ray line emission from cr .
next to cr , we expect line emission from mn from a sub - solar metallicity plasma .
note that the cr and mn solar abundances are about 1% and 0.5% of that of fe , respectively . to search for these lines
, we use the fi spectra integrated over a @xmath55 region in spaced - row charge injection _ on _ and _ off _ modes together . as shown in fig .
[ fig : cr](a ) , by comparing the data with a bremsstrahlung continuum model , we have detected a line from he - like cr clearly and that from he - like mn .
note that in the model used in this fit the si - escape line associated with the he - like fe k@xmath27 line is included .
this si - escape line has a peak at 4.8 kev with a flux of about 1% of the data at that position . using the same data above and gaussian line fitting , we measured these line positions and fluxes as given in table [ tbl : cr - mn ] .
the detection of the cr line is more than 99.9% confidence level , while that of the mn line is more than 98% level .
the fluxes are measured with respect to that of the he - like fe k@xmath27 line at 6.70 kev .
the resonance transitions of he - like cr and mn are at 5.682 kev and 6.181 kev , respectively @xcite .
the obtained line positions are fully consistent with those of redshifted resonance positions .
the equivalent widths of the cr and mn lines are about 5 ev and 2 ev , respectively . in a similar way as above
, we examined spectra from the three regions separately .
the significances of the cr line detections are @xmath56% , @xmath56% , and 94% confidence from the @xmath57 spectra , respectively .
the mn line detection is significant ( @xmath58% confidence ) only from the @xmath24 spectrum . in table
[ tbl : cr - mn ] , we show the obtained line positions and fluxes .
to check if these features are instrumental , we examine the crab nebula and instrumental background data obtained with the xis .
the crab has a bright power - law x - ray emission and is a standard calibration source for x - ray instruments . in the crab spectra , there is no structure at the line positions of cr and mn detected above .
we found no structure at the same positions in the instrumental background .
there are lines from neutral mn k@xmath59 ( 5.895 kev ) and k@xmath60 ( 6.490 kev ) .
however , these position are away from those of he - like cr or mn .
moreover , we confirmed that the _ chandra _ deep spectrum ( @xmath611 ms exposure ) shows a line feature at the same cr position with a flux consistent with the _ suzaku _ data .
these facts , redshifted line positions , and the obtained flux as expected from a sub - solar metallicity plasma ( as calculated in discussion ) indicate that we detected the cr and mn line emission from the cluster . in our knowledge , this is the first significant detection of x - ray line emission of cr and mn ions from an extragalactic source .
@xcite reported the detection of the same cr line from the cluster 2a 0335 + 096 at the @xmath62 level .
llllllll @xmath55 & @xmath63 & @xmath64 & 6070@xmath65 & @xmath66 & & & + @xmath24 & @xmath67 & @xmath68 & 6070 & @xmath69 & @xmath70 & @xmath71 & 4.0 + @xmath72 & @xmath73 & @xmath74 & 6070 & @xmath75 & @xmath64 & @xmath76 & 4.7 + @xmath20 & @xmath77 & @xmath78 & 6070 & @xmath79 & @xmath80 & @xmath81 & 5.6 +
_ suzaku _ xis observations of the perseus cluster provided one of the best quality x - ray spectra from clusters .
we have detected the he - like cr and mn lines .
furthermore , we measured the radial distribution of abundances of ne , mg , si , s , ar , ca , fe , and ni . there is no strong radial variation in relative abundance ratios such as mg / fe and si / fe . the abundance ratios are within a range of @xmath1 times the solar value .
based on observed line fluxes of cr and mn ions , we estimate abundances of these elements . there is no atomic data in the atomdb @xcite nor mekal @xcite providing x - ray emissivities of these lines . then
, following @xcite , we use line emissivities of elements ( si , s , ar , ca , fe , and ni ) available from the atomdb and estimate those of cr and mn . here
we assume that emissivities per single ion of these elements at a given temperature is a smooth function of the atomic number .
we interpolate the function using spline fits , as shown in fig .
[ fig : cr](b ) .
we used temperatures ( table [ tbl : cr - mn ] ) calculated by averaging over the two thermal components in the best - fit model obtained in subsection 3.2 .
errors introduced by the interpolation would be smaller than @xmath82% .
this estimation along with observed fluxes gives cr and mn abundances ( table [ tbl : cr - mn ] ) : cr / fe and mn / fe ratios are @xmath83 and @xmath84 times the solar , respectively .
the above estimated values obtained for cr and mn along with the fe and ni abundances derived in the spectral fitting provide iron - group abundances for the first time in the icm .
we found cr : mn : fe : ni ratios close to the solar value .
these abundance patterns depend on the nature of their origins , supernovae ( sne ) .
for example , @xcite propose to reveal the detailed nature of type ia sn using mn , cr , and fe x - ray lines in sn remnants . in the case of the icm metals
, we can examine sne averaged over many events in member galaxies .
the present result supports the idea that the metals in the icm and solar neighborhood were produced in a similar way .
first , the _ chandra _ result in the perseus cluster @xcite is compared .
we used their abundance pattern obtained from a region ( 40 - 80 kpc ) similar to our data , but accounted for projection .
the ne / fe , si / fe , ar / fe , ca / fe , and ni / fe ratios are consistent with the _ suzaku _ result .
in contrast , the _ chandra _ data shows lower mg / fe ( @xmath85 ) and higher s / fe ( @xmath86 ) values than the solar ratio found in the current study .
we suppose that these inconsistencies are caused partly by calibration errors used in @xcite and partly caused by the difference in the spectral extraction method .
indeed , we found that much longer _
chandra _ data ( @xmath23 1 ms exposure ) with the latest calibration and without the projection correction give mg and s abundances consistent with our result .
second , we compared our result with measurements in other clusters [ fig .
[ fig : abun](c ) ] .
note that including our study , most results are derived from cluster central regions ( within a few 100 kpc ) . in most elements , our results are consistent with those in other clusters .
exceptions are lower ar and higher ca abundances in average values over a sample of 22 clusters reported by @xcite .
we can not find this trend in the perseus cluster .
for all measured elements from ne to fe , the cluster abundances are consistent with the solar ratios within uncertainties .
the ni / fe ratios are relatively high .
this may be a sign of a contribution from type ia sne ( e.g. * ? ? ?
the sub - solar value of o / fe measured in other studies may have a similar origin ( e.g. * ? ? ?
third , our measurement is compared with that from galactic stars with extremely low metallicity [ fig .
[ fig : abun](c ) ; @xcite ] .
the latter abundances should be close to those originated from the first generation of stars and sne in the universe .
the patterns are very different . among elements that we measured , elements lighter than cr are more abundant in the metal poor stars , while cr , mn , and ni are more abundant in the icm .
this comparison , regardless of theoretical models , indicates that the icm is polluted not solely by the first generations of stars but also by additional source(s ) significantly .
the additional source(s ) should add iron group elements to the icm .
the most - likely dominant contributor is type ia sne , as in the galactic chemical evolution model ( e.g. * ? ? ?
the abundance pattern in the perseus cluster is closer to the solar value than to those in extremely low metallicity stars .
therefore the contribution of the first generations of stars into the icm metal is not required largely , in contrast to the model proposed in @xcite . finally ,
the measured pattern is compared with theoretical predictions from two types of sne [ fig [ fig : abun](d ) ] .
the observed abundance pattern can not be produced solely by each type of sne .
all the observed abundance ratios are between the two predictions , except for cr and mn , which have relatively large uncertainties .
consequently , we conclude that the metals in the icm of the perseus and other clusters were produced by a mix of both types of sne .
similar conclusions have been obtained from _ asca _ ,
_ chandra _ , and _ xmm - newton _ studies ( e.g * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the present data strengthen this scenario .
anders , e. , & grevesse , n. 1989 , geochimica et cosmochimica acta , 53 , 197 badenes , c. , bravo , e. , and hughese , j.p .
2009 , , 680 , l33 baumgartner , w. h. , loewenstein , m. , horner , d. j. , & mushotzky , r. f. 2005 , , 620 , 680 bhringer , h. , & werner , n. 2009 , arxiv:0907.4277 cayrel , r. , et al .
2004 , , 416 , 1117 churazov , e. , forman , w. , jones , c. , sunyaev , r. , bhringer , h. 2004 , , 347 , 29 dupke , r. a. , & white , r. e. , iii 2000 , , 528 , 139 fukazawa , y. , makishima , k. , tamura , t. , ezawa , h. , xu , h. , ikebe , y. , kikuchi , k. , & ohashi , t. 1998 , , 50 , 187 gastaldello , f. & molendi , s. 2004 , , 600 , 670 hwang , u. , petre , r. , & hughes , j.p .
, 2000 , , 532 , 970 ishisaki , y. , et al .
2007 , , 59 , s113 iwamoto , k. , brachwitz , f. , nomoto , k. , kishimoto , n. , umeda , h. , hix , w. r. , & thielemann , f .- k . 1999 , , 125 , 439 kaastra , j.s . 1992 , an x - ray spectral code for optically thin plasmas ( internal sron - leiden report , updated version 2.0 ) kobayashi , c. , umeda , h. , nomoto , k. , tominaga , n. , & ohkubo , t. 2006 , , 653 , 1145 koyama , k , et al .
2007 , , 59 , s23 loewenstein , m. 2001 , , 557 , 573 matsushita , k. , finoguenov , a. , bhringer , h. 2003 , , 401 , 443 matsushita , k. et al .
2007 , , 59 , 327 mitsuda , k , et al . 2007 , , 59 , s1 mushotzky , r. , loewenstein , m. , arnaud , k. a. , tamura , t. , fukazawa , y. , matsushita , k. , kikuchi , k. , & hatsukade , i. 1996 , , 466 , 686 de plaa , j. , werner , n. , bleeker , j. a. m. , vink , j. , kaastra , j. s. , & mndez , m. 2007 , , 465 , 345 renzini , a. , ciotti , l. , dercole , a. & pellegrini , s. , , 1993 , 419 sato , k. et al .
2007 , , 59 , 299 sanders , j. s. , fabian , a. c. , allen , s. w. , & schmidt , r. w. 2004 , , 349 , 952 sanders , j. s. , & fabian , a. c. 2006 , , 371 , 1483 serlemitsos , p. et al .
2007 , , 59 , s9 smith , r.k & brickhouse , n.s .
2001 , , 556 , l91 tawa , n. , et al .
2008 , , 60 , 11 tamura , t. , bleeker , j. a. m. , kaastra , j. s. , ferrigno , c. , & molendi , s. 2001 , , 379 , 107 tamura , t. , kaastra , j. s. , den herder , j. w. a. , bleeker , j. a. m. , & peterson , j. r. 2004 , , 420 , 135 tsuru , t. 1991 , phd thesis , university of tokyo werner , n. , et al .
2006 , , 449 , 475
|
the results from _ suzaku _ observations of the central region of the perseus cluster are presented .
deep exposures with the x - ray imaging spectrometer provide high quality x - ray spectra from the intracluster medium .
x - ray lines from helium - like cr and mn have been detected significantly for the first time in clusters .
in addition , elemental abundances of ne , mg , si , s , ar , ca , fe , and ni are accurately measured within @xmath0 ( or 220 kpc ) from the cluster center .
the relative abundance ratios are found to be within a range of @xmath1 times the solar value .
these abundance ratios are compared with previous measurements , those in extremely metal - poor stars in the galaxy , and theoretical models .
|
questions about the nature of chemical reactions , why and how they proceed and how this can be used to form certain desired chemical products are already very old .
in fact they are older than most other of the current research topics in atomic and molecular physics or physical chemistry .
the efforts to answer these questions have lead to numerous achievements over the centuries , starting maybe with the re - discovery of the discreet atomic structure of matter , leading to the invention of chemical catalysis , and including the understanding of the quantum mechanical nature of the chemical bond .
several technological advances have fertilised experimental research on the dynamics of chemical reactions . besides the development of versatile tunable laser source , one notes supersonic single and crossed molecular beams @xcite , multi - dimensional momentum imaging and coincidence detection @xcite , ultrafast time - resolved spectroscopy @xcite , and , most recently
, the preparation of cold and ultracold atoms and molecules @xcite .
today the study of the reaction dynamics of molecules has advanced to precise quantum - state resolved scattering experiments in quantitative agreement with high - level scattering calculations at least for reactive complexes that involve no more than four atoms .
if more atoms contribute to a chemical reaction the dimensionality of the scattering process , i. e. the number of degrees of freedom , increases beyond what quantum scattering calculations can do on current computers . also experiments are challenged by the growing complexity , because individual quantum states are increasingly difficult to separate in the initial as well as the final state of a scattering event .
however , it is precisely this `` complexity limit '' in chemical dynamics that is driving a lot of research , because one would like to understand the details of reactions that are of relevance in organic chemistry , in living cells , or in the earth s atmosphere .
in such many - atom reactions new phenomena may occur that lead to different reaction mechanisms than in triatomic reactions . within the large field of reaction dynamics
an important division follows the electric charge of the collision partners that participate in the reaction . besides the neutral - neutral reaction processes ,
the second largest class are ion - molecule reactions . as will be shown later , the much stronger ion - neutral interaction than the neutral - neutral van - der - waals interaction leads to qualitatively different behaviour .
ion - molecule reactions have often very high rate coefficients and are thus important as soon as ions are present in an environment .
these environments include discharge plasmas @xcite , combustion processes , the earth s ionosphere @xcite , and many other stellar and interstellar regions @xcite .
also in the condensed phase , more specifically in liquid solutions , ion - molecule reactions are important .
reactions of ions are consequently studied almost as long as neutral - neutral reactions .
they require , however , significantly different experimental techniques .
a large portion of the research on ion - molecule reactions is devoted to positively charged ions . in dilute plasmas , ionisation by electron impact or by radiation will directly lead to the formation of cations and they should therefore play an important role there . nevertheless , also negatively charged ions are important constituents of gas phase environments , for example in the lower ionosphere of the earth . here
the negative charge density is mostly represented by atomic and molecular anions ( such as o@xmath0 , o@xmath1 , oh@xmath0 , hco@xmath2 , no@xmath1 and their clusters with water molecules ) and not by free electrons @xcite .
similarly negative ions are omnipresent in liquid solutions and have a strong impact on reactions there .
furthermore , negative ions are precursors for solvated electrons in liquids , which are responsible for strong optical absorption and have been the subject of many studies . in recent years
negative ions have also been detected in other planetary atmospheres in our solar system , such as the atmosphere of saturn s satellite titan , and they have been found in the interstellar medium , in the dark molecular cloud tmc-1 and in a circumstellar shell around the carbon rich star irc+10216 @xcite .
the understanding of the complex chemistry that leads to anions in these areas is still fragmentary .
the questions here are not only related to the formation of the anions , but also if they can add new formation pathways to the complex neutral molecules found in the interstellar medium , such as long unsaturated carbon chains or possibly polyaromatic hydrocarbon molecules .
the study of the reaction dynamics of anion - molecule reactions are the topic of this review .
comparisons with cation - molecule reactions will be made if suitable .
chemical reactions of negative ions have been studied extensively for many decades .
several excellent reviews have appeared a few years ago on this topic @xcite . here
, we will concentrate specifically on the reaction dynamics in contrast to the mere kinetics of anion - molecule reactions . ``
molecular reaction dynamics is the study of elementary processes [ of molecular collisions ] and the means of probing them , understanding them and controlling them . ''
@xcite . in this sense
we focus on processes where isolated molecules collide in a well - controlled reactive event with a negative ion .
we give an overview over all the major experimental techniques that are used to study anion - molecule reaction dynamics .
in particular , we include the recent experimental developments , on the one hand cryogenic ion traps and on the other hand ion - molecule crossed - beam imaging , and show how they provide new insights .
negative ions are also subject to spectroscopic studies in the infrared and optical range , using bound - bound absorption and bound - free photoelectron spectroscopy .
this reveals the stability , i. e. the electron affinity of the neutral , as well as structural information about the anion . since negative ions are still today challenging for quantum chemical calculations , due to the spatial extension of the wavefunction of the excess electron and its interaction with the atomic core , spectroscopy and detachment studies can provide benchmark data @xcite .
photoelectron spectroscopy of negative ions can also reveal the energy level structure of the neutral system that is reached after photodetaching the excess electron .
if the neutral system is not a bound molecule , but a reactive complex , such as for the fh@xmath1 anion , one can even study neutral reactions near the transition state between reactants and products @xcite .
for more information on the spectroscopy of negative ions see refs .
@xcite .
this article will start with the discussion of a few general considerations of anion - molecule collisions and reactions in the next section .
then it will be discussed how measurements of integral cross sections or total rate coefficients , respectively , are performed .
they are carried out either at a given collision energy or a given sample temperature and can already be used to infer information on the underlying reaction dynamics . in the next section , experimental approaches are discussed that allow the investigation of the transient reaction complex . then in section [ crossed - beams : sect ] experiments with crossed beams are presented , which allow the determination of energy- and angle - differential cross sections , under optimum conditions even quantum - state resolved . from these measurements a wealth of information about the collision and reaction dynamics can be extracted . in the final section some of the future directions for the study of anion - molecule reaction dynamics are highlighted .
the understanding of the dynamics of an anion - molecule reaction starts with the description of the interaction potential , usually in the born - oppenheimer approximation .
the long - range and short - range properties of the interaction potential are discussed in the next sections .
the proper theoretical treatment of the dynamics of the colliding molecules and ions on the interaction potential is a quantum scattering calculation @xcite , which is usually , however , limited to a reduced set of dimensions by the computational effort .
alternatively , the atomic motion is treated with classical dynamics @xcite , which is generally a good assumption because of the small de - broglie wavelength of the atoms . in this case
all dimensions can be included , but such effects as the zero - point motion , tunnelling through potential barriers or quantum scattering resonances can usually not be accounted for . at large distances the force between an ion and a neutral molecule is determined by the electrostatic interaction potential of the ion with the induced , and possibly also permanent , electric dipole moment of the molecule .
@xmath4 is the orientationally averaged polarisability and @xmath5 the permanent dipole moment of the neutral molecule .
@xmath6 is the charge of the colliding ion . in certain cases also
the ion - quadrupole interaction and other higher order multipole terms need to be added . if the molecule carries a permanent dipole moment its interaction with the charge , which scales as @xmath7 typically dominates over the induced - dipole interaction , which scales as @xmath8
. however , at large relative separation the free rotation of the molecule leads to a vanishing time - averaged permanent dipole moment . at the largest inter - atomic distances
the ion - induced dipole interaction is therefore most important , which is always attractive .
once the interaction potential then becomes comparable to the rotational energy `` locking '' of the permanent molecular dipole to the incoming ion may occur and will thereby enhance the attractive interaction . at short distances the electronic wavefunctions of the ion and the molecule
determining the short range interaction potential therefore requires solving the schrdinger equation for the collision system .
this is often particularly difficult for negative ions , because the excess electron is so delocalized that a large basis set is required in the numerical solution of the schrdinger equation .
however , it works as an advantage that many negative ions of interest are `` closed - shell '' systems with a spin singlet configuration of the electronic wavefunction , because these are the most stable negative ions with respect to electron detachment . as a consequence ,
only a single born - oppenheimer potential hypersurface governs the reactions of such negative ions with closed - shell neutral molecules . for many negative ion - neutral collision systems
a repulsive potential is obtained at very short range .
this may be attributed to the electron - electron repulsion by electrostatic forces and the pauli exclusion principle .
together with the strong attractive interaction this leads to potential energy landscapes with two characteristic minima along the reaction coordinate , as shown in fig .
[ reaction - pot : fig ] .
the path along the reaction coordinate is defined as the path of minimum potential energy that connects the reactants with the products .
the potential energy minima typically lie several hundred mev below the asymptotic energies .
consequently , substantial short - range energy barriers may have negative transition state energies , as shown in fig.[reaction - pot : fig ] .
but even if the transition state energy is negative , it can have a profound influence on the reactivity as we will see in the next paragraph .
this is in contrasts to most reactions of cations , because there is usually no or at least no substantial short - range barrier .
the kinetics of a chemical reaction , which describes the rate at which products are formed as a function of time , is usually characterised by the reaction rate coefficient . for a bimolecular reaction that occurs in a collision of two species
the second order rate equation @xmath9 describes the time - dependent increase of the product particle density @xmath10 for given reactant densities @xmath11 .
this equation is typically applied to thermal ensembles .
the second order rate coefficient @xmath12 is then dependent on the absolute temperature @xmath13 .
it is related to the scattering cross section @xmath14 for the individual collision events , which depends on the relative velocity , by the thermal average @xmath15 @xmath16 is the thermal probability distribution for the relative velocities at the absolute temperature @xmath13 .
the standard model to estimate collision rates of ions with neutral atoms or molecules is the langevin or capture model @xcite .
the assumption of this model is that a collision occurs with 100% probability if the two collision partners come closer to each other than a critical distance .
only the longest - range attractive interaction , the ion - induced dipole potential ( first term in eq .
[ long - range - pot : eq ] ) , and the repulsive centrifugal potential are taken into account .
this critical distance is then given by the location of the maximum of the centrifugal barrier .
the largest impact parameter for a scattering trajectory that reaches this critical distance and can surmount the centrifugal barrier determines the scattering cross section . for larger impact parameters
the ion - neutral interaction is neglected . with an additional thermal averaging ,
this assumption yields the langevin rate coefficient ( in si units ) @xmath17 which turns out to be temperature independent .
@xmath4 is again the orientation - averaged polarisability of the neutral molecule while @xmath18 denotes the reduced mass of the two - body system ( @xmath6 is the ion charge and @xmath19 the electric constant ) .
typical langevin rate coefficients range between @xmath20@xmath21/s and @xmath22@xmath21/s .
a correction to the langevin rate coefficient has to be introduced if the neutral target carries a permanent dipole moment @xmath5 .
then the second term in eq .
( [ long - range - pot : eq ] ) needs to be included .
the `` average dipole orientation '' approach @xcite introduced an effective `` dipole locking constant '' from which an improved collision rate coefficient was derived . in a more accurate approach
, the rate coefficient for ion - polar molecule collisions is calculated from a series of classical trajectories that are calculated to numerical precision for the exact interaction potential .
the parameterisation of these trajectories leads to a pre - factor @xmath23 to the langevin rate constant eq .
[ langevin : eq ] , which is a function of @xmath5 and @xmath4 @xcite . as an effect
the ion - polar molecule capture rate coefficient is typically a factor of two to four larger than the prediction for the langevin rate coefficient .
the langevin rate constant has proven to be extremely valuable as a guideline and often describes fairly accurately measured reaction rates for cations with nonpolar neutral molecules @xcite . when the observed rate coefficient for an ion - molecule reaction is smaller than the capture rate coefficient , which occurs when the reaction probability after crossing the centrifugal barrier is smaller than 100% , temperature- or energy - dependent rate coefficient measurements can be used to infer information on the reaction dynamics at short internuclear separation . in the next sections
this approach will be illustrated with several studies from recent years .
most major experimental techniques that are used for these studies , drift and flow tubes , guided ion beams , free jet expansions and low - temperature ion traps will be discussed .
ion cyclotron resonance experiments and high - pressure mass spectrometry ( see e.g.@xcite ) , however , will not be covered in this article .
flow and drift tubes have been the classic workhorse for the acquisition of ion - molecule reaction rate coefficients .
they have produced a wealth of atmospherically and astrophysically relevant data over the almost 50 years of their existence @xcite .
today they are found in many different variants e.g.@xcite .
these instruments usually operate at room temperature , but may also be heated up to more than 1000k or cooled down to near liquid nitrogen temperatures ( 77k ) .
ions are created typically by electron impact , eventually chemically transformed and mass selected , and then injected into a constant flow of buffer gas of several m / s velocity .
the pressure of this buffer gas , which is usually helium , is of the order of a few millibar .
consequently , the mean free path of the ions is smaller than the dimension of the tube and the flowing buffer gas transports the thermalised ions downstream through the flow tube .
a distance away from the ion source , neutral reactant gas is injected .
due to reactions of the ions with the neutral gas their density decreases with flow time , which now corresponds to the flow distance .
at the same time , product ions are generated . by sampling ion yield and composition with a quadrupole mass spectrometer as a function of flow distance
absolute reaction rate coefficients can be extracted .
their temperature dependence may be obtained by heating or cooling the buffer gas inlet and flow section .
an electric field gradient may be applied along the flow section turning the flow tube into a drift tube .
this allows studies at elevated kinetic energy rather than temperature .
the rate coefficient gives a first hint at the dynamics of an ion - molecule reaction .
barrier - less proton transfer reactions usually exhibit large rate coefficients in accordance with the capture model @xcite .
in contrast many anion - molecule reactions feature orders of magnitude lower rate coefficients , pointing at dynamical bottlenecks and intermediate barriers on the potential energy hypersurface .
chemically versatile flow and drift tubes are well suited to study the effects of chemical substitution on the rate coefficient @xcite .
different reaction mechanisms leading to the same ( ionic ) products may be distinguished by studying kinetic isotope effects @xcite .
more direct insight into the dynamics can be obtained from a temperature dependent flow tube measurement of the rate coefficient @xcite . moreover , in drift tubes internal and translational degrees of freedom of the reaction partners are decoupled and can be separately controlled .
this allows to investigate how the reaction rate changes if the same amount of energy is provided in different forms , challenging a statistical model of the reaction dynamics @xcite .
probably the most important type of anion - molecule reaction that has been studied up to now is the nucleophilic substitution ( s@xmath242 ) reaction @xcite : @xmath25 where x and y can anything from simple halogen atoms to large macro - molecular systems .
the assumption of statistical energy redistribution was a matter of fierce debate in the 1990s in the framework of s@xmath242 reactions .
a variable temperature flow tube study had shown that the cl@xmath0 + ch@xmath26br reaction rate is strongly dependent on the relative translational energy , while being at the same time insensitive to the internal temperature of the reactants @xcite .
such behaviour is in clear contradiction with the statistical assumption of rapid randomisation of all available energy in the [ cl @xmath27 ch@xmath26br]@xmath0 entrance channel complex .
this observation prompted a large number of theoretical studies , which uncovered a dynamical bottleneck for energy transfer between internal modes of ch@xmath26br and the intermolecular low - frequency modes @xcite .
in a recent application of an ion flow tube instrument , the competition between nucleophilic substitution ( s@xmath242 ) and base - induced elimination ( e2 ) was investigated via chemical substitution and the deuterium kinetic isotope effect @xcite .
villano et al .
studied the reaction of bro@xmath0 and clo@xmath0 with methyl chloride ( ch@xmath26cl ) and its partially and fully methylated form cxyzcl ( where x , y , z can be either h or ch@xmath26 ) .
s@xmath242 and e2 mechanisms lead to the same ionic reaction product , a challenge for all experiments , which rely on charged particle detection .
the authors used that deuteration of the neutral reactant changes the rate of the reaction .
the deuterium kinetic isotope effect is defined as the ratio of perprotio to perdeuterio rate coefficient ( kie=@xmath28 ) .
whereas for ch@xmath26cl an inverse kie was determined ( i.e. kie@xmath291 ) , the kie was found to become increasingly more normal ( kie@xmath301 ) as the extend of methyl - substitution in the neutral reactant is increased . for the reaction of bro@xmath0 with the fully methylated neutral species , an about a factor of three larger reaction rate coefficient was measured for ( ch@xmath26)@xmath26ccl as compared to its deuterated form ( cd@xmath26)@xmath26ccl .
villano et al
. referred to a marked effect of deuteration on the vibrational dynamics near the respective transition states of the s@xmath242 and the e2 mechanism @xcite .
they argued that the e2 pathway is a minor channel for the small neutral reactant ch@xmath26cl , whereas it becomes gradually more important with increasing methylation of the neutral , which sterically hinders and finally impedes nucleophilic substitution .
the oxidation of the trichlorooxyphosphorus anion pocl@xmath2 , which occurs in combustion flames , has been recently studied by kerkines _
despite its very low rate coefficient of only around 1x10@xmath31 @xmath21/s ( at 300k ) , such oxidations can change the chemistry of flames due to the high abundance of o@xmath32 .
the authors employed a turbulent ion flow tube , where orders of magnitude higher neutral gas densities can be applied as compared to the conventional laminar flow tubes @xcite .
they measured the rate coefficient over the range of 300 - 626k and found it to increase slowly with increasing temperature .
an arrhenius fit yielded an activation energy of 50mev . since the oxidation is exothermic by about 1.8ev , the presence of an intermediate potential barrier was concluded .
examination of the reaction pathways at different levels of molecular orbital theory led to the proposal of a multistep reaction mechanism .
it involves product formation via the transformation of the entrance channel ion - dipole complex into a four - membered p @xmath27 o - o @xmath27 cl ring transition state , the highest point on the potential energy surface .
the highest temperature studies of ion - molecule reactions under fully thermalised conditions have been undertaken in a high temperature flowing afterglow apparatus at up to 1800k @xcite .
for an anion - molecule reaction , 1440k has been reached for co@xmath2 + so@xmath32 @xmath33 so@xmath2 + co@xmath32 @xcite .
the main driving force for these technically challenging experiments is to model the chemical environment of the earth s ionosphere and other planetary atmospheres at high altitude .
however , these studies are also interesting from the reaction dynamics point of view . at these high temperatures
even small molecules carry significant amounts of rotational and vibrational excitation .
it can be explored how energy supplied in different forms - translational or internal - affects the reaction rate . for the reaction of co@xmath2 with so@xmath32 , miller et al . found by comparison with drift tube data that the total energy alone controls the reactivity @xcite .
they concluded that the independence of the rate coefficient on the form of energy implies that the reaction is governed by long lived intermediates in which energy equilibrates , even at the very high temperatures .
guided ion beam ( gib ) studies are the method of choice for precise measurements of integral reaction cross sections of ion - molecule reactions for collision energies of the millielectronvolt to the tens of electronvolt range .
these measurements also allow for the determination of the opening and competition of different reactive channels in ion - molecule collisions @xcite .
teloy and gerlich introduced gibs into gas - phase chemistry in pioneering experiments in freiburg in the 1970s @xcite . in this technique
mass selected ions are passed into a long radio frequency ( rf ) multipole ion guide - typically an octupole - with a selected kinetic energy , controlled by the dc potential difference between the ion source and the ion guide .
importantly , the guiding multipole electric rf field contains the ions radially , while at the same time minimising alteration of their kinetic energy .
the latter is crucial to the technique and impedes the use of quadrupole rf fields : micro motion of the ions driven by the oscillating field results in energetic collisions with buffer and reaction gas .
these effects are referred to as radio frequency heating and are minimised in a multipole rf field @xcite .
the created effective potential guides the ions through a scattering cell located at the centre of the long guide .
the neutral reaction partner is introduced into the scattering cell - which may be temperature - variable @xcite - at a well characterised density .
ions collide with the neutrals as they pass through the scattering cell and undergo chemical reactions .
note that the multipole guide also contains the ionic reaction products .
all ions are mass analysed when they reach the end of the guide , typically with a quadrupole mass selector .
counting the number of reactant and product ions allows one to determine absolute integral cross sections for ion - molecule reactions and collision induced dissociation .
these measurements can be done as a function of the relative kinetic energy of the reactants over an extended range of energies from about 0.1 to hundreds of ev .
gib measurements achieve high resolution since they ensure single collision conditions in contrast to drift tubes and employ better defined , narrow ion velocity distributions @xcite . at the same time
the range of relative collision energies is extended since ions are radially contained and can not get lost by drifting to the walls of the tube . to further minimise the energy spread of the reactants , sophisticated ion sources such as flow tubes and multipole traps
are used @xcite .
the bulk of gib studies have focused on positively charged ions .
negative ion chemistry has been investigated in particular by the group of kent ervin , with special emphasis on nucleophilic substitution .
to uncover reaction dynamics and mechanisms , the group heavily employs ab initio calculations and density functional theory in the interpretation of their measurements .
the high resolution of gib measurements allows to reveal the threshold behaviour of endoergic reactions @xcite and of exothermic reactions , which feature an intermediate potential barrier @xcite .
the wide tunability of the relative collision energy enables studies of the opening and interplay of reaction mechanisms leading to different ionic products .
absolute integral cross sections of competing nucleophilic substitution and abstraction reactions were obtained @xcite .
collision induced dissociation of anion - dipole complexes accesses the dynamics at the transition state @xcite . in an advanced mode of operation
, gib experiments may be run in a pulsed manner and the dwell time of the ionic reaction products in the long guide may be used to analyse their axial kinetic energy distribution .
this allows to some degree to distinguish between forward and backward scattering for a more direct insight into ion - molecule reaction dynamics @xcite .
haufler , schlemmer and gerlich carried out a pioneering gib study for anions , investigating the fundamental molecular reaction h@xmath0 + d@xmath32 @xmath33 d@xmath0 + hd and its isotopic variant d@xmath0 + h@xmath32 @xmath33 h@xmath0 + hd @xcite .
integral cross sections have been obtained as a function of collision energy between 0.1 and 10ev translational energy in the centre - of - mass frame .
they show an onset at around 0.3ev , a maximum at around 1ev and a decrease at larger collision energies .
barriers of 350 and 330mev were deduced for h@xmath0 + d@xmath32 and d@xmath0 + h@xmath32 , respectively .
the decrease of the cross section was attributed to the competition with collisional electron detachment of the reactant anion .
interestingly , it was found that the integral cross section of the heavy ion colliding with the light molecule is larger by a factor of two , reaching 2.7@xmath34 at its maximum .
axial product time - of - flight distributions have been measured by recording arrival times and converted to differential cross sections d@xmath35/d@xmath36
( @xmath36 denotes the velocity component of the products along the guide ) .
preferred forward scattering was found for low collision energies just above threshold . by variation of the depth of the effective potential via the rf amplitude , which probes the transversal velocity distribution , the authors concluded that internal excitation of the product molecule is not significant at these energies .
thus they were able to deconvolute d@xmath35/d@xmath36 to obtain angle - differential cross sections d@xmath35/d@xmath37 .
these could be cross - calibrated against crossed - beam data @xcite ( see section [ conv - crossed - beams : sect ] ) .
the angle differential cross sections were found to be quite similar for the two variants of the reaction , despite the big isotope effect observed in the integral cross section .
the authors concluded that the region of the potential energy surface probed in the exit channel has to be similar and the increased reactivity of d@xmath0 + h@xmath32 has to stem from dynamics in the entrance channel .
deturi _ et al .
_ investigated the symmetric s@xmath242 reaction cl@xmath0 + ch@xmath26cl @xmath33 clch@xmath26 + cl@xmath0 @xcite by isotopic labelling .
for this system , a translational energy threshold of 2ev had been determined 10 years before in a pioneering and well - recognised drift tube measurement by the bierbaum group @xcite .
based on that finding , anionic attack on the chlorine side of chloromethane ( `` frontside attack '' ) had been proposed as the reaction mechanism ( potential barrier 2.0ev ) , and subsequently been backed by theoretical studies @xcite . in contrast , deturi _ et al .
_ determined a translational energy threshold of 470@xmath38160mev in their gib experiment .
the authors argued that the previous measurement had been affected by the skewed reactant kinetic energy distribution in the drift tube as well as by collisions with its high density helium buffer gas during the lifetime of the entrance channel complex .
the adjusted threshold , however , energetically excludes frontside attack as the reaction mechanism . on the other hand the conventional s@xmath242 back - side attack mechanism , with inversion of the carbon centre , features a potential barrier height of only around 120mev , substantially lower than the measured threshold by deturi et al .
this puzzle prompted a lot of studies on the theoretical side .
quasiclassical trajectory simulations by the hase group both on an analytical potential energy surface @xcite and with the ab initio direct dynamics method @xcite as well as reduced - dimensionality quantum dynamical calculations @xcite were undertaken .
it was confirmed that frontside attack is not the reaction mechanism and backside attack prevails .
however , in contrast to the conventional s@xmath242 mechanism , it was found that only a very small fraction of the reactive trajectories are indirect , with transient trapping in the ion - dipole wells .
this is assigned to poor coupling of the provided translational excitation to internal degrees of molecular freedom .
a direct pathway of product formation , on the other hand , seems to feature strict dynamical constraints for low collision energy such as a very restricted geometry of approach for passing over the minimum energy barrier .
the power of the gib technique is nicely demonstrated in complementary cross section studies of the exothermic s@xmath242 reaction f@xmath0 + ch@xmath26cl and its counterpart , the endothermic reverse reaction
cl@xmath0 + ch@xmath26f @xcite .
the data from angel and ervin are reproduced in fig .
[ angelervin]a and b. as can been seen , the measured reaction cross sections range over an impressive dynamic range from more than 10@xmath39@xmath34 down to 5@xmath4010@xmath41@xmath34 .
the exothermic s@xmath422 reaction leading to cl@xmath0 product formation is most efficient at the lowest centre of mass collision energies , while its cross section decreases rapidly over the range 0.1 - 2ev .
the reason for this behaviour derives from the s@xmath242 reaction mechanism .
a more impulsive collision allows for less alignment of the reaction partners along the backside attack coordinate by their ion - dipole interaction , i.e. along the near collinear reaction path .
[ angelervin]b shows that the endothermic s@xmath242 reaction leading to f@xmath0 product formation features an energy threshold as expected .
angel and ervin determined a threshold energy of 1.88ev , which is 0.54ev in excess of the reaction endothermicity .
+ ch@xmath26cl ( a ) and the inverse reaction cl@xmath0 + ch@xmath26f ( b ) for different product ions as a function of the relative collision energy .
the calculated capture cross section is shown as solid line .
reprinted with permission from @xcite and @xcite .
copyright ( 2001 ) , american chemical society . ] inspecting the pseudo - collinear pes , the authors interpreted their finding in terms of the polanyi rules @xcite .
an early transition state in the exothermic direction for the collinear reaction pathway renders translational energy efficient in promoting the reaction .
this accounts for the high cross section in excess of 100@xmath34 for low collision energies .
after translational passage over the barrier trajectories need to pass the `` bend '' along the reaction coordinate to form products . hence much of the reactant translation is expected to be converted into excitation of the c - cl stretch vibration of the product molecule .
microscopic reversibility suggests in turn that vibrational excitation of ch@xmath26cl is needed in the endothermic direction to efficiently pass the initial tight bend on the pes before surmounting the late transition state .
translational activation such as in the gib experiment should hence be inefficient at promoting the endothermic reaction .
this is in accordance with the finding of an excess barrier by angel and ervin and a low cross section of maximal 0.6@xmath34 .
fig.[angelervin ] shows that at higher collision energies other reaction channels open up and compete with nucleophilic substitution . for the exothermic direction ( fig .
[ angelervin]a ) , both proton transfer and chlorine abstraction reaction were observed . at collision energies above 2ev ,
the cl@xmath0 cross section was found to increase again .
dissociation of the ch@xmath32cl@xmath0 and fcl@xmath0 reaction products was proposed as an explanation . for the endothermic direction ( fig .
[ angelervin]b ) , methylene and fluorine abstraction as well as a subsequent threebody dissociation were observed .
some of these reactions exhibit dual rising features .
reaction mechanisms have been proposed based on molecular structure calculations .
while gib techniques are strong at determining integral reaction cross sections , the very detailed and insightful study by angel and ervin also clearly shows their limitations .
an experimental method is needed , which can provide more direct insight into the reaction dynamics and is capable of dissecting the partitioning of energy into translational and internal degrees of molecular freedom ( see section [ crossed - beams : sect ] ) .
as first demonstrated by gerlich @xcite and already mentioned above , gib methods can in fact provide to some degree more direct insight into the reaction dynamics by recording the arrival time of the product ions .
angel and ervin applied this technique in a study of the exothermic s@xmath242 reaction cl@xmath0 + ch@xmath26br @xcite .
they reported the relative product velocity distribution along the axis of the ion guide for a series of relative collision energies between 0.1 and 4.0ev . at 0.1ev ,
the s@xmath422 product velocity was found to be symmetrically distributed around the centre of mass velocity and to agree well with the prediction of a statistical phase space theory ( pst ) model .
this can be regarded as strong sign that the reaction is complex - mediated . at 0.25ev ,
the velocity distribution becomes asymmetric with a preference for forward scattering of the ch@xmath26cl product .
this shows that the reaction becomes more direct , with reactant interaction on a time scale less than the rotational period of the complex . at 0.5ev ,
backward scattering arises and becomes dominant above 1.0ev .
the authors interpreted this as introduction of a direct rebound mechanism . at lower and lower temperatures
the dynamics of ion - molecule reactions become more sensitive to details of the interaction potential and to quantum dynamical effects such as tunnelling .
changes in the rate coefficients at low temperature also affect the abundance of ions in atmospheric and interstellar plasmas , which are often characterised by temperatures well - below room temperature .
this has raised substantial interest in low - temperature studies of ion - molecule reactions .
very low temperatures , down to about 0.1k , are reached in free jet supersonic flows employing adiabatic expansion @xcite .
however , the complex cooling process that is induced by the transient flow dynamics of an expanding free jet does lead to temperature disequilibria for the different molecular degrees of freedom and non - thermal rotational state distributions @xcite . despite these difficulties supersonic jets
have been used to study a few reactions of cations , produced by electron impact or laser ionisation .
two examples are the study of vibrational quenching of no@xmath43(@xmath44=1 ) via helium by complex formation below 3kelvin @xcite and the reopening of the bimolecular c@xmath32h@xmath45 channel in the hydrogen transfer of h@xmath32 to c@xmath32h@xmath46 for low temperatures , which is attributed to tunnelling of a collision complex through the reaction barrier @xcite .
an extension of this technique to negative ion reactions is not straight forward and has not been carried out up to now .
uniform expansions from precisely designed laval nozzles can overcome some of the difficulties of free jet flows by maintaining parallel stream lines at constant mach number @xcite .
constant high densities are reached at a fixed temperature in the travelling frame over the entire flow , which makes equilibration of the molecules degrees of freedom more likely . by terminating the supersonic expansion through parallelisation of the stream lines in the nozzle ,
the exceedingly low temperatures of free expansions are unfortunately lost .
this cresu ( a french acronym for cintique de raction en ecoulement supersonique ) technique is applicable to negatively charged ions , created by dissociative electron attachment in the isentropic flow .
cresu was employed by le garrec _ et al . _ to study the s@xmath242 reaction cl@xmath0 + ch@xmath26br from 180 down to 23k @xcite . combining their results with previous measurements obtained from other experimental techniques ,
the authors demonstrated a dramatic increase in the rate coefficient by a factor of 400 upon reduction of the temperature from 500 to 23k .
a quantum scattering method , the `` rotating bond approximation '' has been employed in this work . it was found that at the lowest temperatures the reaction follows the potential energy profile with no activation energy present .
the excitation of intermolecular bending modes of the transition state then induces an activation barrier for increasing temperature . with the exploration of radio - frequency heating in guided ion beams
, it became clear that for the efficient cooling and complete thermalisation of molecular ions with a buffer gas , trapping in a quasi field - free environment is indispensable @xcite .
gerlich pioneered the development of cryogenic storage devices based on electric multipole fields , which are capable of cooling molecular ions to a few kelvin in all degrees of molecular freedom .
the most popular design is the 22pole ion trap @xcite , which is nowadays used in a wide range of applications @xcite . in such a temperature - variable 22-pole ion trap ,
the lowest - temperature for an anion - molecule reaction up to now has been achieved in a recent experiment in our group @xcite .
the specific system under investigation was the proton transfer reaction to the negatively charged amide ion @xmath47 for this reaction a room temperature reaction probability , given by the ratio of the measured rate coefficient to the calculated langevin rate coefficient , of about 2% had been found many years ago @xcite .
thus , in 98% of the collision events the reactants are reproduced and no products are being formed .
this occurs despite the fact that the reaction is exothermic and also the intermediate potential barrier lies below the energy of the entrance channel .
it shows that for this reaction the reaction dynamics at short range are very important . with h@xmath32 ( eq . [ nh2:reaction ] ) as a function of temperature @xcite .
the data show that the probability for reaction increases with decreasing temperature but stays well below the langevin limit .
below 20k the data show an unexpected decrease of the reaction probability that can not be explained by a classical statistical model ( dashed line ) . ] at lower temperatures the probability for reaction ( [ nh2:reaction ] ) increases strongly @xcite .
the data are shown in the right panel of fig .
[ reactions : fig ] . at 20k
the probability has increased by a factor of six .
this increase is a manifestation of the complex - mediated reaction dynamics @xcite : the intermediate nh@xmath48 complex , which is transiently formed during a collision that surmounts the centrifugal barrier , has a longer lifetime with respect to decay back to reactants at lower temperatures , because the number of available decay channels decreases .
the probability to cross the intermediate potential barrier and form products , however , remains approximately constant .
therefore the overall probability to react increases .
the observed decrease of the reaction probability for temperatures lower than 20k ( see fig . [
reactions : fig ] ) can not be explained within the classical dynamics picture of a complex - mediated reaction mechanism .
instead it is expected to represent a signature of quantum mechanical reaction dynamics in low temperature ion - molecule reactions .
in the following section we discuss experiments that investigate the transient complex that the reactants form along the path to reaction products .
experimentally , this is either achieved by analysing ternary , complex - forming reactions , by studying unimolecular decay of anion - molecule compounds or by applying direct time - resolved spectroscopy , as detailed below .
average lifetimes of transient entrance channel complexes formed upon collision of the reaction partners can be determined indirectly from ternary rate coefficients for their formation and collisional stabilisation . for reactions of cations with neutral molecules ,
the extraction of complex lifetimes from ternary association has a long history using flow and drift tubes at high pressure . for reactions of anions
, work has focused on the symmetric s@xmath242 reaction cl@xmath0 + ch@xmath26cl , specifically the formation and unimolecular decay time of the metastable [ cl@xmath49ch@xmath26cl]@xmath50 entrance channel complex @xcite .
ternary association is studied in a thermal environment , where the density of the neutral reaction partner and a buffer gas is well defined and controllable .
it can be understood as a two - step process . @xmath51^ * } & \rightarrow & \mathrm{[cl^-\cdots ch_3cl]}\\ & k_{\gamma } & & & \\
\end{array } \label{threebody}\ ] ] in the first step collisions of the reactants cl@xmath0 and ch@xmath26cl form the metastable ion - dipole complex under investigation , where the relative kinetic energy is transiently transferred into internal energy .
the complex is highly excited with respect to its ground state and decays back to reactants with the unimolecular rate k@xmath52 .
stabilisation of the complex can occur in a second step if a third particle impact removes more internal excitation from the complex than the initial relative translational and internal energy of its constituents .
this is only a small fraction of its internal excitation given the binding energy of ground state [ cl@xmath49ch@xmath26cl ] of about 450mev with respect to the asymptote . in subsequent collisions the internal energy is thermalised with the environment .
re - excitation of the complex above the asymptote for dissociation is then strongly suppressed by a boltzmann factor .
conditions are arranged such that the stabilisation step occurs with a high probability by the buffer gas , which is provided in large excess as compared to the reaction partners . at the same time
the low - pressure limit of the steady state approximation has to be valid , which practically means that it has to be ensured that stabilisation of the complex is a rare event .
then the rates for the association and stabilisation step in ( [ threebody ] ) are well defined and the overall rate of [ cl@xmath49ch@xmath26cl ] formation is r = k@xmath26 n@xmath53n , where k@xmath26 is the ternary rate coefficient k@xmath26 = k@xmath54k@xmath55/k@xmath52 . the ternary rate coefficient is experimentally determined by measuring r as a function of the densities of the reactant ( n@xmath53 ) and the buffer gas ( n ) .
the rate coefficients k@xmath54 and k@xmath55 describe barrier - free ion - molecule reactions ; the method relies on the assumption that these rate coefficients are very close to the capture - limited values . with this , the unimolecular dissociation rate can be extracted , the inverse of which is the average lifetime of the excited collision complex [ cl@xmath49ch@xmath26cl]@xmath50 .
li _ et al .
_ studied ternary association with high pressure mass spectrometry and obtained an average lifetime of 12 to 16ps for [ cl@xmath49ch@xmath26cl]@xmath50 at a fixed temperature of 296k @xcite ( see also @xcite ) . in a recent experiment in our group ,
a 22-pole radiofrequency ion trap has been employed , which provides the advantage of temperature - variability and long interaction times @xcite .
this allows to study the complex lifetime as a function of reactant collision temperature down to 150k , at which point the vapour pressure of the neutral reactant becomes too low to be bearable .
we have found a strong inverse temperature dependence of the average lifetime of the transient [ cl@xmath49ch@xmath26cl]@xmath50 complex in disagreement with a simple statistical model .
longer lifetimes of the entrance channel complex at lower temperatures give s@xmath242 reaction systems more time to exploit the available phase space , randomise energy , cross the central reaction barrier and finally form reaction products .
this substantiates further that enhanced lifetimes of the transient entrance channel complex are likely to underlie the strong inverse temperature dependence of rate coefficients for exothermic s@xmath242 reactions with a submerged barrier .
unimolecular dissociation of the metastable reactant complex into products @xmath56^ * \rightarrow { \rm y}^- + { \rm ch}_3{\rm\,x } \label{unimolecdiss}\ ] ] was studied in a series of experiments on asymmetric s@xmath242 reactions in the bowers group @xcite .
since entrance channel complexes were prepared , the formation of products involves at least one crossing of the intermediate reaction barrier .
relative kinetic energy distributions for the released products were recorded by means of ion kinetic energy spectroscopy .
the measurements were compared to statistical phase space theory .
somewhat controversial in these decomposition experiments is the initial state of the metastable reactant complexes under investigation .
they have to have sufficient lifetime to survive the transport from the high pressure ion source to the field - free high vacuum region of the employed mass spectrometer and still undergo unimolecular dissociation .
this renders partial collisional stabilisation in the ion source likely ( see also @xcite ) .
the kinetic energy release distributions for the investigated s@xmath242 reactions all peak at zero relative kinetic energy of the products and rapidly decrease for increasing energy .
interestingly , despite the long complex lifetimes the experimental distributions are much narrower than calculated ones based on phase space theory .
this means that less energy is partitioned to relative translation of the dissociating products than predicted for a statistical redistribution of energy .
the phase space calculation was brought into agreement with the experimental results only if a significant amount of energy was made unavailable for energy redistribution .
this effect remains if partial collisional stabilisation is taken into account .
the energy missing in translation has to be trapped in internal degrees of freedom of the reaction products .
the authors argued that significant rotational excitation is not present and that hence the neutral reaction product has to be vibrationally hot .
surprisingly , the same result was found for `` simple '' s@xmath242 reactions of methylhalides @xcite as well as for s@xmath242 reactions involving reaction partners with many more internal degrees of freedom @xcite . while for the methylhalide reactions this observation is in agreement with classical trajectory studies @xcite
, it challenges the notion that statistical theories become valid for more complex s@xmath242 reactions with extended complex lifetimes , whereas non - statistical dynamics is restricted to small or highly energised systems with short complex lifetimes on the order of tens of picoseconds @xcite . for extended complex lifetimes , it was speculated that tunnelling through the central barrier might be an alternative way of product formation , leading to a reduced effective barrier height @xcite .
dissociation of reaction complexes is ideally studied upon state - specific excitation of an initially cold complex . for cl@xmath57ch@xmath26br
it was found that excitation of high - frequency intramolecular vibrations in the ch@xmath26br moiety by a co@xmath32 laser leads exclusively to the product formation br@xmath0 + ch@xmath26cl @xcite .
this was predicted for this system by classical trajectory calculations from hase and coworkers @xcite , which demonstrated an enhancement of central barrier crossing upon selective excitation .
similarly , excitation of the doubly degenerate c - h stretch modes in the same complex induces central barrier crossing and the formation of br@xmath0 products @xcite .
craig et al .
created the s@xmath242 intermediate [ cf@xmath26co@xmath32ch@xmath58cl@xmath0]@xmath50 in a highly vibrationally excited state by means of a precursor exothermic association reaction @xcite .
this resulted in an at least four - fold enhancement of the branching ratio for the s@xmath242 reaction pathway as compared to unexcited complexes .
the observation is in disagreement with statistical rrkm calculations and was interpreted as manifestation of a bottleneck for energy transfer between intra- and intermolecular modes on the timescale of the complex lifetime @xcite .
chemical reactions involve the nuclear motion from reactants to products as well as the coupled structural and energetic transformation of molecular orbitals .
ultrafast laser pulses allow to follow half - reactions in real time by photoinitiating the dissociation of a transition state and probing the evolution to products with a second photon in a pump - probe experiment @xcite . in time - resolved photoelectron spectroscopy ( trpes )
the probe laser generates free electrons by photoionisation or photodetachment , whose kinetic energy and eventually angular distribution is measured . since trpes is sensitive to both the electronic configurations and the vibrational dynamics , it has been a particularly successful tool for real - time insight into molecular photodynamics ( see recent reviews @xcite ) . in pioneering trpes studies of negative ions , neumark and coworkers time -
resolved the photodissociation of the i@xmath1 anion @xcite .
an ultrashort laser pulse at 780 nm pumps ground state i@xmath1 ( @xmath60 ) to the first excited electronic state ( @xmath61 @xmath62 ) . following this excitation , i@xmath1 dissociates into ground state products i@xmath0 + i(@xmath63 ) with a 0.6ev kinetic energy release . at variable time delay with respect
to photoinitiation an ultrashort probe pulse is employed , which produces a photoelectron spectrum ( pes ) . for a 260 nm probe wavelength , two photoelectron bands are observed , which asymptotically correspond to photodetachment of the i@xmath0 photo - reaction product . for short pump - probe delays ,
the photoelectron bands shift to smaller kinetic energy for increasing delay , while at the same time their width narrows ( see also @xcite ) .
this effect was attributed to the change in character of the orbital from which the photoelectron is derived . by tracing the transition from the molecular to atomic orbital ,
the pes uncovers that the dissociation is complete within the first 320fs after photoinitiation .
following dissociation , a subtle shift of the photoelectron kinetic energy in the opposite direction is observed , which continues for another 400fs .
this behaviour was assigned to the interaction of the separating fragments , in particular the polarisation - induced charge - dipole attraction between the anion and the neutral atom .
it corresponds to a shallow well on the long - range part of the potential surface , which was characterised in the trpes experiment and determined to be 17mev deep @xcite .
measured photoelectron angular distributions reveale that the localisation of the excess charge on one of the atoms is only complete after about 800fs @xcite .
the observed dynamics in the exit channel of i@xmath1 photodissociation becomes upon time - reversal entrance channel dynamics for an i@xmath0 + i collision .
mabbs _ et al .
_ @xcite extended this work to the related system ibr@xmath0 .
here excitation at 780 nm - to the lowest optically bright excited electronic state - correlates to the second lowest product channel , i@xmath0 + br(@xmath63 ) . also here ,
a well is found in the long range part of the potential , but compared to i@xmath1 it is with about 60mev considerably deeper .
based on the result of a classical trajectory calculation , the authors transformed the time - axis to the intermolecular distance r - thus providing an `` image '' of the potential as a function of reaction coordinate .
@xcite showed very recently that for the same photoexcitation the presence of a single solvent molecule introduces a new product channel , the formation of i + br@xmath0 .
this was traced back to a non - adiabatic transition to one of the lower - energy electronic states driven by the solvent molecule , whose vibrational temperature was found to play a critical role in the process .
ch@xmath26i ( upper panel ) and i@xmath0 ( lower panel ) in the femtosecond photoelectron spectra .
biexponential decay or growth curves , shown as solid lines , are found reproduce these signals @xcite . ]
one of the authors _ et al . _ performed the only time - resolved investigation of a bimolecular s@xmath242 reaction in the gas phase by studying i@xmath0 + ch@xmath26i @xcite .
the reactants were derived from the precursor ion - dipole - bound cluster i@xmath64ch@xmath26i . start trigger for the reaction is a femtosecond pump pulse , which dissociates the i@xmath1 chromophore . upon its photodissociation
, the neutral iodine leaves the cluster , while the s@xmath242 reactants i@xmath0 and ch@xmath26i start to interact due to their charge - dipole interaction .
a probe pulse creates a pes , which identifies reaction transients and products as a function of interaction time . in the experiment two different pump photon energies
have been employed , which result in different kinetic energies of the i@xmath0 reactant . at @xmath65 = 790 nm
, this leads to a relative kinetic energy between i@xmath0 and ch@xmath26i of 0.15ev , where the interaction of ch@xmath26i with i@xmath0 during dissociation has been neglected .
the pes reveal that in this case the dominant process is the production of the vibrationally excited entrance channel complex i@xmath66ch@xmath26i on a time scale of 600fs .
in contrast for @xmath65 = 395 nm , the relative kinetic of the reactants i@xmath0 and ch@xmath26i is 0.32ev ( see fig . [ sn2-fpes : fig ] ) .
contributions of i@xmath66ch@xmath26i and i@xmath0 are now identified in the pes and can be separated .
the complex contribution increases rapidly after the pump pulse with a time constant indistinguishable from the laser cross correlation of 200fs .
interestingly , it then shows an exponential decay on two different time scales , which is mirrored in the i@xmath0 contribution .
since the reactants do not have sufficient energy to cross the central reaction barrier , this is interpreted as decay of the complex back to reactants .
the fast time constant was determined to be 0.75ps , which compares to the vibrational period of the i@xmath49ch@xmath26i stretching mode .
it suggests fairly direct dissociation dynamics in which the i@xmath0 undergoes a quasi - elastic collision with ch@xmath26i before dissociation .
the longer time constant of about 10ps indicates that the complex is stabilised by energy flow from the reaction coordinate into the modes of the complex .
it is comparable to the lifetime of the cl@xmath66ch@xmath26cl entrance channel complex determined in the collisional stabilisation experiments featured above .
in the previous sections we have discussed measurements of the total cross section or the thermally averaged rate coefficient and illustrated how information on the reaction dynamics can be inferred from these measurements .
a much more direct approach to the intrinsic dynamics of anion - molecule reactions is based on the measurement of the angle- and energy - differential scattering cross section @xmath67 . for these experiments collisions of atoms and molecules with well - defined momentum vectors
are prepared .
angle- and energy- or velocity - resolved detection schemes are employed to obtain the differential cross section .
this approach of crossed - beam measurements of the differential scattering cross section is widely used for neutral reactions @xcite .
we will discuss in the following two sections that it also proves to be valuable for anion - neutral reactions .
the classical approach to crossed - beam reactive scattering is to cross two reactant beams at 90@xmath68 and use a rotatable detector to measure the flux and the arrival time of products under a selected set of scattering angles in the laboratory frame of reference . from these laboratory flux data the differential cross section in the centre - of - mass frame of the reaction
is reconstructed , often using numerical simulations of the apparatus function .
this approach has been developed for neutral - neutral reactions and has been used with great success to study the f + h@xmath32 reaction ( see e.g.@xcite ) . in combination with the rydberg tagging technique to measure the h atom product ,
the rotatable - detector setup is still the method of choice for this reaction @xcite .
the same detector concept has been applied to study elastic , inelastic and reactive channels in cation - molecule reactions ( see refs.@xcite for reviews of the early work on cation - molecule reactive scattering ) . here
the main difference to neutral - neutral scattering experiments is , naturally , the production of the ion beam . where the neutral beams are usually produced in supersonic expansions with a well defined narrow range of velocities , the production of ion beams with a narrow velocity spread represents a major experimental challenge . for the study of chemical reactions the most interesting relative collision energies in the centre - of - mass frame range from millielectronvolts to a few electronvolts .
therefore ion beams have to be produced with a low kinetic energy and a correspondingly small energy spread .
this has been achieved for the first time in the 1970s for continuous beams of cations @xcite .
a high degree of control over the electric fields in the experimental setup , including contact potentials between different materials , are important to achieve a low energy spread .
even then , the coulomb repulsion between the reactant ions in the beam ultimately limits the energy resolution for a given reactant density and thus scattering rate . for ion - molecule reactions
this typically leads to a reduction in the ion reactant density by many orders of magnitude in comparison with a neutral supersonic beam .
this is only to a small extend compensated by the higher detection efficiency for product ions in a charged - particle detector when compared to neutral products of neutral - neutral reactions which have to be ionised in the detector .
note that once the control over the ion - beam energy is achieved in an experimental setup , the relative collision energy of the crossed beams can be continuously tuned by means of the ion acceleration potential .
such a tunable scattering energy is much more difficult to achieve for neutral - neutral reactions .
there , different seed gases for the molecules in the supersonic beams and different intersection angles between the supersonic beams are required for the same purpose .
it was only in the 1990s that precise low - energy crossed - beam measurements with negative ions in a rotatable detector setup have been carried out @xcite . in the years before the first detailed scattering experiments on anion - molecule reactions have been carried out using a complex multiparticle coincidence scattering setup . with this setup reactions of heavy , mostly
atomic , negative ions ( f@xmath0 , cl@xmath0 , br@xmath0 , i@xmath0 , s@xmath0 , cn@xmath0 ) with light - weight hydrogen could be studied @xcite .
for example for the cl@xmath0 + h@xmath32 collision , the kinematics of this combination have made it possible to study collisions down to 6ev relative energy with an ion kinetic energies of 110ev @xcite , which is significantly easier to handle experimentally than ion beams with only few electronvolt kinetic energy .
furthermore , under these conditions the kinetic energy of the neutral product in the lab frame is large enough to be detected with a microchannel plate detector . with this technique the different channels for the collision of cl@xmath0 with h@xmath32 , reactive proton transfer , forming hcl + h@xmath0 , reactive detachment , forming hcl + h + e@xmath0 , simple detachment , leading to cl + h@xmath32 + e@xmath0 , and dissociative detachment , leading to cl + h + h + e@xmath0 , could be distinguished and their differential and absolute cross sections could be measured as a function of the relative energy @xcite .
also some information on the vibrational state population of the product molecule could be inferred . from these data
the similarity of the dynamics of reactive proton transfer and reactive detachment could be deduced , where the branching between the two channels depends on the coupling to an intermediate autodetaching hcl@xmath0 state .
different scattering dynamics was found for the simple detachment channel , bearing similarity with anion - rare gas collisions . with the first crossed - beam measurements on anion - molecule reactions using low - energy ion sources and rotatable detectors , relative collision energies down to the millielectronvolt range and an energy resolution
sufficient to resolve product vibrational states became accessible .
in a benchmark experiment on the reaction of h@xmath0 with d@xmath32 zimmer and linder could determine the vibrational state - to - state differential and integral scattering cross section @xcite .
the dominate forward scattering of the d@xmath0 product was attributed to a collinear reaction mechanism that requires scattering at small impact parameter .
wider angular distributions were observed for higher vibrational excitation of the hd product . from the measured product kinetic energy spectra
also rotational state information could be extracted , studied in more detail for the inelastic scattering in h@xmath0 + h@xmath32 collisions in the same laboratory @xcite .
+ d@xmath32 at 89kj / mol relative energy .
the orthogonal black lines indicate the velocities of the incoming beams .
the point in the centre of the circles represents zero velocity in the centre of mass frame . around this point
the circles denote different product velocities .
smaller velocities indicate internal excitation of the hod product molecule due to vibrational excitation of the bending mode ( v@xmath26 ) or the od stretching mode ( v@xmath32 ) .
the measured flux shows that internal excitation , attributed to the bending mode of hod , does occur during the reaction .
reprinted with permission from @xcite .
copyright ( 2005 ) , american chemical society . ]
also in the 1990s , farrar and coworkers started crossed - beam reactive scattering experiments of atomic oxygen with small closed - shell molecules , such as water , ammonia and hydrogen ( see e.g.@xcite ) .
for the o@xmath0 + h@xmath32o reaction , which had already been studied before with crossed beams at higher collision energy and with lower resolution @xcite , they observed in the differential cross section both a direct and an indirect reaction mechanism for the reactive channel of oh@xmath0 formation @xcite .
in addition they could study inelastic scattering via the o@xmath0(h@xmath32o ) reaction complex .
more recently , non - reactive and reactive collisions of oh@xmath0 anions with d@xmath32 molecules have been studied by the same group @xcite . in the isotopic exchange reaction channel ,
the formation of od@xmath0 + hd , no energy dependence was found between 0.27 and 0.69ev relative collision energy @xcite . in the measured narrow angular distribution
it was found that the reaction occurs fast on the time scale of rotation of the oh@xmath0(h@xmath32 ) reaction complex . in a study of the proton transfer reaction channel , forming hod + d@xmath0 ,
also a very narrow velocity distribution was extracted @xcite .
the reconstructed velocity distribution in the scattering plane is reproduced in fig .
[ farrar - data : fig ] .
these data also indicate a direct and fast reaction mechanism .
the flux at smaller product ion velocities is caused by vibrational excitation of the bending mode in the hod product .
this can be attributed to a significant change of the water bond angle during the d@xmath43 transfer .
compared to the conventional crossed - beam experiments with a rotatable detector , higher angular resolution and a much more rapid data acquisition is achieved with an imaging spectrometer .
moreover , such a spectrometer detects products irrespective of their velocity or scattering angle , i.e. it represents a detector with @xmath69 solid angle of acceptance .
ion imaging has been combined with neutral reactive scattering starting with a study of the h + d@xmath32 reaction @xcite and is being successfully employed in a number of laboratories for reactive and inelastic crossed - beam scattering experiments @xcite .
the current spectrometers are based on the technique of velocity map imaging @xcite , which projects ions with the same velocity parallel to the detector surface onto the same spot on the detector .
it thereby avoids broadening of the product ion images due to the finite size of the reaction volume . to study ion - molecule reactions with crossed beam imaging
, we have constructed a velocity map imaging spectrometer and a versatile low - energy ion source @xcite .
this approach is in contrast with an early exploration of ion - molecule crossed beam imaging , where an in - situ production of the reactant ions was used @xcite , an approach that is not usable for negative ions . in our experiment , slow ions with between 0.5 and 5ev kinetic energy are brought to collision with neutral molecules in a supersonic molecular beam in the centre of the velocity map imaging electrode stack .
both reactant beams are pulsed to avoid a heavy gas load in the vacuum system .
once the two beams have crossed , the electric field of the imaging spectrometer is rapidly pulsed on and any product ions are projected onto the position sensitive imaging detector .
ion impact positions , which are proportional to the velocity components in the scattering plane parallel to the detector surface , are recorded with a ccd camera .
recently , we improved the ion imaging electrodes to enhance the velocity resolution and we included a photomultiplier tube to measure the third , vertical , component of the product velocity vector @xcite . with the ion - molecule crossed beam imaging spectrometer we have studied the elementary s@xmath242-reaction of cl@xmath0 with methyl iodide ( ch@xmath26i ) at relative scattering energies between 0.4 and 2ev @xcite . the potential energy curve for this reaction along the reaction coordinate
is closely resembled in fig .
[ reaction - pot : fig ] .
as discussed already in section [ interaction : sect ] the characteristic deep potential minima , separated by an intermediate barrier , are found . in the reaction studied here
this barrier lies submerged below the energy of the reactants .
nevertheless it strongly influences the reactivity in that the reaction occurs at only 10% of the langevin or capture rate .
the imaging data , which is represented by an event list of impact velocity vectors in the laboratory frame , needs only little data processing . on the one hand
, the transformation to the centre of mass frame has to be performed for each scattering event , which involves a translation and a rotation of the image parallel to the relative velocity vector . on the other hand
, the loss of fast moving reaction products needs to be corrected .
product ions with high laboratory velocities have a chance to leave the spectrometer volume that can be imaged onto the position - sensitive detector .
this effect is referred to as density - to - flux correction in neutral crossed beam scattering experiments , where it also includes the correction for the spatial and velocity dependence of the ionisation efficiency . for ion - molecule
reactive imaging , the same correction function is used for all relative collision energies .
it only depends on the magnitude of the laboratory velocity and not on its direction in the lab frame .
measured differential scattering cross section for the reaction of cl@xmath0 + ch@xmath26i giving clch@xmath26 + i@xmath0 . shown
are two images at 0.39 ( left panel ) and 1.9ev ( right panel ) relative collision energy of the i@xmath0 velocity vector in the scattering plane obtained by velocity map slice imaging .
the centre of each image denotes zero velocity in the centre of mass frame .
the circles represent constant product velocities with the largest circle showing the maximum possible product velocity based on the known total energy in the reaction system . at 0.39ev isotropic scattering
is observed , indicative of an indirect reaction mechanism via a long - lived complex . at 1.9ev
most of the flux shows direct scattering with large product velocities ( peak near @xmath70 m/s , but with about 10% probability , small product velocities both forward and backward scattered , are observed .
these events are attributed to the indirect `` roundabout '' mechanism ( taken from @xcite ) . ]
two measured images of the differential cross section in the centre - of - mass frame are shown in fig .
[ sn2-images : fig ] for two relative energies @xcite . at 0.39ev isotropic scattering of the i@xmath0 product ion
is observed , indicative of an indirect reaction mechanism via a long - lived complex . here , also much more energy is partitioned into internal excitation of the neutral ch@xmath26cl product .
this indicates an indirect reaction mechanism with trapping of the collision partners in the minima of the intermolecular reaction potential .
in contrast , at 1.9ev the i@xmath0 ions scatter preferentially backward with respect to the direction of the incoming ch@xmath26i .
also their velocity is found very near the maximum possible velocity .
this is explained by a fast and direct reaction mechanism where the i@xmath0 leaves the reaction approximately co - linearly with the incoming cl@xmath0 anion . in order to understand the details of the measured differential cross sections theoretical calculation
are employed .
however , the theoretical description of polyatomic reactions that involve more than four atoms is very difficult .
the present reaction involves six atoms and therefore twelve internal degrees of freedom .
such a large system can not be calculated quantum mechanically and one has to resort to significant approximations .
these are either quantum scattering calculations in reduced dimensions ( typically four ) @xcite or calculations that treat the electronic structure quantum mechanically but propagate the nuclei classically on the born - oppenheimer surface @xcite .
calculated trajectories show that at 1.9ev collision energy a direct reaction mechanism governs the nucleophilic substitution reaction .
the cl@xmath0 ions moves into the umbrella of the hydrogen atoms and forms a bond with the central carbon atom .
roughly co - linearly to this motion the i@xmath0 product ion is moving away after the three hydrogen atoms have inverted to form the clch@xmath26 product molecule .
this numerical results corresponds directly to the back - scattering observed in the experiment .
the trajectory calculations also revealed another reaction mechanism that occurs with about 10% probability . in this mechanism
the ch@xmath26i reactant undergoes a single 360@xmath68 revolution about an axis perpendicular to the c - i bond . only after this revolution the substitution occurs .
this mechanism , which we named the `` roundabout '' mechanism , is found to go along with a large energy partitioning into internal degrees of freedom of the neutral product molecule .
this agrees with the observed structures in the measured differential cross sections at small i@xmath0 velocity ( see fig . [
sn2-images : fig ] ) , which have therefore been attributed to the roundabout mechanism @xcite .
negative ion reactions have been studied for a long time owing to their importance in many earth - bound , planetary or astrophysical plasmas . in the last two decades research on the detailed dynamics of this class of reactions
has flourished , owing to more and more precise techniques to measure both integral and differential scattering studies . in this article
we have presented an overview of these experimental approaches and how much they have been teaching us on the different , and sometimes peculiar , aspects of anion - molecule reaction dynamics .
we have shown that flow and drift tube measurements are well suited to study the thermal kinetics of anion - molecule reactions .
at low temperature these studies are complemented and extended using cryogenic ion traps , which also can measure much smaller rate coefficients than drift tubes due to the long attainable interaction times . guided ion beam experiments are ideal for precise measurements of integral cross sections at well - defined relative energy over a large dynamic range .
the dynamics of the transient reaction complex can be studied with direct time - resolved spectroscopy and its lifetime can also be indirectly inferred from ternary collision rates .
when it comes to direct imaging the full reaction dynamics , crossed - beam experiments are very revealing . in particular the new opportunities of crossed - beam velocity map ion imaging should be stressed here . despite a wealth of theoretical studies , detailed insight into the flow of energy via the coupling of different vibrational modes during
the reaction continues to be not accessed experimentally .
of particular interest is the regime of breakdown of ergodicity at the transition from a complex mediated , statistical to an impulsive reaction mechanism .
it is in this range of relative collision energies , where crossed beam imaging provides the best resolution .
reactive scattering with vibrationally excited reactants - as successfully used to shed light on chemical reactions of neutrals @xcite - is not yet explored for anion - molecule reactions .
the role of non - reactive degrees of freedom in a reaction , referred to as the `` spectator modes '' , on the integral cross section and the product branching is a matter of debate that needs answers from experiments .
reactive scattering with spatially aligned or oriented molecules would directly access the stereodynamics on the molecular level .
importantly , anion - molecule reactions have very different kinetics in solution .
a bottom - up approach , introducing the solvent molecule per molecule in scattering experiments with microsolvated anions , would highlight the role of the environment on the dynamics .
questions about the role of quantum effects such as feshbach scattering resonances , tunnelling , zero - point vibrational motion or decoherence free subspaces can be dared to be asked now and demand innovative experimental approaches .
we expect that guided ion beam and ion trap studies as well as ion - molecule crossed - beam imaging will be most helpful to address some of these questions in the future .
the combination of an ion trap as a source for internally cold molecular ions and clusters with a scattering experiment will allow for studies of the reactions of complex molecular systems while maintaining good control over their internal quantum states . also merged ion and neutral beams represent a useful approach to study low - energy reactions , as demonstrated most recently for the reaction of h@xmath0 with h @xcite , which is considered to be important for the formation of h@xmath32 in the early universe .
laser - cooling of the reactants may eventually provide even lower collision temperatures than buffer gas cooling , an approach that has been demonstrated for cation - neutral reactions @xcite ; os@xmath0 is a possible candidate for anion laser cooling @xcite .
all in all , there is also a lot of room for further research on anion - molecule reactions , that challenges the `` complexity limit '' and ultimately allows us to understand and control all the subtleties of the chemical dynamics of multi - atom molecular systems .
we would like to thank our collaborators in the experiments on anion - molecule reaction dynamics , in particular sebastian trippel , rico otto , and christoph eichhorn .
we also wish to thank dan neumark and bill hase for the fruitful collaborations on reaction dynamics in general and nucleophilic substitution reactions in particular .
our research is supported by the elitefrderung der landesstiftung baden - wrttemberg , the alexander von humboldt foundation , the deutsche forschungsgemeinschaft and the eu marie curie initial training network iconic .
|
this review discusses the dynamics of negative ion reactions with neutral molecules in the gas phase .
most anion - molecule reactions proceed via a qualitatively different interaction potential than cationic or neutral reactions .
it has been and still is the goal of many experiments to understand these reaction dynamics and the different reaction mechanisms they lead to
. we will show how rate coefficients and cross sections for anion - molecule reactions are measured and interpreted to yield information on the underlying dynamics .
we will also present more detailed approaches that study either the transient reaction complex or the energy- and angle - resolved scattering of negative ions with neutral molecules . with the help of these different techniques
many aspects of anion - molecule reaction dynamics could be unravelled in the last years
. however , we are still far from a complete understanding of the complex molecular interplay that is at work during a negative ion reaction .
|
let @xmath2 , be a configuration of lattice points spanning @xmath3 .
we also denote by @xmath0 the @xmath4 integer matrix with columns @xmath5 .
we say that the configuration @xmath0 is _ regular _ if the points of @xmath0 lie in a hyperplane off the origin .
the _ dimension _ of @xmath0 is defined as the dimension of the affine span of its columns and the _ codimension _ as the rank of the lattice @xmath6 following gelfand , kapranov and zelevinsky @xcite we associate to @xmath0 and a _ parameter _ vector @xmath7 a left ideal in the weyl algebra in @xmath8 variables @xmath9 as follows .
[ def : hypergeom ] given @xmath10 of rank @xmath11 and a vector @xmath12 , the _ @xmath0-hypergeometric system _ with parameter @xmath13 is the left ideal @xmath14 in the weyl algebra @xmath15 generated by the _
toric operators _ @xmath16 , for all @xmath17 such that @xmath18 , and the _ euler operators _ @xmath19 for @xmath20 .
a holomorphic function @xmath21 , defined in some open set @xmath22 , is said to be _
@xmath0-hypergeometric of degree @xmath13 _ if it is annihilated by @xmath14 .
@xmath0-hypergeometric systems include as special cases the homogeneous versions of classical hypergeometric systems in @xmath23 variables .
the ideal @xmath14 is always holonomic and if @xmath0 is regular it has regular singularities .
the singular locus of the hypergeometric @xmath24-module @xmath25 equals the zero locus of the principal @xmath0-determinant @xmath26 , whose irreducible factors are the sparse discriminants @xmath27
corresponding to the facial subsets @xmath28 of @xmath0 @xcite . often , the existence of special solutions to a system of equations imposes additional structure on the data ( see for example a recent preprint @xcite of beukers on algebraic a - hypergeometric functions . ) in this paper we are interested in the constraints imposed on @xmath0 by the existence of rational @xmath0-hypergeometric functions .
all @xmath0-hypergeometric systems admit polynomial solutions for parameters @xmath13 in @xmath29 , which are closely related to the solutions of an integer programming problem associated to the data @xmath30 @xcite . likewise ,
for every @xmath0 there exist laurent polynomial solutions to the @xmath0-hypergeometric system . clearly , these rational solutions are annihilated by a sufficiently high partial derivative .
the goal of this paper is to characterize all codimension - two lattice configurations @xmath0 which admit a rational @xmath0-hypergeometric function none of whose derivatives vanishes .
such rational functions are called _
stable_. we will assume that @xmath0 is not a _ pyramid _ ; that is , a configuration all of whose points , except one , are contained in a hyperplane .
this entails no loss of generality . indeed , suppose the subset @xmath31 lies in a hyperplane not containing @xmath32 , then all @xmath0-hypergeometric functions are of the form : @xmath33 where @xmath34 is @xmath28-hypergeometric .
hence , if @xmath0 is a pyramid over a configuration @xmath28 which admits a stable @xmath28-hypergeometric function then , clearly , so does @xmath0 . in order to state our results we need to describe certain special configurations , which play an important role throughout this paper .
a configuration @xmath35 is said to be a _ cayley configuration _ if there exist vector configurations @xmath36 in @xmath37 such that @xmath38 where @xmath39 is the standard basis of @xmath40 .
note that we may assume that all the @xmath41 s consist of at least two points since , otherwise , @xmath0 would be a pyramid .
a cayley configuration is said to be a _
configuration if all the configurations @xmath41 consist of exactly two points .
thus , up to affine isomorphism , we may assume that @xmath42 , @xmath43 .
it follows from our assumptions that the vectors @xmath44 must span @xmath37 over @xmath45 .
we note that the codimension of a lawrence configuration is @xmath46 .
we say that a cayley configuration is _ essential _ if @xmath47 and the minkowski sum @xmath48 has affine dimension at least @xmath49 for every proper subset @xmath50 of @xmath51 . for a codimension - two essential cayley configuration ,
@xmath52 of the configurations @xmath41 , say @xmath53 , must consist of two vectors and the remaining one , @xmath54 , must consist of three vectors .
if we set @xmath55 , @xmath56 , then it follows from the fact that @xmath0 is essential that the vectors @xmath57 are linearly independent over @xmath58 .
thus , modulo affine equivalence , we may assume without loss of generality that @xmath42 , @xmath56 , where @xmath59 are linearly independent over @xmath58 and @xmath60 with @xmath61 , @xmath62 are not both contained in a subspace generated by a proper subset of @xmath59 . in order to simplify our statements
we will allow ourselves a slight abuse of notation and consider the configuration @xmath63 as a lawrence configuration and the zero - dimensional configuration @xmath64 as a cayley essential configuration .
it has been shown in @xcite that both lawrence configurations and essential cayley configurations admit stable rational @xmath0-hypergeometric functions .
this is done by exhibiting explicit functions constructed as toric residues .
our main result asserts that if @xmath0 has codimension two then these are the only configurations that admit such functions .
[ th : conj ] a codimension two configuration @xmath0 admits a stable rational hypergeometric function if and only if it is affinely equivalent to either an essential cayley configuration or a lawrence configuration . as an immediate corollary to theorem [ th : conj ]
we obtain a proof for the codimension - two case of conjecture 1.3 in @xcite .
we recall that a configuration @xmath0 is said to be _ gkz - rational _ if the discriminant @xmath65 is not a monomial and @xmath0 admits a rational @xmath0-hypergeometric function with poles along the discriminant locus @xmath66 .
such a function is easily seen to be stable .
thus , by theorem [ th : conj ] , @xmath0 must be either a lawrence or a cayley essential configuration .
but , if @xmath67 , the sparse discriminant of a lawrence configuration is @xmath68 , and therefore the only codimension - two gkz - rational configurations are cayley essential as asserted by ( * ? ? ?
* conjecture 1.3 ) .
let us briefly outline the strategy for proving theorem [ th : conj ] .
the fact that an @xmath0-hypergeometric function @xmath69 of degree @xmath13 satisfies @xmath11 independent homogeneity relations , one for each row of the matrix @xmath0 , implies that the study of codimension - two rational @xmath0-hypergeometric functions may be reduced to the study of rational power series in two variables whose coefficients satisfy certain recurrence relations .
now , it follows easily from the one - variable residue theorem that the _ diagonals _ of a rational bivariate power series define algebraic one - variable functions . on the other hand , coming from an @xmath0-hypergeometric function , these univariate functions are classical one - variable hypergeometric functions .
theorem [ notalgebraic ] allows us to reduce the study of these one - variable functions to those studied by beukers - heckman @xcite ( see also @xcite ) . analyzing the possible functions arising as diagonals of a bivariate rational function leads us to conclude that @xmath0 must be affinely equivalent to an essential cayley configuration or a lawrence configuration . in the latter case ,
the stable rational @xmath0-hypergeometric functions have been studied in @xcite where it is shown that an appropriate derivative of such a function may be represented by a multivariate residue . in [ sec : residues ] we show that a similar result holds for essential cayley configurations of codimension two . after recalling the construction of rational @xmath0-hypergeometric functions by means of toric residues , we show in theorem [ th : dim1 ] ,
that if the parameter @xmath13 lies in the so - called euler - jacobi cone @xmath70 ( see ) , the space of rational @xmath0-hypergeometric function of degree @xmath13 is one - dimensional .
this proves ( * ? ? ?
* conjecture 5.7 ) for any codimension - two essential cayley configuration .
finally , in section [ sec : rat - bivariate ] we apply our results to study the rationality of classical bivariate hypergeometric series ( in the sense of horn , see definition [ def : horn ] and remark [ rem : horn ] ) .
theorem [ th : conjclassic ] shows that any bivariate taylor series whose coefficients are quotients of factorials of integer linear forms as in defines a rational function only if the linear forms arise from a lawrence or cayley essential configuration .
we end up by considering the case of horn series supported in the first quadrant .
* acknowledgments : * ec would like to thank the fulbright program and the university of buenos aires for their support and hospitality . ad was partially supported by ubacyt x064 , conicet pip 5617 and anpcyt pict 20569 , argentina .
frv would like to thank the program raices of argentina and the nsf for their financial support .
he would also like to thank the department of mathematics of the universidad de buenos aires , argentina and the arizona winter school , where some of this work was done .
in this section we study algebraic hypergeometric series of the form @xmath71 we are interested in the case when the series ( [ series ] ) has a finite , non - zero , radius of convergence .
hence we assume that @xmath72 the case @xmath73 for @xmath74 , namely , the series @xmath75 has been studied in @xcite .
if @xmath76 all coefficients are equal to @xmath68 and @xmath77 is rational .
assume then that @xmath78 .
using the work of beukers and heckman @xcite it was shown in @xcite that @xmath79 defines an algebraic function if and only if the _ height _ , defined as @xmath80 , equals @xmath68 and the factorial ratios @xmath81 are integral for every @xmath82 .
( in the last case , @xmath79 is not a rational function , in fact , since by stirling the coefficients are , up to a constant , asymptotic to @xmath83 times an exponential . )
beukers and heckman @xcite actually gave an explicit classification of all algebraic univariate hypergeometric series .
as a consequence , we can also classify all integral factorial ratio sequences ( [ ratios ] ) of height @xmath68 ( see @xcite,@xcite , ( * ? ? ?
* theorem 1.2 ) ) .
we may clearly assume that @xmath84 then there exist three infinite families , where @xmath85 is given by @xmath86 @xmath87 or @xmath88 and @xmath89 sporadic cases listed in ( * ? ? ?
* table 2 ) . because of the connections with step functions , it is also interesting to study integral factorial ratio sequences satisfying ( [ samesum ] ) but of height different than one .
partial results in this direction are contained in @xcite .
the connections with quotient singularities and the riemann hypothesis are explored in @xcite .
note that we can write a series @xmath90 as in as follows @xmath91 where @xmath92 is the polynomial @xmath93 and @xmath85 is as in .
we now show that @xmath90 and @xmath79 can only be algebraic simultaneously .
more generally , we have the following . [ notalgebraic ]
suppose @xmath94 where @xmath95 $ ] is non - zero and @xmath85 is as in .
then : \(i ) the series @xmath96 is algebraic if and only if @xmath97 is algebraic .
\(ii ) if @xmath90 is rational then @xmath98 for all @xmath8 and @xmath99 we first prove ( i ) .
one direction is clear as @xmath100 .
suppose then that @xmath90 is algebraic . by a theorem of eisenstein ( see @xcite for a modern treatment and further references ) the coefficients @xmath101 of @xmath90 are integral away from a finite set of primes .
we may assume without loss of generality that @xmath92 is primitive , i.e. , that the @xmath102 of all of its coefficients is @xmath68 .
hence , for any prime @xmath103 there are at most @xmath104 congruences classes @xmath105 for which @xmath106 , where @xmath107 denotes the valuation at @xmath103 .
it follows that for all sufficiently large primes @xmath103 the number of exceptions to @xmath108 is at most @xmath104 , independent of @xmath103 . in other words ,
the valuation at @xmath103 of the coefficientes of @xmath90 is essentially that of the coefficients of @xmath79 .
we will exploit this fact in order to prove the theorem .
it is easy to verify ( see @xcite for details on the following discussion ) that @xmath109 where @xmath110 is the _ landau function _ @xmath111 here @xmath112 denotes the fractional part of @xmath113 .
the following properties of @xmath110 hold : @xmath110 is periodic , with period @xmath68 , locally constant , right continuous with at most finitely many step discontinuities , @xmath114 and away from the discontinuities @xmath115 furthermore , by a theorem of landau , @xmath116 for all @xmath8 if and only if @xmath117 for all @xmath113 . since @xmath118 , for all sufficiently large primes @xmath103 @xmath119 indeed , as @xmath110 is locally constant we have @xmath120 for @xmath121 for some @xmath122 .
if @xmath123 and @xmath124 then @xmath125 more generally , let @xmath126 be a decomposition of @xmath127 into finitely many disjoint subintervals @xmath128 such that @xmath110 is constant on each @xmath129 .
let @xmath130 the minimum length of the @xmath129 s . if @xmath131 for some integer @xmath132 then the number of rationals of the form @xmath133 in each @xmath129 is at least @xmath134 .
taking @xmath135 and combining with we conclude that @xmath117 for all @xmath136 . consequently , @xmath116 for all @xmath8 and also @xmath137 by . if @xmath138 then @xmath139 as @xmath140 by .
it follows that in this case @xmath141 and @xmath142 are both rational and @xmath76 .
hence we may assume @xmath143 .
we can write the series @xmath97 as a hypergeometric series ( recall we assume ) @xmath144 for some @xmath145 in @xmath58 for @xmath146 , with @xmath147 for all @xmath146 and where @xmath148 .
note that the number of @xmath13 s that equal @xmath68 is precisely @xmath11 .
hence , since @xmath149 at least one of factors in the denominator of the coefficient of @xmath150 is @xmath151 and @xmath152 is a classical @xmath153 hypergeometric series .
we remark that the discontinuities of @xmath110 in @xmath154 occur precisely at the @xmath155 s and @xmath156 s .
it follows that @xmath157 , where @xmath158 is the space of local solutions to the corresponding hypergeometric differential equation @xmath159 at some base point @xmath160 .
the nature of the parameters @xmath161 of @xmath162 guarantees that the action of monodromy on @xmath158 is irreducible ( see ( * ? ? ?
* proposition 3.3 ) ) . on the other hand ,
let @xmath163 be the space of local functions at @xmath164 obtained by analytic continuation of @xmath96 .
the map @xmath165 preserves the action of monodromy . by the irreducibility of @xmath158
this map is injective .
we conclude that the monodromy group of @xmath158 must be finite since this is true of @xmath163 given the hypothesis that @xmath90 is algebraic .
this shows , in turn , that @xmath79 is algebraic .
now assume that @xmath90 is rational . if @xmath143 the above argument applies and since the monodromy group of @xmath163 is trivial
so is that of @xmath158 .
therefore @xmath79 is rational contradicting the assumption that @xmath143 . to see this note , for example , that @xmath143 implies that @xmath110 is not identically zero by and hence @xmath166
. in particular , the local monodromies are not trivial .
we conclude that @xmath138 and consequently , as pointed out above , @xmath167 proving ( ii ) .
[ alg ] note that @xmath11 is the multiplicity of the eigenvalue @xmath68 of the local monodromy action on @xmath158 at @xmath168 . by levelt s theorem ( @xcite , ( * ? ? ?
* theorem 3.5 ) ) this monodromy has a jordan block of size @xmath11 ( with eigenvalue @xmath68 ) and hence can not be finite if @xmath169 .
in this section we discuss laurent series expansions for rational functions in two variables .
we prove a lemma which will be of use in [ sec : residues ] and recall one of the key tools to determine whether a bivariate series defines a rational function , namely the observation that a _ diagonal _ of a rational bivariate series is algebraic .
let @xmath170 $ ] be polynomials in two variables without common factors and let @xmath171 .
we denote by @xmath172 the newton polytope of @xmath173 . throughout this section
we will assume that @xmath174 is two - dimensional .
let @xmath175 be a vertex of @xmath174 , @xmath176 the adjacent vertices , indexed counterclockwise and @xmath177 , @xmath178 .
hence , @xmath179 we can write @xmath180 with the support of @xmath181 contained in the cone @xmath182 .
thus we obtain a laurent expansion of the rational function @xmath183 as @xmath184 whose support is contained in a cone of the form @xmath185 for a suitable @xmath186 .
that is , @xmath183 has an expansion @xmath187 whose support @xmath188 is contained in @xmath185 for some @xmath186 .
moreover , the above series converges in a region of the form @xmath189 sufficiently small , as observed in ( * ? ? ?
* proposition 1.5 , chapter 6 ) .
[ supportcone ] given a series ( [ eq : expansion1 ] ) as above then , for each @xmath178 , there exist infinitely many exponents of the form @xmath190 , @xmath191 , @xmath192 , such that @xmath193 .
in particular , the support of the series ( [ eq : expansion1 ] ) is not contained in any subcone @xmath194 , where @xmath195 is properly contained in @xmath196 . we may assume without loss of generality that @xmath197 and that @xmath198 , @xmath199 .
it then suffices to show that for some @xmath200 , the series ( [ eq : expansion1 ] ) contains infinitely many terms with non - zero coefficient and exponent of the form @xmath201 , @xmath202 .
we write @xmath203 , @xmath204 and view them as relatively prime elements in the ring @xmath205[x_1]$ ] . the laurent series expansion ( [ eq : expansion1 ] ) for the rational function @xmath206 may be written as @xmath207 where @xmath208 lie in the fraction field of @xmath205 $ ] , that is , the field of rational functions @xmath209 .
now , it follows from ( * ? ? ?
* lemma 3.3 ) that since @xmath210 is not a monomial and , therefore , not a unit in the laurent polynomial ring @xmath205 $ ] , at least one of the coefficients @xmath211 is not a laurent polynomial and , hence there exist infinitely many non - zero terms with exponents of the form @xmath201 . given a bivariate power series @xmath212 and @xmath213 , with @xmath214 , we define the @xmath215-diagonal of @xmath216 as : @xmath217 the following observation goes back to at least polya @xcite .
we include a proof for the sake of completeness .
[ diag ] if the series ( [ doubleseries ] ) defines a rational function , then for every @xmath218 , with @xmath219 , the @xmath215-diagonal @xmath220 is algebraic .
the key observation is that by the one - variable residue theorem , we can write for @xmath221 and @xmath222 small enough @xmath223 where @xmath224 are integers such that @xmath225 .
thus , @xmath220 , being the residue of a rational function , is algebraic .
we refer the reader to @xcite for generalizations of this result to rational series in more than two variables and to furstenberg @xcite and deligne @xcite for the situation in characteristic @xmath226 where diagonals of rational functions on any number of variables are algebraic .
the laurent expansions of a rational @xmath0-hypergeometric series are constrained by the combinatorics of the configuration @xmath0 . in this section
we sketch the construction of such series .
the reader is referred to @xcite for details .
let @xmath0 be a regular configuration .
as always , we assume , without loss of generality , that the points of @xmath0 are all distinct and that they span @xmath3 .
we also assume that @xmath0 is not a pyramid .
we consider the @xmath227-vector space : @xmath228 of formal laurent series in the variables @xmath229 .
the matrix @xmath0 defines a @xmath230-valued grading in @xmath231 by @xmath232 the weyl algebra @xmath233 acts in the usual manner on @xmath231 .
we will say that @xmath234 is @xmath0-hypergeometric of degree @xmath13 if it is annihilated by @xmath14 , i.e. @xmath235 denote by @xmath236 the vector of differential operators @xmath237 .
since for any @xmath238 we have @xmath239 , it follows that if @xmath234 is @xmath0-hypergeometric of degree @xmath13 then it must be @xmath0-homogeneous of degree @xmath13 and , in particular , @xmath240 . by (
* proposition 5 ) , if a hypergeometric laurent series has a non trivial domain of convergence , then its exponents must lie in a strictly convex cone .
we make this more precise .
let @xmath241 for any vector @xmath242 we define its _ negative support _ as : @xmath243 and given @xmath244 , we let @xmath245 .
we call @xmath246 a cell in @xmath247 .
[ minimalsupport ] we say that @xmath246 is a _ minimal cell _ if @xmath248 and @xmath249 for @xmath250 .
given a minimal cell @xmath251 we let @xmath252 given a non zero @xmath253 , @xmath254 and @xmath255 a @xmath45-basis of the lattice @xmath256
satisfying @xmath257 for all @xmath258 we let @xmath259 be the open set : @xmath260 the following is essentially a restatement of proposition 3.14.13 , theorem 3.4.14 , and corollary 3.4.15 in @xcite : [ laurentseries ] let @xmath253 be such that the collection @xmath261 of minimal cells @xmath246 contained in some half - space @xmath262 is non - empty .
then : \(i ) for @xmath263 sufficiently small the open set @xmath264 of the form is a common domain of convergence of all @xmath265 with @xmath266 and \(ii ) these @xmath265 are a basis of the vector space of @xmath0-hypergeometric laurent series of degree @xmath13 convergent in @xmath264 . since an @xmath0-hypergeometric series of degree @xmath13 satisfies @xmath11 independent homogeneity relations it may be viewed as a function of @xmath23 variables . to make this precise we introduce the _ gale dual _ of the configuration @xmath0 . [
def : gale ] let @xmath267 be a @xmath45-basis of the lattice @xmath256 and denote by @xmath268 the @xmath269 matrix whose columns are the vectors @xmath270 .
we shall also denote by @xmath268 the collection of row vectors of the matrix @xmath268 , @xmath271 , and call it a _ gale dual _ of @xmath0 .
[ gale - remark ] \(i ) our definition of gale dual depends on the choice of a basis of @xmath256 ; this amounts to an action of @xmath272 on the configuration @xmath268 .
\(ii ) @xmath268 is primitive , i.e. , if @xmath273 has relatively prime entries then so does @xmath274 .
this follows from the fact that if @xmath275 for @xmath276 and @xmath277 then @xmath278 .
equivalently , the rows of @xmath268 span @xmath279 .
\(iii ) the regularity condition on @xmath0 is equivalent to the requirement that @xmath280 \(iv ) @xmath0 is not a pyramid if and only if none of the vectors @xmath281 vanishes .
given @xmath282 , and the choice of a gale dual @xmath268 we may identify @xmath283 by @xmath284 with @xmath285 in particular , @xmath286 if and only if @xmath287 , where @xmath288 the linear forms in ( [ linearforms ] ) define a hyperplane arrangement oriented by the normals @xmath289 and each minimal cell @xmath246 corresponds to the closure of a certain connected components @xmath290 in the complement of this arrangement .
let @xmath291 as in ( [ eq:1.1 ] ) .
we can also write for @xmath292 @xmath293 setting @xmath294 we can now rewrite , the series ( [ eq:1.1 ] ) in the coordinates @xmath295 as @xmath296 , where @xmath297 moreover , since changing @xmath282 only changes ( [ eq:1.1 ] ) by a constant , we can assume that in order to write ( [ eq:1.2 ] ) we have chosen @xmath298 and this guarantees that @xmath299 for @xmath300 and @xmath301 for @xmath302 . if @xmath69 is an @xmath0-hypergeometric function of degree @xmath13 , then @xmath303 is @xmath0-hypergeometric of degree @xmath304 . in terms of the hyperplane arrangement in @xmath305
this has the effect changing the hyperplane @xmath306 to the hyperplane @xmath307 .
the cone of parameters @xmath308 is called the _ euler - jacobi cone of @xmath0_. we note that if @xmath309 then @xmath310 for all @xmath311 .
[ ejregions ] given a parameter @xmath13 and @xmath312 as in then @xmath313 if and only if there exists a point @xmath314 such that @xmath315 for all @xmath316 .
this implies in particular that if @xmath317 are such that @xmath318 , @xmath319 , then : @xmath320 in particular , all minimal regions @xmath290 have recession cones of dimension @xmath23 .
we also recall the following result ( * ? ? ?
* corollary 4.5.13 ) which we will use in the following sections : [ th : ejstable ] if @xmath1 is an @xmath0-hypergeometric function of degree @xmath321 then , for any @xmath322 , @xmath323 if and only if @xmath324 . in particular , all non - zero @xmath0-hypergeometric functions whose degree lies in the euler - jacobi cone are stable .
[ ex : running ] let @xmath325 be the configuration @xmath326 @xmath0 is an essential cayley configuration of two dimension one configurations : @xmath327 . for @xmath328 , the ideal @xmath14 is generated by @xmath329 , @xmath330 , @xmath331 together with the three euler operators .
one may verify by direct computation that the function @xmath332 is @xmath0-hypergeometric of degree @xmath333 .
the denominator of @xmath1 is the discriminant @xmath65 which agrees with the classical univariate resultant of the polynomials : @xmath334 a gale dual of @xmath0 is given by the matrix : @xmath335 let @xmath336 .
then @xmath337 and with respect to the inhomogeneous variables : @xmath338 we have @xmath339 the hyperplane arrangement associated with @xmath340 is defined by the five half - spaces @xmath341 , where @xmath342 , @xmath343 , @xmath344 , @xmath345 , @xmath346 .
there are @xmath347 minimal cells in @xmath247 , depicted in figure [ fig : arrangement ] .
they are all two - dimensional and correspond to the negative supports : @xmath348 , @xmath349 , @xmath350 , @xmath351 .
[ fig : arrangement ] the expansion of @xmath69 ( cf . ) from the vertex corresponding to @xmath352 in the newton polytope of the denominator of @xmath1 gives : @xmath353 similarly , the series @xmath354 and @xmath355 correspond to the expansions from the other two vertices of the newton polytope of the denominator of @xmath1 .
in this section we prove theorem [ th : conj ] classifying all codimension - two configurations @xmath0 admitting stable rational @xmath0-hypergeometric functions .
in particular we obtain a description of all gkz - rational configurations @xmath0 of codimension two .
as it has already been pointed out , it is shown in @xcite that an essential cayley configuration is gkz - rational and , therefore , admits stable rational @xmath0-hypergeometric functions .
moreover , it follows from @xcite that codimension - two lawrence configurations , while not being gkz - rational , nevertheless admit stable rational @xmath0-hypergeometric functions .
thus we need to consider the converse statement ; that is , which codimension - two configurations admit stable , rational , @xmath0-hypergeometric functions .
suppose then that @xmath356 is a stable , rational @xmath0-hypergeometric function . @xmath357 and @xmath358 are @xmath0-homogeneous polynomial and consequently , the newton polytope @xmath359 of @xmath358 lies in a translate of @xmath360 .
we choose a vertex @xmath361 of @xmath359 and a @xmath45-basis @xmath362 of @xmath363 such that @xmath364 define @xmath365 as in , @xmath178 , and let @xmath366 be an exponent occurring in @xmath357 .
then @xmath1 has @xmath0-homogeneity @xmath367 and it has a power series expansion supported in a translate of the cone @xmath368 the basis @xmath369 gives rise to a gale dual @xmath268 of @xmath0 as in section [ sec : laurent ] and we may choose @xmath370 to identify @xmath371 .
we can dehomogenize @xmath1 to get a bivariate rational function @xmath372 which verifies @xmath373 it follows from theorem [ laurentseries ] that , without loss of generality , @xmath374 where @xmath375 are minimal cells of the oriented line arrangement defined by @xmath340 contained in the first quadrant . since @xmath1 is stable , no derivative of @xmath1 vanishes and , after appropriate differentiation
, we may assume that the degree @xmath13 lies in the euler - jacobi cone @xmath70 and , consequently , that there are no bounded minimal cells of degree @xmath13 .
we can also suppose that @xmath376 is a two - dimensional pointed cone with integral vertex , which we may assume to be the origin . for each @xmath377 with @xmath378 ,
the @xmath215-diagonal @xmath220 of @xmath216 is algebraic . on the other hand , @xmath379 and
therefore , it follows from ( [ eq:1.2 ] ) that : @xmath380 where @xmath381 .
now , according to theorem [ notalgebraic ] , for all @xmath382 , the series @xmath383 is an algebraic function .
we note that , after cancellation , the coefficients of the series ( [ eq : diagonal_central ] ) no longer involve terms coming from pairs @xmath289 , @xmath281 such that @xmath384 .
we denote by @xmath385 the configuration obtained by removing all such pairs as well as any zero vector and call it the _ reduced configuration of @xmath268_. if @xmath386 then @xmath0 is clearly a lawrence configuration .
next we show that if @xmath0 admits a stable , rational hypergeometric function then @xmath387 can not be a one - dimensional vector configuration .
indeed , suppose @xmath387 is one dimensional , say , @xmath388 .
since @xmath216 is stable , the newton polytope @xmath174 is a two - dimensional polytope .
let @xmath389 be one of its vertices and let @xmath390 , @xmath391 be the adjacent edges .
we may assume without loss of generality that , say , @xmath390 is not orthogonal to @xmath392 .
consequently , it follows from lemma [ supportcone ] that the expansion of @xmath216 from the vertex @xmath389 contains infinitely many non - zero terms whose exponents lie in a ray with direction vector @xmath390 .
the restriction @xmath90 of @xmath183 to such a ray is the specialization of a suitable derivative of @xmath216 and hence a one - variable rational function . on the other hand
, @xmath90 is as in the hypothesis of theorem [ notalgebraic ] with the @xmath393 s and @xmath394 s of the form @xmath395 for some @xmath396 . by ( ii ) of the theorem these must cancel out in pairs but then since @xmath387 is one - dimensional , @xmath386 which is a contradiction . let @xmath397 by construction of @xmath387 there exists infinitely many @xmath398 with relatively prime entries such that @xmath399 has no zero coordinate and no two distinct coordinates adding up to zero .
for such a @xmath215 there is no cancellation of the factorials in the coefficients when we take the @xmath215-diagonal of @xmath216 .
therefore , since @xmath387 has rank two , by the classification of algebraic hypergeometric series in one variable ( see [ sec : univariate ] ) , there exists two pairs of linearly independent vectors @xmath400 and @xmath401 such that @xmath402 and @xmath403 for some @xmath404 .
in other words , there exists @xmath405 such that @xmath406 for some @xmath407 .
in fact , since @xmath408 is primitive , @xmath409 . now with the notation of the previous paragraph @xmath216 restricted to the rays @xmath390 and @xmath391 is rational . by inspection
we see that if @xmath410 there are no two vectors in @xmath411 which give restrictions compatible with theorem [ notalgebraic ] ( ii ) .
( i.e. , such that the coordinates of @xmath412 are of the form @xmath413 for some @xmath414 up to permutation . )
therefore @xmath415 and it is now easy to check that necessarily @xmath0 is affinely equivalent to an essential cayley configuration .
this concludes the proof of theorem [ th : conj ] .
the simplest series with associated matrix @xmath416 @xmath417 was considered by catalan .
it is an algebraic function , in fact , @xmath418 ( see @xcite for an appearance of this series in combinatorics ) .
similarly , the series @xmath419 with associated matrix @xmath420 is algebraic , in fact , @xmath421 ( the quickest way to prove these identities is to use a recursion for the coefficients . for @xmath422 for example @xmath423 where @xmath424 , and @xmath425 . )
note that in addition both @xmath422 and @xmath426 satisfy that all of their @xmath215-diagonals are algebraic .
this is not the typical case for two variable algebraic functions . for @xmath427
the natural series is @xmath428 which is of course rational @xmath429 . in (
* theorem 4.1 ) it was necessary to show that a bivariate series of the form : @xmath430 where @xmath431 are relatively prime positive integers and @xmath432 , does not define a rational function . while the proof presented in @xcite is incomplete , the result is clearly an immediate consequence of theorem [ th : conj ] .
indeed , the @xmath433 univariate diagonal series @xmath434 should be algebraic but , then by theorem [ notalgebraic ] , so should the central series @xmath435 however , this is impossible by ( * ? ? ?
* theorem 1 ) since the height of the series is @xmath436 . we should point out that the argument in @xcite was based on a correct proof for a similar case due to laura matusevich .
the gap appears in adapting that proof to the series .
the purpose of this section is to describe all stable @xmath0-hypergeometric functions in the case of codimension - two configurations . by theorem
[ th : conj ] we may assume that @xmath0 is either a lawrence configuration or an essential cayley configuration .
the first case has been studied , for arbitrary codimension , in @xcite . in particular ,
if @xmath0 is a codimension - two lawrence configuration then @xmath0 is a cayley configuration of @xmath437 two - point configurations in @xmath37 and it follows from ( * ? ? ?
* theorem 1.1 ) that the dimension of the space of stable @xmath0-hypergeometric functions is @xmath438 and that they may be represented by appropriate multidimensional residues .
we refer the reader to @xcite for details .
thus , we will restrict ourselves to the case of essential cayley configurations .
we begin by recalling the construction of rational hypergeometric functions associated with any essential cayley configuration by means of multivariate toric residues ( we refer to @xcite for details and proofs ) and will then show in theorem [ th : dim1 ] that , in the codimension - two case , a suitable derivative of any stable rational hypergeometric function must be a toric residue .
in particular , if @xmath313 , the dimension of the space of rational @xmath0-hypergeometric functions is equal to @xmath68 .
let @xmath439 be an essential cayley configuration .
for each @xmath440 consider the generic laurent polynomial @xmath441 supported in @xmath41 , that is : @xmath442 we set @xmath443 .
generically on the coefficients @xmath444 , given any @xmath445 , the @xmath52-fold intersection @xmath446 is finite and , given any laurent monomial @xmath447 , @xmath448 , we can consider the _ global residue _ : @xmath449 where @xmath450 denotes the local grothendieck residue ( see @xcite ) and @xmath451 is an appropriate real @xmath52-cycle on the torus @xmath452 .
it is shown in ( * ? ? ?
* theorem 4.12 ) that if @xmath453 lies in the interior of the minkowski sum of the convex hulls of @xmath454 then the expression @xmath455 is independent of @xmath407 .
its common value is the _ toric residue _
@xmath456 studied in @xcite .
it is often useful to consider the expression obtained by replacing in ( [ globalres ] ) the polynomial @xmath457 by @xmath458 , where @xmath459 is a positive integer .
this change defines a function @xmath460 , and if @xmath453 lies in the interior of the minkowski sum of the convex hulls of @xmath461 then the expression @xmath462 is independent of @xmath407 .
the toric residue @xmath463 is a rational function on the coefficients @xmath444 and is @xmath0-hypergeometric of degree @xmath464 .
this may be seen , for example , by differentiating under the integral sign in the expression ( [ intres ] ) .
we refer to ( * ? ? ?
* theorem 7 ) for details .
it follows from the arguments in @xcite , that the function @xmath463 does not vanish and , since for a given @xmath465 , a point @xmath448 is in the interior of the minkowski sum of @xmath461 if and only if @xmath466 lies in the euler - jacobi cone @xmath70 ( [ ej ] ) , it follows that @xmath463 is a stable rational @xmath0-hypergeometric function in this case .
we note that @xmath467 where @xmath468 denotes the @xmath407-th vector in the standard basis of @xmath469 .
the family of laurent polynomials @xmath470 associated with a codimension - two cayley essential configuration must consist of @xmath52 binomials and one trinomial .
thus , after relabeling the coefficients and an affine transformation of the exponents we may assume that @xmath471 with @xmath472 .
[ th : dim1 ] let @xmath473 be a codimension - two cayley essential configuration and suppose @xmath474 .
then , any rational @xmath0-hypergeometric function of degree @xmath13 is a multiple of @xmath463 . the gale dual @xmath268 of a codimension - two cayley essential configuration is a collection of @xmath475 vectors @xmath476 which , after renumbering , may be assumed to be of the form : @xmath477 where the vectors @xmath478 are not collinear . as in we denote by @xmath479 the linear functionals in @xmath480 defined by @xmath268 and a choice of @xmath481 such that @xmath337 . for @xmath482
, we set @xmath483 let @xmath356 be any non - zero ( and hence , stable ) @xmath0-hypergeometric function of degree @xmath13 and write @xmath484 , where @xmath485 is as in .
we claim that the newton polytope @xmath174 of the polynomial @xmath486 is a triangle whose inward pointing normals are the vectors @xmath487 , @xmath488 , @xmath489 ( this is indeed the case for the residue @xmath463 since its denominator is a power of the discriminant @xmath65 , whose newton polytope is such a triangle by @xcite ) .
let @xmath490 be a vertex of @xmath174 and @xmath362 the adjacent vertices .
set @xmath491 , @xmath178 .
the laurent expansion of @xmath183 from the vertex @xmath490 is supported in a cone of the form @xmath492 on the other hand , since @xmath183 is the dehomogenization of an @xmath0-hypergeometric function , it follows from theorem [ laurentseries ] that @xmath216 may be written as @xmath493 where @xmath494 runs over all minimal regions of the hyperplane arrangement defined by the linear functionals @xmath312 which are contained in the cone @xmath196 , and @xmath495 .
let @xmath494 be any region appearing in with a non - zero coefficient . since @xmath494 is minimal region , all linear forms @xmath496 have a constant sign in its interior .
as noted in remark [ ejregions ] , @xmath494 must have a two - dimensional recession cone .
let @xmath215 be a rational direction in the interior of @xmath494 and consider the @xmath215-diagonal of @xmath497 . by proposition [ diag ]
, the function @xmath498 must be algebraic .
as the series @xmath497 has the form below , it follows from the discussion in section [ sec : univariate ] that this can only happen if two of the linear forms @xmath499 , are positive ( and the third one negative ) over the interior of @xmath494 .
this proves that @xmath494 is contained in one of the regions @xmath500 .
now , it follows from lemma [ supportcone ] that there must be minimal regions @xmath501 , not necessarily distinct , appearing with non - zero coefficients in the expansion such that @xmath502 contains all points of the form @xmath503 , @xmath178 , for suitable @xmath504 and @xmath505 sufficiently large . consider the series @xmath506 associated to the minimal region @xmath507 .
it follows from that @xmath508 where @xmath509 is a polynomial .
but , since @xmath216 is a rational function we deduce that the univariate function @xmath510 must be a rational function .
but by item ii ) in theorem [ notalgebraic ] this is only possible if @xmath511 for some @xmath512 . as the ray with direction
@xmath390 is in the boundary of the minimal region @xmath507 , this implies that @xmath507 can not be contained in @xmath513 , where @xmath514 .
consequently , @xmath515 is an inward pointing normal to @xmath174 and our claim is proved .
given now any rational hypergeometric function @xmath516 of degree @xmath13 we may now consider the laurent expansion of its dehomogenization @xmath517 from the vertex of @xmath174 defined by the edges with inward - pointing normals @xmath487 and @xmath488 .
when that expansion is written as in there must be , by lemma [ supportcone ] , a minimal region @xmath494 whose recession cone has a boundary line orthogonal to @xmath487 and the corresponding coefficient @xmath518 must be non - zero .
hence , the map @xmath519 is @xmath520 and the space of rational @xmath0-hypergeometric functions of degree @xmath13 has dimension at most one .
as we have already recalled , it follows from @xcite that the toric residue @xmath463 is a non zero rational @xmath0-hypergeometric function of degree @xmath13 , which thus spans the vector space of all rational @xmath0-hypergeometric functions of this degree .
[ ex : running3 ] we continue with example [ ex : running ] .
let @xmath69 be as in and @xmath521 $ ] as in .
then we have @xmath522 we showed in example [ ex : running ] that in inhomogeneous coordinates @xmath523 for the minimal regions
@xmath524 contained in the first quadrant . according to theorem [
th : dim1 ] neither @xmath525 nor @xmath526 can be rational functions .
indeed , one can check by direct computation that , up to sign , @xmath525 and @xmath526 agree with the pointwise residues : @xmath527 where @xmath528 are the roots of @xmath529 : @xmath530 and , in the inhomogeneous coordinates @xmath531 , @xmath532 , we have : @xmath533
in this section we will apply the previous results to study the rationality of power series in two variables which generalize the univariate series discussed in section [ sec : univariate ] , that is , series whose coefficients are ratios of products of factorials of linear forms defined over @xmath45 .
our starting data will be a support cone @xmath196 which will be assumed to be a two - dimensional rational , convex polyhedral cone in @xmath480 and linear functionals @xmath534 where @xmath535 , @xmath536 .
we will denote by @xmath537 the primitive integral vectors defining the edges of @xmath196 and by @xmath538 the corresponding primitive inward normals .
[ rem : horn ] let @xmath541 be a horn series as in .
then , the coefficients @xmath542 satisfy a horn recurrence ; that is , for @xmath543 , and any @xmath544 such that @xmath545 also lies in @xmath196 , the ratios : @xmath546 are rational functions of @xmath547 ( recall that @xmath548 denote the standard basis vectors ) .
we are interested in studying when a horn series defines a rational function @xmath549 .
we will assume that @xmath550 and note that implies that converges for @xmath551 , @xmath552 for any small @xmath553 .
[ horn = hypergeom ] every horn series is the dehomogenization of an @xmath0-hypergeometric function for some regular configuration @xmath0 .
more precisely , there exists a codimension - two configuration @xmath554 , a vector @xmath555 , a @xmath45-basis @xmath362 of @xmath556 , and an @xmath0-hypergeometric function @xmath69 of degree @xmath557 such that : @xmath558 this may be seen as follows : the linear forms @xmath559 define an oriented hyperplane arrangement associated with the vector configuration @xmath560 and the vector @xmath561 .
we can enlarge @xmath268 to a new configuration @xmath562 by adding to @xmath268 pairs of vectors @xmath563 where @xmath564 ranges over all @xmath565 , @xmath362 , and the standard basis vectors @xmath566 , @xmath567 .
@xmath562 is the gale dual of a configuration @xmath0 and , for a suitable choice of parameter @xmath568 , @xmath569 , every region in the hyperplane arrangement defined by @xmath570 is minimal .
and the series is the dehomogenization of an @xmath0-hypergeometric series of degree @xmath571 .
[ th : conjclassic ] let @xmath572 , @xmath316 , be linear forms on @xmath480 defined over @xmath45 and @xmath196 a two - dimensional rational , convex , polyhedral cone in @xmath480 .
let @xmath573 be a horn series satisfying .
set @xmath574 .
if @xmath575 is a rational function then either * @xmath576 is even and , after reordering we may assume : @xmath577 * @xmath268 consists of @xmath578 vectors and , after reordering , we may assume that @xmath579 satisfy and @xmath580 , @xmath581 , @xmath582 , where @xmath369 are the primitive , integral , inward - pointing normals of @xmath196 and @xmath583 are positive integers . as noted in remark [ horn = hypergeom ] , the series @xmath575 may be viewed as the dehomogenization of an @xmath0-hypergeometric function , for a suitable regular configuration @xmath0 whose gale dual @xmath562 is obtained from @xmath268 by adding pairs of vectors @xmath563 , @xmath584 . since @xmath0 admits a stable rational hypergeometric function , it follows from theorem [ th : conj ] that @xmath0 is either a lawrence configuration or a cayley essential configuration .
it is now clear that in the first case , @xmath268 satisfies , while in the second , @xmath585 and we may assume that @xmath579 also satisfy , while @xmath586 moreover , if @xmath0 is a cayley essential configuration then it is shown in the proof of theorem [ th : dim1 ] that @xmath587 where @xmath588 are canonical series , as in , associated with the minimal regions of the hyperplane arrangement of @xmath562 , and the sum runs over all minimal regions contained in one of the sectors defined by the half - spaces @xmath589 , @xmath590 , @xmath591 .
but then , since the expansion is not supported in any proper subcone of @xmath196 , it follows that @xmath196 must agree with one of those sectors .
hence , after reordering if necessary , we have that @xmath592 the series @xmath595 is a horn series .
it follows from theorem [ th : dim1 ] that @xmath596 may be represented as a residue .
indeed , following the notation of theorem [ th : conjclassic ] , the configuration @xmath268 is defined by the vectors @xmath597 , @xmath598 , @xmath599 .
we enlarge it to a configuration @xmath562 by adding the vectors @xmath600 , and @xmath601 .
now , @xmath562 is the gale dual of the cayley essential configuration @xmath602 and @xmath596 is the dehomogenization of an @xmath0-hypergeometric toric residue associated to @xmath603 .
explicitely , in inhomogeneous coordinates we have : @xmath604 where @xmath605 runs over the three cubic roots of @xmath606 ; that is , @xmath607 is the global residue with respect to the family of polynomials @xmath608 of the rational function of @xmath222 ( depending parametrically on @xmath295 ) defined by @xmath609 .
the assertion is evident if either @xmath612 or @xmath613 since in this case becomes : @xmath614 as well as in the case when @xmath615 since @xmath616 more in general , given any @xmath617 , consider the cayley essential configuration : @xmath618 and @xmath619 . consider the hyperplane arrangement associated with the vector @xmath620 and the gale dual @xmath268 of @xmath0 with rows @xmath621 , @xmath622 , @xmath623 .
the first quadrant is a minimal region and the corresponding laurent @xmath0-hypergeometric series is : @xmath624 thus , the rationality @xmath1 implies that of @xmath625 .
but , since the first quadrant is the only minimal region contained in the open half space @xmath626 and @xmath0 is a cayley essential configuration , the series @xmath69 must agree with a laurent expansion of the toric residue @xmath627 which is a rational function .
in fact , we can write explicitly : @xmath628 write @xmath634 , where @xmath635 are relatively prime polynomials in @xmath636 $ ] . for any @xmath637 such that @xmath638 we have @xmath639 hence @xmath640 for some non - zero constant @xmath641 .
since @xmath642 is nonzero ( as we assume @xmath216 is holomorphic at the origin ) evaluating at @xmath643 shows @xmath644 and the result follows .
our last result shows that , up to the action of differential operators of the form [ eq : p ] below , all rational horn series with support on all the integer points of the first quadrant are given by the functions @xmath657 .
then , there exist differential operators @xmath659 of the form @xmath660 such that @xmath661 where @xmath662 in case @xmath663 is a lawrence configuration and @xmath664 if @xmath268 is cayley essential .
it follows from theorem [ th : conjclassic ] that @xmath268 must be either a lawrence or a cayley essential configuration . in the latter case
, we have moreover that @xmath585 , @xmath579 are as in while @xmath665 , @xmath666 , @xmath667 for @xmath583 positive integers .
therefore , we can find a differential operator @xmath668 as in such that @xmath669 for suitable integers @xmath670 .
thus taking @xmath671 we get @xmath672 the argument in the lawrence case is completely analogous .
we return to the rational function in example [ ex : gessell ] : @xmath673 and its laurent expansion .
let @xmath674 ( so that @xmath675 ) then @xmath676 where @xmath162 denotes the sublattice @xmath677 and @xmath678 .
thus , we get an expansion similar to that of @xmath629 but the sum is only over the points in the first quadrant that lie in the sublattice @xmath162 of index @xmath679 rather than all of @xmath680 .
f. rodrguez villegas , hypergeometric families of calabi - yau manifolds , calabi - yau varieties and mirror symmetry ( toronto , on , 2001 ) , 223231 , fields inst .
, 38 , amer .
soc . , providence , ri , 2003 .
v. i. vasyunin , , zap .
nauchn . sem .
otdel . mat .
inst . steklov .
( pomi ) * 262 * ( 1999 ) , no . issled .
po linein .
27 , 4970 , 231232 , translation in j. math .
sci.(new york ) 110 ( 2002 ) , no . 5 , 29302943 .
|
we describe the structure of all codimension - two lattice configurations @xmath0 which admit a stable rational @xmath0-hypergeometric function , that is a rational function @xmath1 all whose partial derivatives are non zero , and which is a solution of the @xmath0-hypergeometric system of partial differential equations defined by gelfand , kapranov and zelevinsky .
we show , moreover , that all stable rational @xmath0-hypergeometric functions may be described by toric residues and apply our results to study the rationality of bivariate series whose coefficients are quotients of factorials of linear forms .
|
despite the fact that ultra - high energy ( uhe ) cosmic rays are known for decades , their sources and the acceleration mechanisms are still under debate .
sources are only detectable by @xmath1-rays produced in interactions near the sources . in the very - high energy ( vhe )
range near 1tev more and more @xmath1-ray sources are revealed by the imaging cherenkov technique . on the contrary ,
no clear source detections have been made in the uhe domain above about 100tev , except perhaps a few episodic cases .
mainly for reasons of the required power , the dominant sources of cosmic rays up to about 100tev and probably up to the _ knee _ of the cosmic - ray energy spectrum at a few @xmath0ev are believed to be supernova remnants in the sedov phase .
the change of the spectrum near the knee presumably reflects a change in the origin and the takeover of another , yet unclear type of sources at energies above the knee .
a change in the cosmic - ray propagation with a decreasing galactic containment has also been considered . in either case , the change in the slope of the spectrum should be accompanied by a change in the mass composition of cosmic rays . in the case of a change of sources across the knee ,
the composition could change dramatically .
a much less spectacular change of the composition is expected in the case of decreasing containment .
direct measurements well below the knee @xcite show indeed a substantial change in the mass composition already at energies around 100tev .
in particular , the fraction of protons seems to diminish with increasing energy . due to their small collection areas
the balloon - borne direct experiments run out of statistics above several hundred tev .
indirect methods using ground - based experiments are , so far , not able to classify individual cosmic rays unambiguously by their mass .
such methods are more appropriate for evaluating some average mass number .
most notably , the results of the fly s eye group @xcite indicate a rather heavy composition well above the knee .
if taken at face value , their results represent a composition of mainly very heavy nuclei , like iron , at @xmath2ev .
experiments measuring the composition right at the knee of the spectrum obtained either ambiguous or even conflicting results .
results with no significant change @xcite or a slight increase of heavy elements @xcite have been reported .
other groups found more significant increases of heavy elements @xcite or , on the other hand , predominantly protons @xcite .
further measurements are needed , if the cosmic - ray composition should help to resolve the question where cosmic rays are accelerated .
the method proposed in this paper should be relatively easy to implement at sites where an air - shower array with an angular resolution below one degree already exists .
the method is mainly a measurement of the longitudinal shower development by the angles of muons with respect to the shower axis .
it does not require to measure many muons in a single shower because it uses only the _ inclusive _ angular distributions , nor is accurate timing required .
indeed , if several muons are measured in one event , they are treated separately .
not only the average muon angles but also the average electron angles with respect to the shower axis are sensitive to the primary composition . because the detector response is , in general , better understood for muon tracks than for electron tracks , this paper focuses mainly on the muons . using the tracking detectors and the air - shower array , the method can be supplemented by traditional methods like the average @xmath3 ratio or the muon lateral distribution .
although one can think of complex experiments dedicated to measuring the cosmic - ray composition , like kaskade @xcite , where more pieces of information are collected , the purpose of this paper is to demonstrate that even the muon angles alone provide significant information about the composition .
the proposed method relies on shower simulations to obtain an absolute value of an average mass number , just like other indirect methods . if suitable tracking detectors are used , it has the advantage that essentially all required detector parameters for the simulation can be obtained from the measured data as a function of shower size
. therefore , it can be particularly sensitive to changes in the composition across the measured shower size spectrum , even if a comparison with simulations using different interaction models may yield different absolute values .
an analysis of data taken with ten crt detectors tracking detectors of 2.5m@xmath4 sensitive area each @xcite and the hegra air - shower array @xcite on la palma is in progress @xcite . in the data of crt and hegra both muon and electron tracks
are analysed .
although the intention of this paper is to promote the method to other air - shower experiments , the simulations shown in this paper are specific to the combination of the crt and hegra experiments .
these simulations take into account the response of both components the tracking detectors and the air - shower array in much detail .
after an outline of the method , the specific simulations are described to demonstrate that the method can be very well applied with existing detector technology .
experimental requirements for application of the method and limitations by shower simulations are shown in a more general context , not specific to crt and hegra .
first measurements of the angles of muons with respect to the corresponding air showers or , in alternative terms , of the apparent height of origin of the muons were done in the 1960s @xcite . at that time neither the angular resolutions of air - shower arrays nor the accuracy of shower simulations were appropriate to use such measurements for investigating the mass composition of primary cosmic rays . for an early comparison of measured and simulated height of muon origin see @xcite .
the proposed method is based on the fact that showers initiated by heavy primaries , like iron nuclei , develop on average at smaller atmospheric depths , i.e. higher altitudes , than proton showers .
muons from showers of heavy primaries arrive , therefore , at smaller angles with respect to the shower axis for any given core distance .
the proposed method is not intended to measure individual primaries but to measure accurately an average number of the composition , like @xmath5 , the average logarithm of the primary mass numbers .
measuring many tracks in a single shower is , therefore , not necessary and only helps to improve statistics .
= 0.6 the angles of muons or other particles with respect to the shower axis are best expressed in terms of their _ radial angle _ and _ tangential angle _ components .
the radial angle is the projection onto the plane defined by the geometry of the shower axis and the location of the muon detector . throughout this paper , a negative radial angle
is assigned to a particle flying away from the shower axis .
in fact , at large core distances most particles have such a negative radial angle . the tangential angle ( also called the transverse angle )
is the projection onto the plane perpendicular to the ` radial ' plane and parallel to the shower axis ( see figure [ fig : geometry ] ) . for almost vertical showers a more convenient definition with projections into vertical planes defined only by the positions of the shower core and
the detector is essentially equivalent .
the radial projection can be used to derive an _ apparent height of origin _ of a particle ( figure [ fig : geo - height ] ) .
although radial angle and apparent height of origin are almost equivalent , the radial angle is the more robust variable .
this is due to the fact that a small scattering or experimental smearing around zero radial angle corresponds to @xmath6 apparent height .
in addition , the derived height for each muon depends directly on the resolution of the core position .
in particular , the last point turns out to be a major drawback of an analysis in terms of the apparent height .
the core position resolution is usually not very precisely known as a function of shower size , zenith angle , and position within the array .
an analysis in terms of the apparent height is , for this reason , limited to core distances much larger than the core position resolution .
= 0.6 the distribution of tangential angles ( see figure [ fig : tang+rad ] ) is symmetric and its width is dominated by multiple scattering and the lateral distribution of the parent particle generation . for the muons in uhe air showers
the intrinsic distribution is very narrow , apart from tails due to some low - energy muons .
this distribution shows little difference between proton- and iron - initiated showers , even for an ideal angular resolution . under certain experimental constraints
the tangential angle can be used to measure the combined angular resolution of air - shower array and tracking detectors .
= 0.8 the radial angle , representing the longitudinal shower development , is the quantity sensitive to the composition .
the _ average _ ( which may be mean , median , or mode ) radial angle depends in a characteristic way on the core distance . to compare the simulated average radial - angle curves with measurements requires to know the combined angular resolution of the air - shower array and tracking detectors and to some extent also the core position resolution of the array
however , the average radial angle at any given core distance changes very little with shower size or zenith angle .
the figures in this paper are intended as examples .
they are based on simulations using the corsika code ( see section [ sect : shower+det - sim ] ) . the simulations assume an experimental resolution as observed for crt detectors at the hegra array @xcite . see figure [ fig : rad - ang - primaries ] for the median radial angles of different primary types . at any core distance
the average radial angle is almost a linear function of @xmath5 , where @xmath7 is the mass number of the primary .
= 0.8 although a similar difference between showers initiated by protons or iron nuclei is also present in the radial - angle curves of electrons , the electron component is less usefull for composition studies than muons .
first of all , a very detailed simulation of the behaviour of the tracking detectors to electrons , gammas , muons , and hadrons is required to compare with experimental data . for muons
the detector is generally much better understood , although punch - throughs of other particles have to be considered .
the second reason in favour of muons is based on the shower selection introduced by the air - shower array . to achieve the same shower size ( @xmath8 ) at ground level ,
showers initiated by heavy primaries require a larger primary energy than proton showers .
due to the steep spectrum , conventional air - shower arrays will , therefore , mainly see proton showers at any shower size , even if protons account for only half of the mass composition at a given energy per particle .
muons can almost compensate for this bias because , at the same shower size , a proton shower contains less muons than a shower of a heavy primary . because of the compensation of the selection bias by the larger number of muons for heavy primaries the quantity @xmath9 is even for a mixed composition essentially proportional to @xmath5 : @xmath10 in this context @xmath11 is the measured ( or simulated ) median radial angle in the core distance interval @xmath12 , @xmath13 and @xmath14 are the expected median values for pure protons and pure iron nuclei from the simulations , and the weights @xmath15 take the statistical accuracy of measured as well as simulated average values into account . in the simplest possible formulation , ignoring statistical errors of the simulation and correlations of errors in different radial bins and taking into account only @xmath16 as the measured accuracy of @xmath11 , @xmath17 for compositions as reported by direct measurements , equation [ eq : lna - lambda ] holds to better than 0.05 at all investigated shower size intervals at least for crt and hegra to which the simulations correspond .
although this relation is quite coincidental , it is expected to hold as well for most other existing air - shower arrays in combination with tracking detectors of a low energy threshold for muons . in the case of extreme compositions like pure helium ,
however , the relation may be wrong by up to 0.26 in terms of @xmath5 .
= 0.8 = 0.8 figure [ fig : mu - rad - ang - composite ] shows that the median ( or mean ) radial angle of muons is indeed sensitive to changes in the composition as indicated by direct measurements @xcite below the knee .
even the electron radial angles can be used for that purpose ( see figure [ fig : elec - rad - ang - composite ] ) but a more accurate knowledge of the detector response is required in that case .
in addition , @xmath18 ( defined in analogy to @xmath19 ) , is not proportional to @xmath5 but is more sensitive to the fraction of light nuclei than to heavy nuclei .
the scope of this paper is beyond the measurements carried out with crt detectors and the hegra array .
although the simulations presented here serve only as examples , experimental effects of crt detectors and the hegra scintillator array are carefully taken into account .
it may be instructive to see which of these effects turn out to be the most relevant ones .
the corsika simulation code @xcite was used in version 4.068 for all shower simulations presented in this paper . except for tests of the accuracy of the shower simulations
( see section [ sect : shower - accuracy ] ) the venus hadron - nucleus and nucleus - nucleus interaction code @xcite was used for high - energy hadronic interactions , gheisha @xcite for hadronic interactions below 80 gev , and egs4 @xcite for the electromagnetic component all in the framework of corsika .
the simulations take also the geomagnetic field at la palma into account .
a total of 5580 showers ( 1860 proton showers and 930 showers for each of the other four types of primaries , he , n , mg , and fe ) were generated isotropically in the zenith angle range 0@xmath20 to 32@xmath20 .
the energy ranges were selected to cover approximately the same shower sizes , from 20 tev to 2 pev for protons increasing to 40 tev to 4 pev for iron nuclei .
showers were generated with a @xmath21 differential energy spectrum to have sufficient numbers of showers at all energies and weighted by @xmath22 for the analysis .
the analysis was restricted to shower sizes where contributions from showers outside the simulated energy ranges are negligible . concerning the crt detectors @xcite
, it should be mentioned that muons are identified as pairs of tracks in two drift chambers , one above and one below a 10 cm thick iron plate .
the measured angles between the two tracks of an identified muon are required to be less than 2.5@xmath20 in two perpendicular projections .
electrons are identified as non - muon tracks in the upper drift chamber .
results of very detailed simulations of the response of crt detectors to various types of particles ( electrons , gammas , muons , pions , and protons ) as outlined in @xcite and described in detail in @xcite and the measured detector performance @xcite were carefully parametrized .
this includes detection efficiencies and angular resolutions as functions of particle energies , zenith angles , and track densities
independent for each particle type and for the identification as an electron track and as a muon track . as a consequence ,
the simulations include not only punch - through of electrons but also of hadrons .
it should be noted that the energy dependence for detecting genuine muons is fully described by multiple scattering and energy loss in the iron plate @xcite .
crt detectors are mainly sensitive to muons above about 1 gev .
electrons , on the other hand , are detected above some 10 mev .
the effect of multiple scattering in the detector container and the iron plate are included in the angular resolutions .
for a detailed simulation of the hegra array see @xcite .
for the simulations presented here , a much simpler approach is sufficient because the simulations are restricted to shower sizes where the hegra array is fully efficient , independent of shower age , core position , and zenith angle .
the resolutions for shower sizes and core positions as obtained by the detailed simulations were parametrized .
the hegra angular resolution is implemented in the simulations in such a way that the combined angular resolution of crt and hegra as measured by the muon tangential angles as a function of shower size and zenith angle @xcite are fully reproduced .
it may be instructive to note that the median radial - angle curves presented in this paper are not significantly altered by modifying the assumed angular resolution within the experimental limits .
modifying the assumed core position resolution has an impact on the median radial angles only within a few ten meters from the core .
punch - throughs , although included in some detail , turned out to be of minor importance with crt detectors , due to their excellent punch - through rejection .
the most important experimental parameters in the simulations are the energy dependence of the detection efficiencies and the combined angular resolution of tracking detectors and air - shower array . in the case of crt detectors and the hegra array both are very well known for muons and even sufficiently well known for electrons .
it appears very reasonable that the response of other types of tracking detectors and other air - shower arrays can also be understood well enough to take advantage of the radial - angle method .
however , a few experimental requirements as outlined in section [ sect : requirements ] should be observed for that purpose .
apart from the detector response , a major uncertainty in the interpretation of any indirect measurement of the cosmic - ray mass composition is due to the limited accuracy of the shower simulations . despite much progress in recent years ,
the nucleus - nucleus and hadron - nucleus interaction models in air - shower simulations lead to non - negligible theoretical uncertainties .
the corsika simulation code attempts a particularly detailed and accurate shower simulation . in the version 4.068 ,
as used for this paper , it offers several independent interaction models .
= 0.8 results for different interaction models , using the venus and gheisha options on one side and dpm on the other side , and for different fragmentation ( evaporation ) of the projectile nuclei have been compared in the energy range @xmath23@xmath0ev .
systematic differences in the average shower sizes of up to 20% are found and also differences in the shower fluctuations .
if the median radial - angle curves are compared in the same ranges of shower sizes , the differences between models are very small within core distances up to about 200 m , increasing to about 10% of the proton - iron difference at a core distance of 400 m ( figure [ fig : rad - ang - syserr ] ) .
that increase of systematic uncertainties is due to the fact that hard interactions resulting in high-@xmath24 secondaries are not yet included very accurately . with all interaction models
the shower - size dependence of the radial - angle curve is small .
a 30% systematic error in the relation of simulated to measured shower sizes would result only in a systematic error of the median radial angles of 3% of the proton - iron difference .
moreover , this dependence is essentially the same for different models .
another uncertainty in the longitudinal shower development is due to the sparse particle physics data in the very forward rapidity region at high energies . until more such data is available and incorporated into the interaction models it seems desirable to have also indirect composition measurements in an energy range overlapping with direct measurements .
when searching only for a change of the composition near the knee of the cosmic - ray spectrum , the poor knowledge of the very forward region should not be a limitation .
different parametrizations of parton densities which pose problems for shower simulations at extremely high energies @xcite are not of concern near @xmath0ev primary energy .
to exploit the described method for composition studies several experimental requirements should be met .
these are requirements for the _ particle identification _ with a tracking detector and the _ resolution _ of angles and positions . because the particles of the electromagnetic shower component , @xmath25 and @xmath1 , have angular distributions much different from those of muons , a good electron rejection is important .
hadrons only matter very close to the shower core . due to the broad muon lateral distribution
it is generally not such a problem to identify muons at large core distances but more of a problem to reject random , non - shower muons .
this can be easily solved by a timing information of a few hundred nanoseconds accuracy and by the angular correlation of shower muons with the shower axis .
another requirement for the muon identification is due to the fact that the radial angles are well correlated with the muon momentum , with higher - energy muons usually coming from further up in the atmosphere and being better aligned with the shower axis .
the muon identification should , therefore , be well understood also as a function of momentum and zenith angle . for a high muon - energy threshold ,
a very good angular resolution is required , while for a low threshold the multiple scattering and magnetic deflection of muons can not be neglected .
the combined angular resolution of air - shower array and tracking detector should be better than about 1@xmath20 in each projection but has to be known as a function of shower size to compare measurements with simulations .
definitely sufficient would be a combined resolution of some 0.5@xmath20 if the muon - energy threshold is as low as about 1 gev
. this would be difficult to achieve with a tracking detector entirely below thick shielding material which also scatters the muons .
a good uniformity of the resolution over azimuth angles would be of advantage for the important detector calibration . only with a uniform resolution
can the angular resolution be derived from the tangential angle distribution of the same data set which is used to compare the radial angles with simulations .
otherwise , shower - size dependent systematic errors can arise .
the alignment of the tracking detectors with respect to the reference frame as defined by showers reconstructed from array data should be known to better than about 0.1@xmath20 .
this may require some careful calibrations but is certainly feasible .
the core position resolution which also enters into the simulations has to be determined from the air - shower array alone . keeping selection effects of the air - shower array as small as possible demands to select only showers well contained in the array and with shower sizes well above the trigger and reconstruction thresholds . for many of the existing air - shower arrays that would correspond to threshold energies of some 80200tev , just below where direct measurements are running out of statistics .
reasonable errors in the shower size and energy calibrations of the array do not compromise the method as long as the same experimental resolution is used in the simulations as seen in the data to be compared with these simulations . with the 25 m@xmath4 sensitive area of the crt detectors the statistics with data of about one year
would be the limit for primary energies above several @xmath26 ev .
a few hundred square meters area would be sufficient in order to achieve some overlap with the fly s eye composition measurements .
all these requirements can be fulfilled with present technology for large - area tracking detectors and existing air - shower arrays .
although simultaneous precise ( nanosecond ) timing information of the muons could supplement the described method ( see for example @xcite ) , precise timing is not required and might be difficult to accomplish with large - area detectors .
the average radial angle of particles are sensitive to the cosmic - ray mass composition .
muons have , among other particles in air showers , many advantages .
their intrinsic angular distribution is very narrow and , thus , the tangential - angle distribution can be used to calibrate the radial - angle resolution as a function of shower size .
muons can be distinguished from other particle types very well .
the larger number of muons in showers initiated by heavy primaries compensates for the @xmath8 shower selection bias which favours light primaries . like any other indirect measurement of some average of the cosmic - ray composition ,
that method requires comparison of measured data with simulations .
compared to many other indirect methods , the muon radial - angle method has two major advantages : first , the most important detector effects to be included in the simulation can be measured very well and , second , systematic errors , for example due to the interaction model , can be easily checked by comparing measurements at low energies ( a few hundred tev ) with simulations for directly measured compositions as in @xcite . despite some systematic uncertainties in the shower simulations an average mass number ( @xmath5 )
can be derived and compared with direct measurements well below the knee .
regardless of that , the muon radial - angle method can be very sensitive to changes in the composition .
the radial - angle method would not require a very large experimental effort if existing air - shower arrays are supplemented with suitable tracking detectors .
|
measuring the angles of muons and electrons in air showers is proposed as a method for studying the primary cosmic - ray mass composition near the knee of the cosmic - ray energy spectrum at a few @xmath0ev .
conventional tracking detectors at existing air shower arrays could serve this purpose , like the crt detectors at the hegra array .
when the average radial muon angles are examined as a function of shower core distance , the experimental resolution can be very well calibrated from the tangential angle distribution .
the method is particularly promising for measuring changes in the average mass number of the primary cosmic rays with energy .
the method is described and experimental and theoretical constraints are discussed .
|
quantum technology protocols exploit the unique properties of quantum systems to achieve communication , computing and metrology tasks that are impossible , inefficient or intractable for classical systems @xcite . in many cases the distribution of entanglement ,
correlations between subsystems that exceed those possible for classical systems , is a necessary condition for quantum technology protocols to succeed .
however , entanglement is fragile and can easily be degraded by the communication or storage of the entangled quantum systems .
one solution to this problem is entanglement distillation @xcite . given an ensemble of weakly entangled quantum states , distillation techniques allow one to select or distill a smaller sub - ensemble of quantum states that are more strongly entangled .
this can be achieved using only local operations and classical communication . in this way
strong entanglement can be established between remote locations under conditions where it would be impossible without distillation such as losses that are common in quantum communication channels .
there are two broad classes of quantum optical technology protocols : those using quantum observables with a discrete spectrum , such as the spin of an electron , and those using quantum variables with a continuous spectrum , such as the position and momentum of a harmonic oscillator @xcite .
our focus here is distillation of continuous - variable ( cv ) states .
the primary entangled resource in cv systems is the two - mode squeezed vacuum state , also known as the einstein - podolsky - rosen state @xcite because its idealized version was introduced by epr in the early days of quantum mechanics to illustrate quantum nonlocality .
the epr state can be used to implement many quantum protocols including continuous versions of teleportation and quantum key distribution @xcite .
an advantage of the cv approach to quantum communication is its universality : it is capable of transmitting arbitrary states of light , in contrast to the single - photon subspace of the hilbert space to which the discrete method is limited .
further , homodyne detection used in cv systems does not require expensive single - photon detectors and is more tolerant to ambient light .
additionally , in contrast to their discrete - variable counterparts , high - quality epr states are readily available on demand at a high rate from parametric amplifiers . on the other hand , those communication protocols that use single photons do not need a special distillation procedure to counter the effect of the losses .
this is because if a photon is lost , it is not registered by the detector , so a loss event is automatically eliminated from further analysis . in cv protocols , quadrature detection occurs independently of the losses , so recovering an entangled resource suitable for use in a teleportation or repeater protocol requires a dedicated distillation step . in this paper we present experimental results demonstrating the distillation of optical cv entanglement in two settings : for very low initial squeezing and after transmission through a lossy channel . in the second
setting , we directly observe an entanglement strength of our distilled state that exceeds anything possible via deterministic transmission of the states through the same channel .
that is , even if a perfectly pure , infinitely entangled epr state was passed through that channel , the resulting entanglement would be inferior to what we observe for our distilled state .
we will refer to this as breaking the deterministic bound .
our protocol relies on the technique of noiseless linear amplification ( nla ) @xcite , in contrast to previous cv entanglement distillation demonstrations based on photon subtraction @xcite .
photon subtraction is unable to enhance entanglement in the epr state by more than a factor of two , which is by far not sufficient to compensate for a loss occurring in a typical communication line .
nla does not suffer from this limitation , and in principle allows restoring the entanglement to a macroscopic level after an arbitrarily high loss @xcite .
it is this feature of nla that enables us to break the deterministic bound .
it represents a major step forward in realizing protocols that can enhance quantum technologies under practical conditions .
a key feature of our experiment is that heralded , free propagating distilled epr states are produced by our protocol .
this is in contrast to experiment @xcite , where nla has been emulated by postselecting the quadrature measurements of an asymmetrically attenuated epr state .
this virtual distillation protocol can not herald free - propagating epr pairs and hence can not be used for quantum teleportation or repeaters .
the principal scheme of our setup is shown in the inset to fig .
we explain the idea of our method in the practically relevant case of low initial squeezing and high losses .
the initial epr state prepared in a pair of modes @xmath0 and @xmath1 is then given by @xmath2 where we have limited the analysis to the first order of the squeezing amplitude @xmath3 .
a loss channel with amplitude transmissivity @xmath4 applied to mode @xmath1 will degrade the non - vacuum component of this state , transforming it into a mixture @xmath5 we then subject this state to nla .
in contrast to previous nla experiments @xcite , where either the method of generalized quantum scissors @xcite or coherent superposition of photon subtraction and addition was used , here we employ a technique known as quantum catalysis @xcite . the mode to be amplified is reflected from a low - reflectivity beam splitter , on which it is brought into interference with an ancillary single photon .
the amplification event is heralded by detecting a single photon in the other output port of that beam splitter ( fig .
[ f1 ] , inset ) .
this will happen if either ( a ) the ancilla photon is reflected ( probability amplitude of this event equals the amplitude reflectivity @xmath6 of the beam splitter ) or ( b ) a photon from the epr state is transmitted ( this occurs with probability amplitude @xmath7 ) . in the latter case , the ancilla photon is likely to transmit through the beam splitter into the reflected channel of the epr state .
if these possibilities are made indistinguishable by proper mode matching of the ancilla photon and the epr state , their amplitudes add coherently . as a result
, the epr state is transformed into @xmath8 the second term in the above state is of the higher order in @xmath9 , @xmath10 , and @xmath11 than the second term , so the final state is well approximated by pure superposition @xmath12 we see that the reduction of the single - photon component due to low initial squeezing and/or loss in the communication is compensated by the reduction of the vacuum part due to the catalysis .
this is equivalent to nla with an amplitude gain of @xmath13 .
the amount of entanglement to be recovered depends on the ratio @xmath14 and reaches one ebit at @xmath15 .
if we instead aim to optimize the two - mode squeezing of the resulting state , this goal is reached at a slightly lower gain of @xmath16^{-1}$ ] resulting in a two - mode squeezing factor of @xmath17 with respect to the standard quantum limit .
the precise value of the optimal gain may vary dependent on the efficiency of the ancilla photon source as well as the chosen continuous - variable entanglement measure .
the bottom line is , though , that however high the loss , and however low the initial squeezing may be , one can choose the nla gain to recover a macroscopic amount of entanglement in the distilled state @xcite .
the epr state ( fig . [ f1 ] ) is prepared in a periodically poled potassium titanyl phosphate crystal ( ppktp@xmath18 ) , where the type ii parametric down - conversion is taking place .
the crystal is pumped with frequency - doubled pulses at @xmath19 nm , generated by a ti : sapphire laser with a repetition rate of 76 mhz and a pulse width of @xmath20 ps .
we spatially split the outgoing orthogonally - polarized light modes on a polarizing beam splitter .
each mode is subjected to balanced homodyne detection @xcite .
the phase of one of the local oscillator phases is varied by a piezoelectric transducer ( pzt ) .
this allows us to obtain a quorum of the two - mode quadrature measurements and therefore completely characterize the state .
we observe the variances in the difference and sum of the position quadratures of the initial state to be @xmath21 and @xmath22 , measured in units of the vacuum state level [ fig .
[ f2](a ) ] .
this corresponds to the squeezing parameter of @xmath23 .
prior to distillation , we degrade the initial state s entanglement by two alternative methods .
first , we turn the polarization of the pump beam by @xmath24 , resulting in the reduction of the squeezing parameter @xmath9 by a factor of @xmath25 .
this reduces the epr correlations , bringing the difference and sum quadrature noise levels to @xmath26 and @xmath27 , respectively .
the second method consists of introducing asymmetric attenuation imitating propagation of mode 2 through a lossy optical channel with an intensity transmissivity of @xmath28 .
then the difference and sum quadrature noise levels become @xmath29 and @xmath30 , respectively . such low squeezing levels
could not be reliably measured by homodyne detection , so the variance values quoted above are based on the measured initial squeezing and the known attenuation factor .
the single photon state required for the nla is obtained from parametric down - conversion in another crystal , ppktp@xmath31 , at a rate of 50100khz , heralded by the clicks of a single - photon counting module ( spcm@xmath18 ) @xcite .
the variable beam splitter that enacts the nla is implemented by a combination of two polarizing beam splitters and a half - wave plate between them .
one output of that beam splitter is directed onto another photon detector ( spcm@xmath31 ) while the other one is subjected to homodyne detection .
the distillation events correspond to both spcms firing synchronously , and occur at a rate of @xmath32 . to characterize the distilled state ,
the quadrature data from both detectors should be accompanied by knowledge of the sum @xmath33 of the local oscillator phases for each data sample @xcite .
we acquire this information by utilizing the fraction of mode @xmath1 of the original epr state that is reflected from the nla beam splitter towards the homodyne detector in the absence of the heralding event .
this field exhibits phase - dependent quadrature correlations with mode @xmath0 .
because it is supplied at the master laser repetition rate , we observe a sufficiently clear correlation signal to retrieve @xmath33 at any moment in time .
it is the availability of this phase evaluation technique that has motivated us to choose the catalysis scheme in favor of the more common scissors scheme for the nla . to mitigate the degradation of this reference signal by the loss channel
, we implement that channel using an electro - optical modulator ( eom ) followed by a half - wave plate .
the eom is controlled by a square wave of a 0.2 s period ( during which the phase is assumed to vary linearly ) and a duty cycle of 0.35 .
the off " period , when the transmission through the eom is maximized , is used to measure @xmath34 . during the on " period ,
the eom voltage is set to attenuate the propagating signal , and the quadrature samples associated with the distilled epr state are acquired .
our experimental results are summarized in fig . [ f2 ] . part
( a ) shows samples of the raw data for the original and distilled states while the left and right columns in ( b ) and ( c ) correspond to the distilled states after applying the two methods of degrading the entanglement .
panel ( b ) shows the variances of the sum and difference of the position quadratures ( the behavior of the momentum quadrature is similar to that of the position ) .
the amount of recovered squeezing is maximized at @xmath3567 for the first method and @xmath351012 for the second , which is consistent with the theoretical expectation in the ideal case @xmath16^{-1}$ ] . in both cases ,
the two - mode squeezing of the state recovered by distillation is at the same level as the initial state , and dramatically higher than that of the epr state with reduced squeezing , or the attenuated epr state before distillation .
to quantitatively estimate the degree of entanglement at different stages of our experiment , we use the duan _
inseparability parameter @xcite , defined as @xmath36 for all our data sets , we found the optimal value of @xmath37 to be close to unity , resulting in the plot shown in fig .
[ f2](c ) .
the value of this parameter in the distilled state is a factor of about five greater than that in the degraded epr states . to our knowledge , this factor of entanglement increase achieved in a distillation protocol is higher than anything reported to date , not only in the continuous- , but also in the discrete - variable domain @xcite . for distillation
after asymmetric loss , we can also compare this parameter with that of the perfect epr state that has experienced the same loss , in which case @xmath38 @xcite .
the entanglement observed in the distilled state is consistently stronger than this benchmark , demonstrating that our method is able to overcome the deterministic bound .
it is instructive to estimate our distillation procedure by finding a hypothetical equivalent " epr state that , after appropriate attenuation of both channels , would exhibit the same quadrature statistics as our distilled state @xcite .
we found the squeezing parameter of that state and the equivalent efficiency of the loss channel in mode @xmath1 to be , respectively , about two and three times greater than those in the actual experiment .
this illustrates the capability of our procedure to largely compensate for the losses and low initial squeezing @xcite .
the cv entanglement distillation procedure demonstrated here allows us to compensate for the low initial squeezing and losses in the transmission channel , however significant these detrimental factors may be .
for the first time , we have broken the deterministic bound for the distribution of cv entanglement .
these properties make our method promising for the construction of a cv quantum error correction protocol @xcite and hence ultimately a cv quantum repeater .
however , our results also demonstrate the limitations in the amount of entanglement that can be recovered .
there are two reasons giving rise to these limitations .
first , the single - channel linear - optical scheme implemented here constitutes a good approximation to nla only for output states with mean photon numbers much less than one , thereby imposing a fundamental limit on the amount of squeezing and entanglement attainable through the distillation .
second , we are not approaching even that limit because of imperfect preparation of the epr and the ancilla photon , mode matching on the nla beam splitter and detection inefficiencies .
the first of the above issues can be resolved by switching to multichannel nla as envisioned in the original proposal @xcite . addressing the second point
improving the experimental efficiencies is possible by a number of techniques . in particular , switching from the pulsed to continuous - wave domain would allow eliminating the intrinsic efficiency issues associated with pulsed squeezing @xcite and enable significant preparation and detection efficiency improvements @xcite .
however , before these measures are to be undertaken , it is important to theoretically understand the requirements for a distillation device that would enable construction of a practical continuous - variable quantum repeater as was done for the discrete - variable case @xcite .
this understanding will clarify if and how the method presented here must be improved in order to qualify for practical applications .
99 m. nielsen and i. chuang , _ quantum computation and quantum information _ , ( cambridge university press , cambridge , uk 2000 ) .
c. bennett _ et al .
_ , purification of noisy entanglement and faithful teleportation via noisy channels .
lett . * 76 * , 722(1996 ) .
s. l. braunstein and p. van loock , quantum information with continuous variables , rev .
* 77 * , 513 ( 2005 ) .
m. d. reid , demonstration of the einstein - podolsky - rosen paradox using nondegenerate parametric amplification .
a * 40 * , 913(1989 ) . c. weedbrook , s. pirandola , r. garcia - patron , n. j. cerf , t. c. ralph , j. h. shapiro , s. lloyd , gaussian quantum information , rev .
phys . * 84 * , 621 ( 2012 ) .
t. c. ralph , and a. p. lund , nondeterministic noiseless linear amplifcation of quantum systems , quantum communication measurement and computing , proceedings of 9th international conference , lvovsky , a. ( editor ) , 155160 ( aip , new york 2009 ) .
h. takahashi , j. s. neergaard - nielsen , m. takeuchi , m. takeoka , k. hayasaka , a. furusawa , m. sasaki , non - gaussian entanglement distillation for continuous variables , nature photonics * 4 * , 178 ( 2010 ) .
y. kurochkin , a. s. prasad , and a. i. lvovsky , distillation of the two - mode squeezed state , phys .
lett . * 112 * , 070402 ( 2014 ) .
t. j. bartley and i. a. walmsley , directly comparing entanglement - enhancing non - gaussian operations , new j. phys . *
17 * , 023038 ( 2015 ) .
h. m. chrzanowski , n. walk , s. m. assad , j. janousek , s. hosseini , t. c. ralph , t. symul , and p. k. lam , measurement - based noiseless linear amplification for quantum communication , nature photonics * 8 * , 333 ( 2014 ) .
g. y. xiang , t. c. ralph , a. p. lund , n. walk and g. j. pryde , noiseless linear amplification and distillation of entanglement , nature photonics * 4 * , 316 ( 2010 ) .
f. ferreyrol , et al . , implementation of a nondeterministic optical noiseless amplifier , phys
lett . * 104 * , 123603 ( 2010 ) .
a. zavatta , j. fiurasek , and m. bellini , a high - fidelity noiseless amplifier for quantum light states .
nature photon .
* 5 * , 52(2010 ) .
s.kosis , g.y.xiang , t.c.ralph and g.j.pryde , heralded noiseless amplification of a photon polarization qubit , nature physics * 9 * , 23 ( 2013 ) d. pegg , l. phillips , s. barnett , optical state truncation by projection synthesis .
* 81 * , 1604(1998 ) .
s. a. babichev , j. ries , and a. i. lvovsky , quantum scissors : teleportation of single - mode optical states by means of a nonlocal single photon , europhys
. lett . * 64 * , 1 ( 2003 ) .
a. i. lvovsky , j. mlynek , quantum - optical catalysis : generating nonclassical states of light by means of linear optics , phys.rev.lett .
* 88 * , 250401 ( 2002 ) .
see the supplementary information for the details of all theoretical calculations .
r. kumar , e. barrios , a. macrae , e. cairns , e.h .
huntington , and a.i .
lvovsky , opt .
commun . , * 285 * , 5259 ( 2012 ) .
s. r. huisman , n. jain , s. a. babichev , f. vewinger , a .-
zhang , s .- h .
youn , and a. i. lvovsky , instant single - photon fock state tomography , opt
. lett . * 34 * , 2739 ( 2009 ) .
duan , g. giedke , j. i. cirac , and p. zoller , inseparability criterion for continuous variable systems , phys .
* 84 * , 2722 ( 2000 )
. j. w. pan , s. gasparoni , r. ursin , g. weihs , a. zeilinger , experimental entanglement purification of arbitrary unknown states , nature * 423 * , 417 ( 2003 ) .
t. c. ralph , quantum error correction of continuous - variable states against gaussian noise , phys .
a * 84 * , 022339 ( 2011 ) .
w. wasilewski , a. i. lvovsky , k. banaszek , and c. radzewicz , pulsed squeezed light : simultaneous squeezing of multiple modes phys .
a * 73 * , 063819 ( 2006 ) .
a. i. lvovsky , w. wasilewski and k. banaszek , decomposing a pulsed optical parametric amplifier into independent squeezers , j. mod
. opt . * 54 * , 721 ( 2007 ) .
u. l. andersen , j. s. neergaard - nielsen , p. van loock , a. furusawa , hybrid quantum information processing , arxiv:1409.3719 ( 2014 ) .
a. scherer , b. c. sanders , w. tittel , opt .
express * 19 * , 3004 ( 2011 ) .
d. w. berry and a. i. lvovsky , linear - optical processing can not increase photon efficiency , phys .
lett . * 105 * , 203601 ( 2010 ) .
d. w. berry and a. i. lvovsky , preservation of loss in linear - optical processing , phys .
a * 84 * , 042304 ( 2011 ) .
suppose a two - mode state prepared in modes @xmath0 and @xmath1 is subjected to one - sided loss in mode @xmath1 .
this loss results in quadrature transformation @xmath39 where @xmath10 is the amplitude transmissivity of the loss channel and @xmath40 is the position quadrature of the vacuum state with variance @xmath41 . we define the duan _
_ inseparability parameter @xcite as
@xmath36 the perfect epr state has @xmath42 and @xmath43 , but each quadrature taken individually has infinite variance .
because of the latter fact , the only value of @xmath37 for which @xmath44 takes a finite value is @xmath45 .
then for the position quadratures we have @xmath46
the same value obtains for the momentum term : @xmath47 accordingly , we have @xmath48 for our experiment , @xmath49 and hence @xmath50 .
we begin by considering the idealized case described by eq .
( 3 ) in the main text .
the variances of the individual position or momentum quadratures in that state are @xmath51}\nonumber\end{aligned}\ ] ] while their correlation equals @xmath52 the variance of the quadrature sum and difference is therefore @xmath53 where @xmath54 , @xmath55 being the nla gain .
the position difference ( momentum sum ) variance is minimized for @xmath56 , at which point it takes the value @xmath57 .
let us now develop a more complete theory , taking into account the efficiency @xmath58 of the ancilla photon as well as the efficiencies @xmath59 and @xmath60 of the detectors in both channels , according to fig .
our goal is to calculate the covariance matrix of the distilled state .
we continue to work in the subspace of the hilbert space spanned by states @xmath61 and assume the perfect efficiency of the heralding spcm .
both these assumptions are well - justified for @xmath62 and @xmath63 .
the inefficient ancilla photon source generates the mixture of the single - photon state with probability @xmath58 and the vacuum with probability @xmath64 . in the former case ,
the distilled epr state is @xmath65 , while in the latter case , the click of the heralding detector will occur only if the photon from the epr state propagates through the beam splitter towards the spcm , resulting in the unnormalized state @xmath66 in the distilled output .
taking into account both possibilities , we obtain the following output state @xmath67 where we have not yet taken into account the homodyne detector efficiencies @xmath59 and @xmath60 . in the second term of the above equation we have @xmath69 from these results , as well as from eqs . and
, we find the covariance matrix elements of state : @xmath70 accounting for the detector efficiencies affects these results as follows : @xmath71 the covariance matrix elements related to the momentum are transformed in a similar fashion .
in addition to the above analytical calculation , we performed a numerical modeling of our experiment using the density matrix in the fock basis , in the subspace spanned by the zero- to three - photon fock states .
figure s2 shows the variances @xmath72 predicted by the two techniques
. as expected , the predictions are very similar as long as conditions @xmath62 and @xmath63 hold .
the single - photon approximation breaks down for low amplifications because for macroscopic nla beam splitter reflectivities @xmath73 we obtain a significant likelihood of hong - ou - mandel type events when both the ancilla photon and the photon from the epr state exit the beam splitter through the same port .
we used this numerical model to calculate the theoretical fit to the data shown in fig . 2 of the main text .
note that our model does not take into account a possible mode mismatch between the to - be amplified epr state mode and the ancilla photon .
this is a possible explanation for the discrepancy between the measured and best - fit values for the initial squeezing ( @xmath74 and @xmath75 , respectively ) .
the variances of the sum and difference of the position quadratures of a pure epr state with squeezing parameter @xmath76 are @xmath77 , respectively @xcite .
if this state experiences two - sided loss by propagating through channels with efficiencies @xmath78 and @xmath79 , covariance matrix of the output can be calculated from eq .
( [ losscovar ] ) : @xmath80;\\ { \left\langle x_b^2\right\rangle}_{eq}&={\left\langle p_b^2\right\rangle}_{eq}=\frac 1 2[\eta_{b , eq}\cosh 2\gamma_{eq}+(1-\eta_{b , eq})];\\ -{\left\langle x_ax_b\right\rangle}_{eq}&={\left\langle p_ap_b\right\rangle}_{eq}=\frac 12\sqrt{\eta_{a , eq}\eta_{b , eq}}\sinh 2\gamma_{eq}.\end{aligned}\ ] ] accordingly , the variances of the sum and difference of the position quadratures become @xmath81 knowing the experimentally measured quantities @xmath72 for the distilled state , one can solve the above equation to find the parameters of the equivalent state that would exhibit similar quadrature statistics . in the calculations shown in fig .
[ f3 ] , we assumed the efficiency @xmath78 of the equivalent state to equal the actual efficiency @xmath59 of the undistilled channel obtained from the fit to the experimental data , and calculated @xmath76 and @xmath79 . in this way , the quantity @xmath79 describes the equivalent channel efficiency , thereby serving as a benchmark of our distillation protocol .
the high uncertainty bars in fig .
[ f3 ] are due to the relatively low degree of squeezing obtained after the distillation .
small variations of quadrature variance measurements have significant effect on the solutions of eq .
( [ equivvar ] ) .
|
entanglement distillation is a process via which the strength and purity of quantum entanglement can be increased probabilistically .
it is a key step in many quantum communication and computation protocols . in particular
, entanglement distillation is a necessary component of the quantum repeater , a device which counters the degradation of entanglement that inevitably occurs due to losses in a communication line .
here we report an experiment on distilling the einstein - podolsky - rosen ( epr ) state of light , the workhorse of continuous - variable entanglement , using the technique of noiseless amplification .
in contrast to previous implementations , the entanglement enhancement factor achievable by our technique is not fundamentally limited and permits recovering an epr state with a macroscopic level of entanglement no matter how low the initial entanglement or how high the loss may be . in particular , we recover the original level of entanglement after one of the epr modes has passed through a channel with a loss factor of 20 . the level of entanglement in our distilled state is higher than that achievable by direct transmission of any state through a similar loss channel .
this is a key bench - marking step towards the realization of a practical continuous - variable quantum repeater and other cv quantum protocols .
|
according to theoretical suggestions , a regular secular decrease in the pulsar rotation frequency can be described by the relation @xmath2 , where @xmath3 is the braking index @xcite . for the vacuum dipole model , when the secular spin - down is due to magnetic dipole radiation , @xmath4 .
the direct way of defining the braking index , @xmath5 , is based on a measurement of the second derivative @xmath0 from the pulsar observations .
the early timing observations showed that the rotation rate of many pulsars is subject to irregularities such as glitches ( discrete jumps in the rotation rate ) and timing noise ( variations in the pulse arrival times ) .
these irregularities disturb the steady secular spin down and hinder a measurement of the deterministic values of @xmath0 . for ordinary pulsars ,
the values of @xmath0 due to slowdown are very small in comparison to the measurement uncertainties and a rotation phase should be well described by a simple @xmath1 spin - down model . in actuality , most ordinary pulsars show significant deviations from a simple spin - down model .
an analysis of timing noise for these pulsars has shown that these deviations are due to a random , continuous wandering of the pulse phase that produces long - term cubic polynomial components in the timing residuals .
these components lead to the large observed @xmath0 values and accordingly to anomalously high values of the braking indices @xcite .
the properties of timing noise have been studied by numerous authors over the past 40 yr @xcite .
it has been established that timing noise is widespread in pulsars .
the strength of the timing noise is correlated with the period derivative and is quantified by a parameter based on the non - zero second derivative @xmath0 .
in fact , the observed @xmath0 value is the measure of timing noise in pulsars @xcite .
it is widely believed that timing noise can be explained by random fluctuations in the pulsar rotation rate .
a description of timing noise is constructed in the context of physical models that are based on noise - like processes .
important results were recently obtained by @xcite who presented a detailed study of the timing behavior of 366 pulsars observed at the jodrell bank observatory ( jbo ) over a long time span of 36 yr .
it was shown that a variety of different structures visible in the timing residuals can be classified into groups . out of a sample of 366 pulsars ,
approximately 37% pulsars have timing residuals where noise - like variations dominate , about 36% pulsars have residuals where cubic structures with both signs dominate , and the remaining 27% of pulsars show more complex structures .
@xcite pay attention to the necessity of using long data spans to analyze the timing residuals , as otherwise the structure of the residuals will depend on the length of the data span analyzed .
@xcite discussed in detail the problems that are related to timing noise and the large values of the measured second derivatives @xmath0 in pulsars , and paid attention to the important properties of timing noise .
first , there is a large anti - correlation between the amount of timing noise and characteristic age .
pulsars with smaller characteristic ages have more timing noise than older pulsars .
second , there is a relation between the signature of timing residuals and characteristic age : the timing residuals observed for the pulsars with the smaller characteristic ages exhibit mainly cubic signatures , whereas the timing residuals seen for older pulsars have quasi - periodic characteristics .
the main conclusion of this paper is that the timing residuals can not be explained by the models that are based on noise - like processes because such models are not consistent with observations over long time scales . in this paper
, we analyze a 33.5 yr period of pulsar timing observations made at the pushchino radio astronomy observatory ( prao ) between 1978 and 2012 .
the prao data sets for 10 pulsars that are in common with the jet propulsion laboratory ( jpl ) sample were combined with the archival jpl data spanning the years 19681983 @xcite .
this procedure extends the individual data spans for some pulsars up to 43.5 yr .
the common pulsars are b0329 + 54 , b0823 + 26 , b0919 + 06 , b0950 + 08 , b1133 + 16 , b1642@xmath603 , b1737 + 13 , b1929 + 10 , b2016 + 28 , and b2217 + 47 .
the combined data set for psr b1642@xmath603 includes also the jbo data collected over the interval 19811999 taken from an earlier paper @xcite .
the data set for psr b1822@xmath609 includes the hartebeesthoek radio astronomy observatory data collected over the interval 19851998 taken from a published paper @xcite . while our sample of 27 pulsars is small
, these objects nevertheless reflect the properties of a large number of pulsars .
figure [ gist ] shows a histogram of the characteristic ages of known pulsars based on the atnf pulsar catalogue @xcite .
the 27 studied pulsars have ages ranging from @xmath7 to @xmath8 yr ; this interval is marked by the horizontal line .
more than 80% of the global pulsar population has ages within this range . on the basis of a detailed analysis of the timing residuals of these pulsars , we obtained the following results .
first , we detected a new phenomenon in the timing behavior of two pulsars , b0823 + 26 and b1929 + 10 , that demonstrates a rapid change of pulsar rotation parameters in which the sign of @xmath0 is reversed .
this phenomenon may be interpreted as a modulation process that changes the sign of @xmath0 .
the second result is that we showed that a long - term structure of timing noise is produced by four types of rotational irregularities that have an evolving nature and form three evolutionary stages in pulsar rotation .
according to the evolutionary scenario , certain types of rotational irregularities can only occur at certain stages of pulsar rotation evolution .
pulsar timing observations were carried out at the pushchino observatory from 1978 july to 2012 february with a four - year gap between 1987 and 1990 .
the observations were made with the bsa radiotelescope , which operated at frequencies close to 102.7 mhz until 1998 may and around 111.3 mhz since 1998 november . a 64-channel radiometer with a channel bandwidth of 20 khz
was used for the pulsar observations .
the data were sampled at intervals of 2.56 or 1.28 ms .
the bsa radiotelescope , made up of a linearly polarized transit antenna with a beam size of about ( @xmath9 ) min , provided the duration of the observing session ranging from 3 to 11 minutes at different pulsar declinations @xmath10 . during this time
, individual pulses were summed synchronously with a predicted topocentric pulsar period to form the mean pulse profile in each 20 khz channel .
after dispersion removal , all the channel profiles were summed to produce a mean pulse profile for the given observing session .
the topocentric arrival times of the pulses for each observing session were calculated by cross - correlating the mean pulse profile with a standard low - noise template .
the parameters of the prao timing observations are listed in table [ inform27 ] . in column order
, the table lists the pulsar s j2000 and b1950 names , the mjd range of the prao observations , the number of pulse time of arrival ( toa ) , the mean measurement error for each observing session , and the dispersion measure ( dm ) parameter .
the references in the last column mark the dm value that provided the best adjustment between the two sets of the timing residuals obtained for each pulsar at the two observing frequencies of 102.7 mhz ( until 1998 may ) and 111.3 mhz ( since 1998 november ) in the prao observations .
timing parameters were determined using the program tempo and the jpl de200 ephemeris .
the topocentric arrival times collected at the prao and the geocentric arrival times obtained from the archival jpl timing data were all corrected to the barycenter of the solar system at infinite frequency .
the position and the proper motion required for this correction were taken from @xcite and were held constant during the fitting procedure .
the positions of most of the pulsars from our sample do not require correction since their residuals ( defined as the observed minus predicted arrival times ) obtained after the fitting procedure did not show a sinusoid with a period of 1 yr within the precision of the measurements .
position corrections were needed for pulsars b1508 + 55 , b2020 + 28 , and b2224 + 65 , which are fast moving pulsars , but a timing solution did not provide improved positions for these pulsars because of the large timing noise . as the timing behavior of these pulsars
was analyzed over a long interval , the presence of a weak annual sine wave at the end of this interval does not affect the large - scale structure of the timing residuals .
the regular change of pulsar s rotation frequency with time is well described by a polynomial including several frequency derivatives @xcite . in accordance with a taylor expansion , the pulse phase @xmath11 at the barycentric arrival time @xmath12
is expressed as : @xmath13 where @xmath14 , @xmath15 , @xmath16 , and @xmath17 are the pulse phase , rotation frequency , and the first and second frequency derivative at some reference time @xmath18 , respectively .
the timing residuals , obtained as the differences between the observed times and the times predicted by a best - fit model , are used for the analysis of the rotation behavior of the pulsar . in order to study the time changes in a pulsar s rotation frequency in more detail
, we show the plots of the timing residuals together with the plots of the frequency residuals @xmath19 relative to the different timing models . at first , the timing residuals are calculated relative to a second - order polynomial , including the mean rotation parameters @xmath15 and @xmath20 defined over the whole data span .
the corresponding frequency residuals @xmath19 are obtained as a result of calculating the parameters @xmath15 and @xmath20 from the local fits , performed to the pulse arrival times over intervals from 150 to 300 days .
the @xmath19 residuals are given relative to the mean parameters @xmath15 and @xmath20 and are referenced to the first point of each interval .
next , the timing and frequency residuals are calculated relative to a third - order polynomial for @xmath1 , and @xmath0 using a similar procedure .
the results of the timing analysis for the observed pulsars are given in table [ param27 ] .
the columns list , respectively , the pulsar s b1950 name , the rotation frequency , @xmath15 , the first frequency derivative , @xmath20 , and the second frequency derivative , @xmath0 . the next four columns give the epoch of the @xmath15 measurement , the mjd range of the observations , the unweighted rms value remaining after a second - order fit @xmath21 , and the unweighted rms value remaining after a third - order fit @xmath22 .
@xcite found that changes in pulse shape parameters are correlated with changes in the spin - down parameters of some pulsars . at the prao ,
the pulsar signals are received with a linearly polarized meridian telescope . for the majority of pulsars from our sample ,
the observed shape of the mean pulse profile varies from day to day because of either the influence of the faraday effect or the lack of a sufficient number of individual pulses for an observing session . under such conditions
, we can not determine our pulse shape changes and repeat a similar analysis of the pulse shape parameters and the spin - down parameters .
therefore , the pulse shape changes are not discussed in the later sections . the timing behavior of the studied pulsars is shown in figures [ resid27](a)(h ) .
the residual plots for each pulsar are shown in two panels .
the timing residuals relative to a simple @xmath23 spin - down model are shown in the left - hand panels .
the corresponding frequency residuals @xmath19 are shown in the right - hand panels .
these plots are arranged in order of increasing characteristic age of the pulsars , @xmath24 . the pulsar s b1950 name and the pulse period ( seconds )
are indicated in each left - hand panel .
the pulsar age in years is indicated on a logarithmic scale in each right - hand panel .
the periods of the studied pulsars are between 0.1 and 3 s. it is clearly seen in figure [ resid27 ] that the timing behavior of most pulsars in our sample significantly deviates from a simple @xmath15 , @xmath20 spin - down model
. such deviations , known as timing noise , may be a result of significant irregularities in the rotation rate of the pulsars .
table [ param27 ] shows that the timing residuals for all the observed pulsars are characterized by significant second derivatives that are too large to be attributed to pulsar spin - down .
these values of @xmath0 testify that all our pulsars have a high level of timing noise .
detailed inspection of the timing residuals plotted in figure [ resid27 ] shows that there is an appreciable correlation between the amplitude of the residual curve and the characteristic age of the pulsar the smaller the amplitude of the residual curve , the older the pulsar . as seen in figure [ resid27](a ) , the timing residuals for younger pulsars have complex structures including cubic polynomial components with large amplitudes . as the age of a pulsar increases ,
we more often observe residuals that exhibit a quasi - periodic structure with smaller amplitudes , as is shown in figure [ resid27](c ) for psr b1642@xmath603 and b1133 + 16 , for example .
the older pulsars mainly have the structures where noise - like variations dominate , as in figure [ resid27](g ) , for example , for the pulsars b0320 + 39 and b0809 + 74 . in order to study the effect of timing noise on the @xmath0 measurements , we analyze the timing residuals together with the frequency residuals and
try to determine the causes of the large values of the measured @xmath0 for each given pulsar .
we distributed all the observed pulsars into four groups according to the signatures of their timing residuals .
the first group includes eight pulsars , in which the timing residuals take the form of a cubic polynomial .
five of these pulsars have experienced glitches in the past according to the literature .
these pulsars have the largest @xmath0 values , as shown in table [ param27 ] .
the second group includes seven pulsars that show quasi - periodic variations in their timing residuals .
the third group includes 10 pulsars that exhibit timing residuals where noise - like variations dominate .
the fourth and final group is unique .
this group includes two pulsars , b0823 + 26 and b1929 + 10 , whose timing behavior demonstrates a rare phenomenon a change in the sign of the second derivative @xmath0 .
studying this phenomenon will help us to understand why pulsars with characteristic ages greater than @xmath25 ys have both positive and negative @xmath0 values .
the first group includes eight pulsars that show cubic components in their timing residuals .
the amplitude of the timing residuals for these pulsars varies from one quarter up to more than one pulse period over spans ranging from 19 to 34 yr .
the characteristic ages of these pulsars range from @xmath26 to @xmath27 yr . * ( 1 ) b1822@xmath609*. @xcite was the first to detect slow glitches in the rotation frequency of the pulsar b1822@xmath609 .
slow glitches represent a unique glitch phenomenon that results in regular oscillations in the rotation frequency .
a full pattern of the timing and frequency residuals for the pulsar b1822@xmath609 over the 26 yr span between 1985 and 2012 is shown in figures [ resid27](a ) and [ resid27](b ) , respectively .
these plots show that the pulsar experienced three discrete glitches ( marked by the three arrows in the left - hand plot ) and suffered a sequence of five slow glitches that occurred over the 19952004 interval .
an oscillatory character of the timing and frequency residuals seen in both the plots over the interval 19952004 is due to the presence of slow glitches .
the signature of the large discrete glitch that occurred in 2007 january @xcite is clearly visible on the right side of the @xmath19 plot .
the difference between the signature of a slow glitch and that of a discrete glitch is clearly seen in figure [ freqnew1822 ] , where the time behavior of the frequency derivative @xmath20 and the @xmath19 residuals are presented relative to the timing model 19911994 throughout the whole time span from 1985 to 2012 . here
, the epochs of the three discrete glitches of 1994 , 2006 , and 2007 are marked by arrows pointing upward and the epochs at which the slow glitches occurred are marked by arrows pointing downward .
it is seen in figure [ freqnew1822](b ) that while the discrete glitch of 2007 represents a sudden jump in the @xmath19 residuals , the slow glitches exhibit gradual exponential increases in the rotation frequency with a time scale of 200300 days .
figure [ freqnew1822](a ) displays the time behavior of @xmath20 , which has some surprising features .
first , a gradual increase in @xmath19 is accompanied by a rapid decrease in the magnitude of @xmath20 by @xmath281%2% of the initial value and a subsequent exponential increase back to its initial value .
second , the peaks of @xmath29 , which characterize the steepness of the front in the residuals @xmath19 , lie on a parabolic curve that is the envelope of these peaks .
the existence of the envelope indicates that all of the slow glitches are components of one process that acted during the 10 yr between 1995 and 2004 .
the glitch history of psr b1822@xmath609 was described in detail in a series of papers @xcite .
the glitch parameters of this pulsar from the prao measurements are given in table [ glit1822 ] .
the final part of the oscillatory process over the interval 2000
january2003 november in psr b1822@xmath609 was also observed by @xcite .
the authors have independently detected a glitch event with an unusual signature that occurred over this interval and have interpreted this glitch in terms of a slow glitch , as suggested by @xcite . in a sequence of five slow glitches , shown in figure [ freqnew1822 ] , this glitch is designated as the third one .
the fourth glitch is very small and its signature is not seen .
the glitch events in this pulsar were studied in a number of other papers @xcite .
examining these papers shows that the glitch epochs , indicated in our table [ glit1822 ] , are generally in reasonable agreement with the measurements of other authors , but the interpretations of the glitch events are different . according to table 1 of @xcite ,
psr b1822@xmath609 has experienced six glitches between mjd 49940 and 54115 and none of these glitches was interpreted as a slow glitch .
we can not compare a full pattern of the prao glitch residuals , presented in figure [ freqnew1822](b ) , with that of the jbo glitch residuals because figures 6 and 7 of @xcite show separate residual plots for four different glitches .
the idea of slow glitches has recently received support .
@xcite suggested that the slow glitch - like events that result in the quasi - sinusoidal features in the timing residuals should be labelled as slow glitches .
figure [ postglt1822 ] displays the timing residuals for psr b1822@xmath609 after the glitch of 2007 relative to a simple spin - down model .
the observed cubic term points to a large positive value @xmath30 s@xmath31 that can be explained by the recovery from the discrete glitch .
we conclude that rotational irregularities , such as slow glitches and discrete glitches , are the cause of the deviation of the timing behavior of this pulsar from a simple @xmath1 spin - down model . *
( 2 ) b0919 + 06*. as reported by @xcite , the slow glitch phenomenon was also detected in the rotation frequency of the pulsar b0919 + 06 .
figure [ resid27](a ) shows the timing residuals relative to a timing model of @xmath1 , and @xmath0 .
it is seen that the residual curve displays a large quartic term with the amplitude approximately equal to half the pulsar period .
rapid oscillations that are imposed on this curve are interpreted as a sequence of slow glitches .
the arrow in figure [ resid27](a ) marks the epoch of a large glitch of magnitude @xmath32 that occurred in 2009 november 5 ( mjd 55140 ; @xcite ) .
the cubic component of this complex structure corresponds to a large value @xmath33 s@xmath31 which leads to a large braking index , @xmath34 .
a sawtooth - like character of the frequency residuals @xmath19 , shown in figure [ resid27](b ) , implies that the rotation frequency of this pulsar underwent rapid oscillations with a time scale of about 600 days during the whole 30 yr period of observations . as discussed by @xcite ,
the cause of these oscillations lies in the continuous generation of similar slow glitches .
these glitches are characterized by small absolute amplitudes of @xmath35 hz , long rise times of 200 days , and relaxation time intervals of 400 days .
the sequence of the observed slow glitches is well approximated by a periodic sawtooth - like function with a period of 600 days .
further observations of this pulsar are needed to study the relation between the discrete glitch and a process that continuously generates rapid oscillations identified with slow glitches .
we conclude that rotational irregularities , such as long - term quasi - periodic oscillations and a sequence of slow glitches and a large discrete glitch , are responsible for the large measured @xmath0 values in this pulsar . * ( 3 ) b0355 + 54*. figure [ resid27](a ) displays the timing residuals obtained for the pulsar b0355 + 54 at the prao during the two observational intervals 19821986 and 19912012 .
this pulsar underwent two glitches in the 1980s .
as reported by @xcite , the first glitch occurred in 1985 ( mjd 46077 ) and was quite small with @xmath36 .
the second occurred in 1986 ( mjd 46497 ) and was a giant glitch with @xmath37 .
these glitches were also registered in the prao observations @xcite .
the glitch epochs are marked by two arrows in figure [ resid27](a ) .
figure [ resid27](a ) shows that the timing residuals for psr b0355 + 54 over the post - glitch interval 19912012 take the form of a cubic polynomial and are characterized by a significant @xmath0 value .
the plotted cubic trend presents the residuals calculated relative to a simple @xmath1 timing model 1991 - 2010 because all of the pulse arrival times gathered over the entire 19912012 interval can not be described by a simple spin - down model within half the pulse period because of the presence of a large @xmath0 .
this plot shows that the amplitude of the cubic structure , equal to @xmath28 125 ms , exceeds half the pulse period .
the observed cubic trend corresponds to a positive value @xmath38 s@xmath31 that gives a braking index of @xmath2833 .
it is evident that the observed cubic structure in the timing residuals is caused by the recovery from the 1986 glitch .
the corresponding frequency residuals presented in figure [ resid27](b ) also indicate the presence of a large @xmath0 . from this plot
, it is seen that all the @xmath19 points lie on a parabolic curve that is due to the presence of a large @xmath0 .
@xcite studied the timing behavior of this pulsar during @xmath28 6 yr between 1999 and 2005 and found four micro - glitches with sizes of @xmath39 , and @xmath40 , respectively .
the epochs of these events are marked by vertical lines in figure [ resid27](b ) .
it is seen that these small glitch events are consistent with the level of fluctuations in the @xmath19 curve .
therefore , they do not influence large - scale structure in the frequency residuals and they do not contribute to the measured value of @xmath0 .
we can not compare our results with the jbo measurements because psr b0355 + 54 is not included in the jbo sample @xcite .
the timing and frequency residuals for psr b0355 + 54 after a fit of a third - order polynomial for @xmath1 , and @xmath0 are presented in figure [ cubic6 ] .
it is seen that the amplitude of the remaining residuals has considerably decreased .
the @xmath19 curve is straightened .
this plot , as well as figure [ resid27](b ) , shows that four micro - glitches , whose epochs are marked by the vertical lines , are local events and do not influence the measured value of @xmath0 . in figure
[ cubic6 ] , it is evident that the second derivative @xmath0 dominates the timing and frequency residuals over the whole span of observations of this pulsar .
thus , a large cubic trend visible in the timing residuals of psr b0355 + 54 is caused by the recovery from the 1986 glitch . *
( 4 ) b2224 + 65*. the timing residuals of psr b2224 + 65 over the interval 19822012 are presented in figure [ resid27](a ) .
this pulsar suffered a large glitch with a magnitude of @xmath41 in 1976 @xcite .
the glitch epoch is marked by an arrow in this plot .
the plot shows that the timing residuals have a complex structure with a large amplitude of about 270 ms .
the cubic component of this structure is characterized by a positive @xmath0 , equal to @xmath42 s@xmath31 , that leads to a braking index of 106 .
the corresponding frequency residuals @xmath19 are presented in figure [ resid27](b ) .
we pay specific attention to an arc - like extended detail clearly seen in the @xmath19 curve between 1999 and 2007 ( @xmath28 mjd 5120054100 ) .
four vertical lines plotted on the @xmath19 panel mark the epochs of three micro - glitches with magnitudes of @xmath43 , and @xmath44 , respectively , reported by @xcite , and the epoch of a fourth glitch with @xmath45 , reported by @xcite .
the plot shows that all of these lines are in the range of the arc - like extended detail .
most likely , these small glitches are a consequence of a complex structure of the @xmath19 curve . the pulsar b2224 + 65 is associated with the guitar nebula and is the fastest known pulsar .
it has a large transverse velocity of 8001600 km s@xmath46 for a distance of 12 kpc .
the motion of psr b2224 + 65 through a medium with significant density fluctuations can cause changes in the dm that can be measured in timing observations @xcite .
we suppose that the arc - like extended detail in @xmath19 can be an important indication of the existence of dm variations for the pulsar b2224 + 65 .
there are a number of arguments in favor of the dm variations .
first , there is a discrepancy among the dm values measured by different authors at different epochs .
as follows from table [ dm ] , the difference among the measured dm values significantly exceeds the estimated measurement uncertainties . at the prao
, the dm value was measured using the phase shift between the two sets of the timing residuals obtained at the two observing frequencies of 102.746 ( until 1998 may ) and 111.646 mhz ( after 1998 november ) .
the timing residuals for psr 2224 + 65 , plotted in figure [ resid27](a ) , were calculated for the value dm=36.226(5 ) pc cm@xmath31 . secondly , the jbo residuals , plotted in figure 3 of @xcite , show a flatter structure compared with that of the prao residuals , plotted in figure [ resid27](a ) .
the low - frequency timing measurements are very sensitive to changes in the dm .
for example , a continuous increase in the dm by @xmath48 dm @xmath28 0.2 pc cm@xmath31 during several years of observations will cause delay in the pulse arrival times registered at 112 mhz during this interval , where the delays can be as large as 66 ms .
it accordingly follows that changes of the dm changes with time , if they exist , will more considerably influence the form of the residuals obtained at lower frequencies , for example , 112 mhz .
besides , as is shown in figure [ cubic6 ] , where the timing and frequency residuals are presented after a fit of a third - order polynomial , a remnant structure of the arc - like extended detail has the same shape , as in figures [ resid27](a ) and [ resid27](b ) .
this result means that the @xmath19 variations over the interval 19992007 can not be attributed to the presence of @xmath0 .
reliable information about the dm variations can be obtained from a comparison of the high- and low - frequency timing data collected over the period 19822012 in the quasi - simultaneous observations of this pulsar at different observatories .
figure [ cubic6 ] also shows that the micro - glitches , whose epochs are marked by the vertical lines in the @xmath19 plot , do not influence the large - scale structure of the frequency residuals . as in the case of psr b0355 + 54 , small glitch events are local events and do not contribute to the long - term behavior of @xmath0 .
we conclude that a large positive value of @xmath0 measured for the pulsar b2224 + 65 may be explained by the recovery from a large glitch in 1976 .
it is also very likely that the dm variations additionally contribute to the measured @xmath0 value . *
( 5 ) b1508 + 55*. the pulsar b1508 + 55 experienced a small glitch of @xmath49 that occurred in early 1973 @xcite .
figure [ resid27](a ) displays the timing residuals for 18 yr of observations after 1993 .
the epoch of the 1973 glitch is marked by an arrow .
the plotted timing residuals take the form of a cubic polynomial that is characterized by the negative value @xmath50 s@xmath31 ; this result leads to a braking index of -562 .
figure [ resid27](b ) shows the time behavior of the frequency residuals @xmath19 . from this plot ,
it is seen that all the @xmath19 points are well described by a parabolic curve despite noticeable quasi - sinusoidal deviations of these points from the approximating curve .
a parabolic pattern of the @xmath19 changes means that the second derivative @xmath0 dominates the timing residuals during the entire observational interval .
a quasi - sinusoidal structure in the rotation rate is clearly visible in figure [ cubic6 ] , where the timing and frequency residuals are shown after the removal of a cubic polynomial .
it is seen that the post - fit timing residuals show two cycles of a quasi - periodic structure with a variable amplitude of about 3010 ms .
the corresponding @xmath19 residuals also exhibit noticeable quasi - periodic variations in the rotation frequency . comparing with the results of @xcite
shows that the signature of the timing residuals , presented in our figure [ resid27](a ) , is unlike that of the residuals , shown in their figure 3 .
according to table 1 of @xcite , this pulsar was observed at the jbo over the 19822006 interval . at the prao , b1508 + 55
was observed at another interval , between 1993 and 2012 . as is shown in figure 5 of @xcite
, the timing residuals obtained from different section of the entire data span can have a different form .
it is very likely that we observe this effect for this pulsar .
thus , we found that the timing residuals of psr b1508 + 55 are characterized by a large negative @xmath0 .
its origin is uncertain .
it is unlikely that the complex structure seen in the residuals of this pulsar can be a consequence of a small discrete glitch of 1973 . *
( 6 ) b2020 + 28*. the pulsar b2020 + 28 is not known to have glitched in the past .
nevertheless , the timing residuals seen in this pulsar have the form of a cubic polynomial corresponding to a positive @xmath0 . figures [ resid27](a ) and ( b ) exhibit the timing and frequency residuals that characterize the rotation of the pulsar over the 21 yr span from 1991 to 2012 .
the observed quasi - cubic trend in the timing residuals corresponds to a large value @xmath51 s@xmath31 , which gives a braking index @xmath52 .
figure [ resid27](b ) clearly displays that the changes in the @xmath19 residuals with time are well approximated by a parabolic curve that indicates the dominant character of @xmath0 in the rotation rate of the pulsar .
the timing and frequency residuals after subtraction of a cubic polynomial from the pulse arrival times are shown in figure [ cubic6 ] .
we see that the remaining timing residuals have cyclical variations with small amplitude of @xmath28 8 ms .
the corresponding @xmath19 residuals , plotted on the right panel , show noticeable quasi - periodic variations that range between @xmath53 hz .
we can not compare our results with the jbo measurements because psr b2020 + 28 is not included in the jbo sample @xcite .
we hypothesize that the cubic form of the timing residuals corresponding to a large positive derivative @xmath0 could be the consequence of an unobserved glitch event that occurred in b2020 + 28 before our observations began in 1991 . *
( 7 ) b0943 + 10*. figure [ resid27](c ) shows the timing residuals for the pulsar b0943 + 10 over the 30 yr span from 1982 to 2012 .
we see that the timing residuals are characterized by a large cubic trend corresponding to a negative @xmath0 .
the plotted residuals are presented relative to the timing model 19822008 , as not all the pulse arrival times can be described by a simple spin - down model within half the pulsar period because of the presence of a large @xmath0 .
this plot shows that the deviations from the simple timing model are very large , about @xmath281500 ms , and that they are more than one pulse period .
the cubic trend in the timing residuals corresponds to a negative value @xmath54 s@xmath31 .
this value yields a braking index @xmath55 .
the frequency residuals , presented in figure [ resid27](d ) , show a parabolic change of the @xmath19 points with time .
this plot confirms that a large @xmath0 dominates the timing residuals of this pulsar over the whole observational interval .
the timing residuals after the removal of a cubic term are shown in the left panel of figure [ cubic6 ] .
a large quartic term visible in the remaining timing residuals points to the presence of a third significant frequency derivative .
the corresponding frequency residuals @xmath19 are plotted in the right panel of figure [ cubic6 ] .
the noticeable downward slope of the @xmath19 curve also indicates the presence of a third frequency derivative . comparing our timing residuals , shown in figures [ resid27 ] and
[ cubic6 ] , and the jbo timing residuals , shown in figure 3 and 13 of @xcite , reveals a discrepancy between the signatures of these residual curves .
the reason for this discrepancy is related to the presence of the third derivative in the rotation rate of this pulsar and the different length of the data sets analyzed .
the jbo data set covers the interval 19832006 ( table 1 of @xcite ) ; the prao data set spans more time and covers the interval 19822012 .
the influence of the third derivative becomes noticeable after 2006 , as is seen in figure [ resid27](c ) .
if we analyze the timing data over the same interval 19832006 , then the discrepancy between the prao and jbo residuals disappears .
thus , we found that the timing residuals for psr b0943 + 10 have the form of a cubic polynomial corresponding to a significant negative second derivative @xmath0 .
most likely , the pulsar b0943 + 10 could have had glitches in its rotation frequency in the past .
then , a large value of @xmath0 could be the consequence of these glitch events that occurred prior to our observations .
however , the reason for the negative sign of @xmath0 is unclear . * ( 8 ) b1737 + 17*. in the literature
, there are no indications that the pulsar b1737 + 17 glitched in the past , but figure [ resid27](e ) shows a clear cubic trend in the timing residuals that can be considered the signs of a recovery from a glitch .
the plotted timing residuals are presented over the interval 19782012 .
the first point from the jpl data set belongs to 1978 december .
it is possible that this pulsar suffered a glitch before the jpl timing observations started .
the cubic term in the timing residuals corresponds to a positive value @xmath56 s@xmath31 .
this value gives a braking index of @xmath28 5181 .
the frequency residuals presented in figure [ resid27](f ) show that all the @xmath19 points lie on a parabolic curve .
this result means that a large @xmath0 dominates the rotation behavior of this pulsar over the entire interval of the observations .
the timing and frequency residuals after the removal of a cubic term are given in figure [ cubic6 ] .
the remaining residuals show a noticeable long - term quasi - periodic structure .
comparison with the results of @xcite shows that the timing residuals , presented in our figures [ resid27](e ) and [ cubic6 ] , are in good agreement with the residuals given in their figures 3 and 13 .
we may suppose that a large positive @xmath0 measured in the pulsar b1737 + 13 could be caused by the recovery from an unobserved glitch that occurred prior to 1978 , before the jpl observations had begun .
* summary*. we found that all eight pulsars whose timing residuals take the form of a cubic polynomial have large second frequency derivatives that range between @xmath57 s@xmath31 .
the long - term structure of these timing residuals is determined by such rotational irregularities as discrete glitches , slow glitches , and quasi - periodic oscillations .
the second group includes seven pulsars that show cyclical or quasi - periodic signatures in their timing residuals .
the maximum amplitude of the cyclical changes in the timing residuals for the different pulsars varies from 4% to 40@xmath58 of their period over the 3044 yr spans of the observations .
the characteristic ages of these pulsars range from @xmath59 to @xmath60 yr . * ( 1 ) .
b2217 + 47*. figure [ resid27](c ) shows the timing residuals for psr b2217 + 47 over the 42 yr time span between 1970 and 2012 . from this plot
, we see that the timing residuals have a quasi - periodic structure over the interval 1970 january2011 october .
this structure is characterized by maxima spacing between cycles of about 1014 yr and an amplitude of about 510 ms .
we detected that this pulsar experienced a small glitch that occurred in 2011 october 24 ( mjd 55858 ) .
the glitch epoch is marked by an arrow .
the @xmath19 residuals , plotted in figure [ resid27](d ) , show that the glitch has a small absolute amplitude of @xmath61 hz .
this value corresponds to a fractional increase in the rotation frequency of @xmath62 .
the glitch parameters are given in table [ glt2217 ] .
the value of the second derivative measured before the glitch occurred is @xmath63 s@xmath31 , which gives a braking index @xmath64 .
figure [ cyclic7 ] shows the timing and frequency residuals after the removal of the cubic trend over the interval before the glitch occurred .
the quasi - periodic structure of the post - fit residuals is clearly seen in these plots .
the @xmath19 curve changes within a range of @xmath19 @xmath65 hz .
comparison with the results of @xcite shows that our residuals , plotted in figure [ resid27](c ) before the glitch occurred , are in good agreement with their timing residuals , given in their figure 3 .
we conclude that the measured negative value of @xmath0 is associated with the long - term quasi - periodic variation observed in the timing residuals for psr b2217 + 47 over the interval 19702011 . *
b1642@xmath603*. @xcite reported the detection of slow glitches in the rotation rate of the pulsar b1642@xmath603 and showed that the observed pattern of cyclical timing residuals is a result of the continuous generation of slow glitches whose amplitudes are modulated by some periodic large - scale process .
slow glitches in this pulsar show striking properties that allow us to predict the epochs and the amplitudes of new glitches in the pulsar s rotation rate .
the cyclical character of the timing and frequency residuals for psr 1642@xmath603 is well seen in figures [ resid27](c ) and [ resid27](d ) , respectively , which present these residuals over the 43.5 yr time span from 1968 to 2012 . with respect to the results of @xcite , we extend the observational interval by including three years of new observations . in @xcite
, we showed that the cause of the cyclical oscillations in the timing residuals of psr b1642@xmath603 is the continuous generation of slow glitches .
these glitches are characterized by long rise times of about 500 days and exponential relaxation over several years after a glitch .
the observed slow glitches have two surprising properties .
first , the amplitude of these glitches and the time interval until the following glitch obey a strong linear relation the larger the glitch amplitude , the larger the time interval to the next glitch .
this property allows us to predict epochs of new glitches .
secondly , the amplitude of these glitches is modulated by a periodic large - scale sawtooth - like function . as a result of such modulation ,
the glitch amplitudes change discretely in a strictly certain sequence assigned by the sawtooth - like function .
this property allows us to predict the amplitudes of new glitches . in our previous paper @xcite
, we analyzed the properties of eight slow glitches . according to the predictions of that paper ,
the ninth glitch will occur in @xmath28 2013 .
figures [ resid27](c ) and ( d ) show that this prediction has come true .
the maximum of the ninth glitch is clearly seen in the timing residuals , plotted in figure [ resid27](c ) .
the maximum of the @xmath19 residuals , plotted in figure [ resid27](d ) , occurs around mjd 56100 .
the observed epoch is slightly earlier than the predicted epoch of mjd 56300 .
nevertheless , these epochs are in agreement within the time resolution of @xmath28300 days .
further observations are needed to determine the amplitude of the ninth glitch .
we should measure the relaxation time interval after the ninth glitch . according to table 2 of @xcite , for the 43 yr modulation scheme including eight glitches ,
this information will be obtained at the end of 2013 or in the beginning of 2014 . in this case
, the observed amplitude of the ninth glitch should be equal to @xmath66 hz . in the case of the two other modulation schemes with modulation periods of 53 yr and 60 yr
, the relaxation time interval will be known in the middle of 2015 .
then , the observed amplitude of the ninth glitch should be equal to @xmath67 hz .
these two schemes will differ starting with the 10th glitch that should occur in @xmath28 2018 .
the cyclical structure in this pulsar is characterized by a positive value @xmath68 s@xmath31 that yields a braking index @xmath69 .
we suppose that the large measured value of @xmath0 in psr b1642@xmath603 is caused by the continuous generation of slow glitches over the entire period of observations 19682012 .
another point of view was presented by @xcite , who suggested that the observed oscillations in psr b1642@xmath603 can be accounted for by magnetospheric regulated dual - state switching . in figure [ resid27](c )
, it is seen that the predictability of the cycles in oscillatory behavior of this pulsar testify that the interpretation of this process by slow glitch events is preferable . *
b1133 + 16*. the bottom panel of figure [ resid27](c ) presents the timing residuals for the pulsar b1133 + 16 over the time span of 43 yr between 1969 and 2012 .
we see that the timing residuals have a quasi - periodic character with an amplitude of @xmath2840 ms and maxima spacing of @xmath2828 yr .
these residuals are characterized by a positive value @xmath70 s@xmath31 that gives a braking index @xmath71 .
the corresponding frequency residuals @xmath19 , presented in figure [ resid27](d ) , have a similar quasi - periodic structure .
the rotation frequency changes within a range @xmath19 of @xmath72 hz .
the observed quasi - periodic pattern of the timing residuals is in good agreement with that of the timing residuals , given in figure 3 of @xcite .
the timing and frequency residuals of psr b1133 + 16 after the subtraction of the polynomial for @xmath1 and @xmath0 maintain a quasi - periodic structure , as is seen in figure [ cyclic7 ] .
thus , an appreciable @xmath0 value , measured for psr b1133 + 16 , is associated with the long - term quasi - periodic variation observed in the timing residuals of this pulsar . * ( 4 ) b0329 + 54*. the timing residuals of psr b0329 + 54 are shown in figure [ resid27](e ) over the 43.5 yr time span from 1968 to 2012 .
we see that the first half of the timing residuals exhibit a long - term quasi - periodic variation that is superimposed on a cubic trend with a rather large amplitude .
the parabolic character of the @xmath19 residuals , which are plotted in figure [ resid27](f ) , also testifies that a significant @xmath0 dominates in the pulsar rotation rate over the entire interval of observations , despite a noticeable quasi - sinusoidal structure visible in the first part of the @xmath19 curve .
the observed timing residuals are characterized by a positive value @xmath73 s@xmath31 that leads to a braking index of @xmath74 .
@xcite were the first to argue for the presence of a planet around the pulsar b0329 + 54 .
the authors reported the existence of a quasi - sinusoidal modulation with a period of 3 yr in pulse arrival times and supposed that a small planet around the pulsar could cause such a modulation .
the planetary interpretation was supported by @xcite , who reported the existence of a quasi - sinusoidal modulation with a period of 16.9 yr in the timing residuals and showed that a 3 yr periodicity is clearly seen in the jpl data after removing the main 16.9 yr periodicity .
further observations by @xcite showed that there is no evidence for the presence of two planet - mass bodies on the 3 yr and 16.9 yr orbits around this pulsar .
the authors concluded that the observed quasi - periodicity in the timing residuals of this pulsar may be attributed to the variations in the rotation frequency or free precession of a neutron star .
according to @xcite , the jbo data set shows no evidence for either a 3 yr or a 16.9 yr periodicity and therefore there is no evidence for a planetary companions to this pulsar .
we agree with this conclusion of @xcite . in figure
[ cyclic7 ] , which shows the residuals for psr b0329 + 54 after the removal of a cubic term from the arrival times , we see that a 16.9 yr periodicity existed over the interval 19681995 and is not repeated in further observations after 1995 .
note that similar periodicities can be observed in the rotation frequency of other pulsars .
for example , figure [ sin1541 ] shows the timing residuals for two pulsars , b0823 + 26 and b1541 + 09 , which contain two cycles of a quasi - periodic structure . in the case of b1541 + 09
, the two cycles of the oscillatory residual curve are well described by a sinusoidal function with a period of @xmath287 yr .
the observed quasi - periodic structure is not repeated in further observations .
this example illustrates how a quasi - sinusoidal structure in the timing residuals can confuse the observer if the observational interval of the given pulsar is not sufficiently long .
it is very likely that the cubic trend seen in the timing residuals of figure [ resid27](e ) and the parabolic character of the corresponding @xmath19 residuals , presented in figure [ resid27](f ) , both are the result of a large glitch that took place in the distant past . in this case
, the measured @xmath0 value in the pulsar b0329 + 54 can be an echo of that previous event . in figure [ resid27](f )
, we see that the presence of cyclical variation in the rotation rate influences the pulsar timing behavior only on rather short intervals . we conclude that the large @xmath0 value of this pulsar is due to the presence of long - term cyclical variations and probably the influence of an odd glitch event . *
( 5 ) b0950 + 08*. figure [ resid27](e ) shows the timing residuals for the pulsar b0950 + 08 over the 43.5 yr time span from 1968 to 2012 .
the plotted timing residuals demonstrate a structure varying with time .
this structure is characterized by a negative value @xmath75 s@xmath31 that gives a braking index @xmath76 .
the @xmath19 residuals , plotted in figure [ resid27](f ) , also indicate some changes in the rotation frequency ranging from 0 to @xmath77 hz .
the removal of a cubic term from the arrival times reduces the amplitude of the timing residuals , as is shown in figure [ cyclic7 ] .
thus , we suppose that a long - term cyclical structure is the cause of the significant @xmath0 measured in the timing residuals of this pulsar . *
( 6 ) b1541 + 09*. the timing and frequency residuals for psr b1541 + 09 are given in figures [ resid27](g ) and ( h ) , respectively , over the 30 yr interval from 19822012 .
the timing residual plot shows a cyclical structure that changes its amplitude and period with time . a quasi - periodic oscillation seen over the interval 19972010
was shown on a large scale before , in figure [ sin1541 ] .
this plot shows that the two cycles of the observed quasi - periodic structure are well described by a sinusoidal function with a period of @xmath287 yr and an amplitude of 10 ms .
this periodic structure is not repeated in further observations , as follows from figure [ resid27](g ) .
the frequency residuals , plotted on the right panel , also exhibit the presence of cyclical changes in @xmath19 that are within the range of @xmath78 hz .
the observed timing residuals are characterized by a positive value @xmath79 s@xmath31 that gives a braking index @xmath80 .
figure [ cyclic7 ] shows that the timing and frequency residuals maintain a cyclical structure after the removal of a cubic term from the arrival times .
comparison with the results of @xcite shows that the signature of the prao timing residuals is generally in reasonable agreement with that of the jbo residuals , plotted in their figure 3 up to 2006 .
one cycle of the quasi - sinusoidal periodicity examined above is present at the very end of the jbo data span .
it is evident that the presence of cyclical changes in the rotation frequency is the cause of the appreciable value of @xmath0 in psr b1541 + 09 .
the resemblance of the timing residuals to the quasi - sinusoidal structure with a period of @xmath287 yr is an interesting feature of the residuals .
this result requires further research on the rotation frequency of this pulsar . *
( 7 ) b2016 + 28*. figure [ resid27](g ) displays the timing residuals for psr b2016 + 28 over the 43.5 yr time span between 1968 and 2012 . in this plot
, we see an appreciable cubic trend with an amplitude of about 30 ms .
the parabolic character of the frequency residuals , presented in figure [ resid27](h ) , also indicates the dominant influence of a rather significant @xmath0 over the entire observational interval .
the observed timing residuals are characterized by a positive value @xmath73 s@xmath31 that gives a braking index @xmath81 . in figure [ cyclic7 ] that shows the timing and frequency residuals after the removal of a cubic term from the arrival times , we see that the remaining residuals exhibit a long - term quasi - periodic structure . the signature of the timing residuals , presented in our figure [ resid27](g ) , is in agreement with that result , as shown in figure 3 of @xcite .
we suppose that the parabolic character of the @xmath19 curve for this pulsar , shown in figure [ resid27](h ) , can be a consequence of the recovery from an old discrete glitch . as in the case of psr b0329 + 54
, the deviation of the timing behavior of psr b2016 + 28 from a simple @xmath1 spin - down model can be explained by the presence of long - term quasi - periodic variations and probably the influence of a prior glitch event .
* summary*. we found that all seven pulsars whose timing residuals show a quasi - periodic signature have measurable @xmath0 values that range between @xmath82 s@xmath31 .
the structure seen in the timing residuals of these pulsars is determined by such rotational irregularities as small discrete glitches , slow glitches , long - term quasi - periodic processes , and probably the influence of prior glitch events .
the third group includes 10 pulsars whose timing residuals have noise - like variations that dominate at a level of less than 1@xmath58 of a pulse period over time spans from 21 to 34 yr .
the characteristic ages of these pulsars are more than @xmath59 yr . * ( 1 ) b0834 + 06*. the top panel of figure [ resid27](c ) displays the timing residuals for psr b0834 + 06 over the 34 yr span between 1978 and 2012 .
this pulsar has a long period of 1.27 s and a sufficiently large period derivative @xmath83 , indicating that it is a relatively young pulsar with @xmath84 yr .
the study of its rotation showed that this pulsar has very stable timing behavior in which rms residuals are @xmath28300 @xmath85s over the entire time span .
figure [ resid27](d ) shows that weak variations in the @xmath19 residuals range between @xmath86 hz .
the timing residuals of this pulsar are characterized by a positive value @xmath87 s@xmath31 .
this value yields a braking index @xmath88 that is very close to the canonical braking index @xmath4 , expected from the simple magnetic dipole braking .
it is very likely that the timing noise in this pulsar can be accounted for by magnetic dipole braking . in this case
, a measurement of the deterministic @xmath0 value will be possible in the near future . *
( 2 ) b2110 + 27*. figure [ resid27](e ) exhibits the timing residuals for this pulsar over the 28 yr time span from 1984 to 2012 .
the observed residuals take the form of a cubic polynomial with a very small amplitude of @xmath284 ms .
figure [ resid27](f ) shows that variations in the @xmath19 residuals are rather small and range between @xmath89 hz .
the timing residuals are characterized by a positive value @xmath90 s@xmath31 that yields a braking index @xmath91 .
we see that the observed shape of the timing residuals resembles a damped cubic trend .
* ( 3 ) b2303 + 30*. figure [ resid27](e ) displays the timing residuals for psr b2303 + 30 over the 21 yr span between 1991 and 2012 .
a weak cubic trend is visible in these residuals .
the observed variations in the @xmath19 residuals are within the range of @xmath92 hz , as is shown in figure [ resid27](f ) .
the timing residuals are characterized by a negative value @xmath93 s@xmath31 that gives a braking index @xmath94 . *
( 4 ) b1919 + 21*. the timing residuals for psr b1919 + 21 are presented in figure [ resid27](e ) over the 34 yr data span from 1978 to 2012 .
the pulsar shows a weak timing noise in which rms residuals are @xmath281 ms .
the @xmath19 residuals , plotted in figure [ resid27](f ) , exhibit random variations that range between @xmath95 hz .
the observed timing residuals are characterized by a negative value @xmath96 s@xmath31 that gives a braking index @xmath97 . *
( 5 ) b1839 + 56*. the bottom panel of figure [ resid27](e ) exhibits the timing residuals of psr b1839 + 56 over the 29 yr data span from 1983 to 2012 .
the timing residuals take the form of a cubic polynomial with an appreciable amplitude of about 7 ms , which is still less than 0.5% of a pulse period .
the corresponding changes in the @xmath19 residuals , plotted in figure [ resid27](f ) , are weak and are within a range @xmath19 of @xmath89 hz .
we may be observing the damped oscillations of a more long - term cyclical process .
the timing residuals are characterized by a negative value @xmath98 s@xmath31 that gives a braking index @xmath99 . *
( 6 ) b2315 + 21*. the top panel of figure [ resid27](g ) shows the timing residuals for this pulsar over the 21 yr data span from 1991 to 2012 .
the timing residuals exhibit a noise - like structure .
the corresponding changes in the @xmath19 residuals , presented in figure [ resid27](h ) , range between @xmath100 hz .
the timing residuals are characterized by a negative value @xmath101 s@xmath31 , that gives a braking index @xmath102 . *
b0138 + 59*. the timing residuals for this pulsar are shown in figure [ resid27](g ) over the 29 yr time span from 1983 to 2012 .
a noise - like structure is visible in the timing residuals .
the variations of the @xmath19 residuals , plotted in figure [ resid27](h ) , are within a small range @xmath19 of @xmath95 hz .
the timing residuals are characterized by a positive value @xmath103 s@xmath31 that gives a braking index @xmath104 . *
( 8 ) b1821 + 05*. figure [ resid27](g ) displays the timing residuals for psr b1821 + 05 over the 21 yr time span between 1991 and 2012 .
the timing residuals show a noise - like structure .
figure [ resid27](h ) shows that the variations of the @xmath19 residuals are within a range of @xmath100 hz .
the timing residuals are characterized by a positive value @xmath105 s@xmath31 that gives a braking index @xmath106 . *
( 9 ) b0320 + 39*. figure [ resid27](g ) shows the timing residuals for this pulsar over the 32 yr time span from 1980 to 2012 .
noise - like variations are visible in the timing residuals .
the variations in the @xmath19 residuals range between @xmath95 hz , as is shown in figure [ resid27](h ) .
the timing residuals are characterized by a positive value @xmath103 s@xmath31 that gives a braking index @xmath107 . *
b0809 + 74*. the bottom panel of figure [ resid27](g ) displays the timing residuals for psr b0809 + 74 over the 33 yr data span from 1979 to 2012 .
the oldest pulsar in our sample demonstrates timing residuals with a noise - like structure .
figure [ resid27](h ) shows that variations in the @xmath19 residuals range between @xmath89 hz .
the timing residuals are characterized by a positive value @xmath108 s@xmath31 .
thia value yields a braking index of 3717 .
figure [ quiet10 ] shows the timing residuals for the 10 pulsars after the removal of a cubic term from the pulse arrival times .
we see that most pulsars in this group exhibit random variations in the remaining timing residuals .
* summary*. we found that all 10 pulsars whose timing residuals are dominated by noise - like variations show @xmath0 values that range between @xmath109 s@xmath31 .
the measured @xmath0 values are on the boundary of the precision of our measurements , which is estimated to be @xmath110 s@xmath31 .
we suppose that the observed @xmath0 values for these pulsars arise because of the limited precision of our measurements , which is due to short observing intervals . as discussed in @xcite , hundreds of years of observations are required to measure @xmath0 due to deterministic spin - down law in the middle - aged pulsars .
the fourth group includes two pulsars whose timing residuals exhibit more complex structure compared with the previous groups .
the main feature of the rotation of these pulsars is that their rotation parameters @xmath111 can undergo a rapid change .
this change is accompanied by a change in the sign of the second frequency derivative @xmath0 . as a consequence of this phenomenon ,
the rotation behavior of these pulsars can not be described by one @xmath111 timing model within one pulse period over a long time span of tens of years .
glitch events have nothing to do with this situation .
we show that two different timing models , each of which includes @xmath0 of an opposite sign , are required for the description of the rotation behavior of these pulsars over the 43 yr span of data analyzed .
the characteristic age of psr b0823 + 26 is @xmath112 yr and that of b1929 + 10 is @xmath113 yr .
figure [ resid27](c ) displays the timing residuals for psr b0823 + 26 over the 43 yr span from 1969 to 2012 .
the combined data set includes the jpl data between 1969 and 1983 and the prao data between 1981 and 2012 .
a detailed analysis of the data showed that all the pulse arrival times gathered over the 19692012 interval can not be described by a simple @xmath1 model within a pulse period because of the presence of a large @xmath0 .
these arrival times also can not be described within a pulse period by a timing model including the second derivative @xmath0 .
the cause of this situation is not connected with glitch events . in order to demonstrate the complex character of the rotation behavior of this pulsar over the entire observational period 19692012
, we plotted the timing residuals relative to a simple spin - down model 19811986 , as is shown in figure [ resid27](c ) . from this plot
, we see that the deviation of the residual curve with respect to a zero line comprises approximately 1700 ms , more than three pulse periods . in order to keep track of the pulsar rotation over a long time span , we eliminated the phase jump in the residual curve by adding one period when the curve deviated by more than one pulse period .
the cause of such a deviation becomes more clear if we consider the frequency residuals shown in figure [ resid27](d ) . here
, the @xmath19 residuals are presented relative to the same spin - down model 19811986 .
we see that the @xmath19 curve has a distinct break in 1974 november ( mjd 42364 ) .
the break separates two regions with different rotation frequency behaviors and is a cross point of the two curves having convex and concave shapes .
it is known that such curves are characterized by opposite signs of @xmath0 .
this property means that the break in the frequency residuals is accompanied by a change in the sign of the second frequency derivative @xmath0 .
further analysis confirms that the sign of @xmath0 is reversed at this point .
our purpose is to show that : ( 1 ) the two timing models , including @xmath0 of the opposite signs , are required to describe the rotation behavior of psr b0823 + 26 over the entire 19692012 interval , ( 2 ) the break in the @xmath19 curve causes a change in the sign of the second derivative @xmath0 , and ( 3 ) the break in the @xmath19 curve causes a distinct jump in the first derivative @xmath20 .
figure [ sign0823 ] shows a detailed picture of the timing behavior of the pulsar b0823 + 26 .
the top panel demonstrates how the entire data set for this pulsar can be described by two independent timing models , including @xmath0 with opposite signs .
the first interval * ( i ) * indicates a maximum interspace from the first point , in which the data set can be described by one timing model of @xmath15 , @xmath20 , and @xmath0 within half a pulse period .
this interval is 19691984 ( mjd 4026445873 ) .
the obtained timing residuals are marked by open circles .
an increase of this interspace to the right at one following point from the data set already causes a phase jump of more than a pulse period .
the second interval * ( ii ) * indicates a maximum interspace from the last point of the data set ( the rightmost point on the @xmath114-axis ) to the earlier point in the data set ( the leftmost point on the @xmath114-axis ) , in which the data set can be well described by another timing model of @xmath15 , @xmath20 , and @xmath0 within half a pulse period .
the indicated interval is 19712012 ( mjd 4127955973 ) .
the obtained timing residuals are marked by solid circles .
an increase of this interspace to the left at one previous point from the data set already causes a phase jump of more than a pulse period .
the measured rotation parameters for these timing models include second derivatives @xmath0 with opposite signs : @xmath115 s@xmath31 over the 19691984 interval and @xmath116 s@xmath31 over the 19712012 interval .
so , we have showed that two timing models are required to describe the pulsar s rotation behavior over the 19692012 interval and these models include @xmath0 of opposite signs .
the middle panel of figure [ sign0823 ] shows how a change in the sign of @xmath0 has an effect on the rotation frequency of the pulsar .
the vertical line marks the time of the break in the frequency residuals which , according to figure [ resid27](d ) , took place in 1974 november ( mjd 42364 ) .
the plotted frequency residuals @xmath19 are presented relative to the two @xmath111 timing models : 19691974 ( before the break ) and 19742012 ( after the break ) .
the mean rotation parameters for these timing models are given in table [ param27 ] , from which it can be seen that these models include second derivatives of @xmath0 with opposite signs .
at first , we present the @xmath19 residuals relative to the first timing model 19691974 , including @xmath117 ( open circles ) . from this plot
, we see that the @xmath19 residuals correspond well with the zero line in section a ( before the break ) . beyond section
a , the @xmath19 curve sharply deviates with respect to the zero line .
this result means that the values of @xmath15 calculated from the model 19691974 do not correspond to the actual rotation frequencies existing after the break .
a similar picture is observed for the other model 19742012 , including @xmath118 ( solid circles ) . in section
b ( after the break ) , the @xmath19 residuals correspond well with the zero line .
ahead of section b , the @xmath19 curve abruptly declines downwards with respect to the zero line .
both of these plots , as well as figure [ resid27](d ) , show that the point of the break is formed as a result of a sudden change in the slope of the @xmath19 curve within the framework of the given model .
the sharp turn in the @xmath19 curve is the cause of a change in the sign of the second derivative @xmath0 .
the break is a rapid event as it happens within @xmath28300 days , which is the time resolution .
so , we have shown that the break in the @xmath19 curve is accompanied by a change in the sign of the second frequency derivative @xmath0 .
the bottom panel of figure [ sign0823 ] shows the first frequency derivative @xmath20 as a function of time .
the time behavior of @xmath20 is identical for all the timing models that were mentioned above .
the observed changes in the @xmath20 behavior are the effect of the changes in the @xmath19 curve .
this statement is confirmed by a visual comparison of this @xmath20 plot with the @xmath19 plot , presented in figure [ resid27](d ) . from the bottom panel of figure [ sign0823 ]
, we see that the observed appearance of the @xmath20 curve testifies to a change in the sign of @xmath0 .
the long - term behavior of the @xmath20 curve can be described as a fast increase of @xmath119 before the break ( section a ) , a distinct jump in the magnitude of @xmath20 at the point of the break , and the gradual decrease of @xmath119 after the break ( section b ) .
the break in the @xmath19 curve produces a distinct jump in the first derivative @xmath20 .
the magnitude of the jump in @xmath20 equals @xmath120 s@xmath121 .
it makes up @xmath122 of the mean value of @xmath123 s@xmath121 .
so , we have shown that the break in the @xmath19 curve causes a jump in the first derivative @xmath20 .
it should be noted that the jump in @xmath20 for the pulsar b0823 + 26 was known earlier .
@xcite detected an event in @xmath20 in early 1975 ( jd 2442500 ) .
this time agrees well with our time of the break in the @xmath19 curve ( jd 2442364 ) .
according to their figure 11 , the size of the jump equals @xmath124 , which agrees well with our measurement of the jump in @xmath20 .
@xcite also noticed that b0823 + 26 exhibits a large discontinuity in both @xmath125 and @xmath126 near jd 2442600 .
these events are clearly visible in their figure 6 . later , the timing noise of psr b0823 + 26 was analyzed by @xcite .
the break in the frequency residuals of this pulsar is clearly seen in their figure 1 .
the time of the break can be estimated from this plot to be @xmath28 jd 2442400 .
this time agrees well with our result .
we pay attention to the short - term quasi - periodic variations seen in the timing behavior of psr b0823 + 26 in all three panels of figure [ sign0823 ] .
two cycles of a quasi - periodic structure of the timing residuals over the interval 19912008 were shown before on a large scale in figure [ sin1541 ] .
short - term quasi - periodic variations in the behavior of @xmath20 , seen clearly in the bottom panel of figure [ sign0823 ] , are the cause of variations in the @xmath0 values measured during different time spans .
this plot illustrates why the measured values of @xmath0 may be different at different parts of the data span analyzed , as was earlier remarked by @xcite ( see their figure 7 ) . on the contrary ,
the long - term behavior of the second derivative @xmath0 is associated with a linear trend that is observed in the first derivative @xmath20 over the entire 38 yr time span between 1974 and 2012 .
the pulsar b1929 + 10 is the second pulsar in our research , after psr b0823 + 26 , for which the phenomenon of a change in the sign of the second derivative @xmath0 was revealed .
figure [ resid27](c ) shows the timing residuals for b1929 + 10 over the 43 yr data span from 1969 to 2012 .
this plot includes the jpl data spanning between 1969 and 1982 and the prao data spanning between 1999 and 2012 .
a 16 yr data gap seen in the residual curve complicates the analysis of the rotation frequency for this pulsar .
nevertheless , there are strong arguments for the existence of a change in the sign of @xmath0 for this pulsar .
figure [ resid27](c ) presents the timing residuals relative to a simple @xmath1 model 19691982 .
we see that the residuals have different signatures for the two observational intervals . in the second interval over 19992012 ,
the residual curve suddenly changed its slope with respect to the zero line . at the end of this interval ,
the deviation of the timing residuals exceeded 1000 ms , more than five pulse periods .
comparing of this plot with the plot of the timing residuals presented in figure 3 of @xcite shows that there is good agreement between the picture of the prao timing residuals and that of the jbo residuals despite a 16 yr gap in our data set . in figure 3 of @xcite
, we see that glitch events were absent during a 16 yr gap between 1982 and 1999 .
therefore , the observed change of the slope of the residual curve is not connected with glitches .
the variable character of the timing residuals is probably due to a sudden change in the conditions of the pulsar rotation .
estimates show that a sudden change in the pulsar rotation parameters could have taken place in approximately 1997 .
the picture of the frequency residuals , plotted in figure [ resid27](d ) , also indicates a variable behavior of the rotation frequency .
we clearly see that the @xmath19 residuals , presented relative to the timing model 19691982 , show different structures for the two observational intervals 19691982 and 19992012 . in order to reveal the cause of the dissimilar timing behavior of psr b1929 + 10 over the two intervals , we consider figure [ sign1929 ] . figure [ sign1929](a )
presents the timing residuals relative to the simple @xmath1 timing model 19691982 ( on the left side ) and relative to the simple @xmath1 timing model 19992012 ( on the right side ) .
we see that the post - fit residuals exhibit a weak cyclical structure over the first interval and a large cubic term over the second interval .
the cubic term indicates a large positive value of @xmath0 .
this result means that the change of conditions in the pulsar rotation have resulted in a large @xmath0 value . in the following two plots , we present evidence that a different signature of the @xmath19 residuals over the two observational intervals is due to a change in the value and the sign of the second frequency derivative @xmath0 .
figures [ sign1929](b ) and ( c ) display the frequency residuals @xmath19 relative to the @xmath111 timing models 19691982 and 19992012 , respectively .
the mean rotation parameters for these two models are given in table [ param27 ] , which clearly shows that the models include @xmath0 of opposite signs .
figure [ sign1929](b ) shows that the @xmath19 residuals relative to the first timing model have a quasi - periodic character over the interval 19691982 and show a significant deviation with respect to the zero line over the second interval .
a parabolic form of the @xmath19 residuals over the 19992012 interval indicates the presence of a large @xmath0 , considerably exceeding @xmath0 of the first interval .
still greater discordance between the calculated and actual rotation frequencies is observed for another timing model , plotted in figure [ sign1929](c ) .
the frequency residuals @xmath19 relative to the second @xmath111 model have a noise - like character over the second interval 19992012 and show a very large deviation with respect to the zero line over the first interval .
this result means that the calculated rotation frequencies @xmath15 can not be compared to the actual rotation frequencies existing for the first interval .
both of these plots support a sudden change in the magnitude of @xmath0 between the two observational intervals .
the timing behavior of the first derivative @xmath20 is presented in figure [ sign1929](d ) .
we see that the observed signature of @xmath20 differs between the two intervals . in the second interval ,
the magnitude @xmath119 decreases with time , indicating the presence of a large positive second derivative @xmath0 . in this plot ,
the solid line indicates the best fit with a linear function .
we have presented evidence that the pulsar b1929 + 10 experienced a significant change in its rotation parameters between 1982 and 1999 .
a rapid change of the parameters @xmath1 , and @xmath0 was accompanied by a change in the sign of @xmath0 .
this event could have taken place around 1997 ( @xmath28mjd 50400 ) ; this year is marked by a vertical line in figure [ sign1929 ] .
the scheme of the time behavior of @xmath0 for two pulsars , b0823 + 26 and b1929 + 10 , is plotted in figure [ modulff ] . as was shown above , the observed change in the value and
the sign of @xmath0 is due to a rapid change of the pulsar rotation parameters @xmath1 , and @xmath0 . from this plot
, we see that the time behavior of @xmath0 for psr b0823 + 26 is well described by a rectangular function . for the other pulsar , b1929 + 10 ,
the behavior of @xmath0 is shown by the same function .
the plot suggests that we probably observe a small fragment of some long - term process that can modulate the second derivative @xmath0 in pulsars in such a way that the sign of @xmath0 is reversed .
if such a modulation of @xmath0 is a periodical process , then the time behavior of @xmath0 could be described by the periodic rectangular function over long time scales . in our observations , the phenomenon of a change in the sign of @xmath0 is observed for two pulsars out of the 27 studied .
this result means that the period of possible modulation in @xmath0 insignificantly exceeds the 43 yr time span of our observations by two to four times and probably reaches 100200 yr .
so , we have revealed a phenomenon of a rapid change of the pulsar rotation parameters @xmath15 , @xmath20 , and @xmath0 .
a rapid change in the rotation frequency @xmath15 is accompanied by a distinct jump in @xmath20 and a change in the value and the sign of @xmath0 .
the indication of the existence of such a phenomenon is that the timing behavior of the pulsar can not be described by one @xmath111 timing model within one pulse period during a long period of observations spanning decades . for this case ,
two models that include @xmath0 values of opposite signs are required for the fitting procedure .
glitches have nothing to do with this situation . we suppose that in the future this phenomenon a rapid change of pulsar rotation parameters accompanied by a change in the sign of @xmath0will be revealed in the rotation frequency of many other pulsars .
figure 3 of @xcite presents the timing residuals for a sample of 366 pulsars .
the large cubic trends seen in the residuals of many pulsars , presented in this plot , can have different origins .
some part of these cubic trends may result from glitch events .
the other part of them may serve as indicators that these pulsars experienced a rapid change in the conditions their rotation .
the cubic trend visible in the timing residuals of psr b1929 + 10 , shown in our figure [ sign1929](a ) over the interval 19992012 , indicates this case .
figure [ zavis27 ] shows the relation between the measured value of the second frequency derivative @xmath0 and the characteristic age @xmath127 for the 27 studied pulsars .
these parameters are listed in table [ deriv27 ] .
we see that the plotted points form three sequences according to the signatures of the timing residuals observed in these pulsars , as was discussed above .
a detailed inspection of figure [ zavis27 ] shows that certain signatures of the timing residuals are correlated with certain values of the second derivatives @xmath0 and certain age ranges .
this correlation implies that pulsar timing irregularities have an evolutionary character .
we suggest that the three plotted sequences , corresponding to different kinds of signatures in the timing residuals , indicate the three evolutionary stages of pulsar rotation . in other words , these sequences show that certain signatures in the timing residuals can occur only at a certain stage of pulsar rotation evolution . from this fact
, it follows that each evolutionary stage is characterized by certain types of rotational irregularities . * discrete glitches and long - term quasi - periodic oscillations . *
the extensive jbo timing observations and observations of other researchers have shown that phenomena such as discrete glitches and quasi - periodic processes represent two main types of rotational irregularities in pulsars @xcite .
glitches occur as sudden , discrete jumps in a pulsar s rotation frequency , followed by an exponential recovery to the pre - glitch value .
cubic trends in the timing residuals corresponding to the positive values of @xmath0 arise as a result of relaxation after a discrete glitch @xcite . to date , a total of 422 glitches have been revealed in 138 pulsars @xcite .
the second type of rotational irregularities occurs in the form of long - term quasi - periodic oscillations . in figure 3 of @xcite
, we see that many pulsars show different quasi - periodic structures in their timing residuals .
both types of rotational irregularities can occur in the rotation frequency of one pulsar . according to @xcite ,
the timing residuals of many pulsars represent a combination of a cubic trend and a quasi - periodic structure . from our data ,
this effect is clearly visible in figure [ cubic6 ] .
the examples of quasi - periodic changes seen in the timing residuals of the pulsars in our sample are shown in figure [ cyclic7 ] . *
slow glitches . * slow glitch phenomena present the third type of rotational irregularities .
the term `` slow glitch '' was introduced by @xcite to explain a glitch of an unusual , gradual signature that occurred in psr b1822@xmath609 over the interval 19951997 . according to the literature ,
the second slow glitch in this pulsar was observed by @xcite and the third glitch was independently observed by @xcite and @xcite .
the oscillatory process in psr b1822@xmath609 was finished in 2004 .
slow glitches were also investigated in detail in the rotation frequency of psr b1642@xmath603 and b0919 + 06 @xcite .
these results showed that slow glitch events occur in series ; they produce regular oscillations in a pulsar s rotation frequency that looks like a continuous sequence of glitch - like events .
these oscillations in the rotation frequency give rise to rapid , regular oscillations in the timing residuals .
as reported by @xcite , slow glitches have yet to be identified for two pulsars , j0631 + 1036 and b1907 + 10 .
an extension of this work would be to study the properties of these events and to find evidence that they are glitch events , rather than timing noise .
according to @xcite , slow glitches are not a unique phenomenon and they can be caused by the same process that causes timing noise in pulsars .
another point of view was suggested by @xcite .
these authors proposed that it is reasonable to label as slow glitches the observed discrete events that result in quasi - sinusoidal features in the timing residuals .
this fact means that the idea of slow glitches has received confirmation from @xcite .
apparently , the phenomenon of slow glitches is not widespread among pulsars .
a sample of 10 pulsars whose timing residuals demonstrate a regular cyclical structure is shown in figure 1 of @xcite . according to a model proposed by these authors ,
these timing data are explained by quasi - periodic switching in the spin - down rate of these pulsars . as discussed by @xcite ,
the harmonic structure seen in the timing residuals of these 10 pulsars is consistent with the precession interpretation .
we suppose that most pulsars from this sample are candidates for having slow glitches in their rotation frequencies . it should be noted that regular oscillations were observed earlier in the rotation rate of the crab pulsar from jbo observations over the interval 19821989 .
the period of a regular component of the timing residuals was estimated to be @xmath28 20 months @xcite or @xmath128 days @xcite .
@xcite claimed that the observed cyclical residuals with this period were consistent with a physically real quasi - periodic process . in our opinion
, the periodic process with the 568 days period observed in the crab pulsar resembles the phenomenon of slow glitches producing a periodic sawtooth - like modulation with a period of 600 days in the timing residuals of the pulsar b0919 + 06 @xcite .
it is quite possible that the phenomenon of slow glitches presents an intermediate position between discrete glitches and quasi - periodic processes . on the one hand ,
slow glitch events show properties that should characterize them as discrete glitch events .
for example , a linear relation between the amplitudes of glitches and the post - glitch intervals was revealed both for the x - ray pulsar j0537@xmath66910 @xcite and for the pulsar b1642@xmath603 ( see figure 3 of @xcite ) . on the other hand
, a sequence of slow glitches produces regular cyclical changes in the timing residuals .
regular oscillations may be a particular case of more general phenomenon related to long - term quasi - periodic oscillations observed in the rotation rate of many pulsars . in any case , the phenomenon of slow glitches is due to real oscillatory processes rather than random variations in the pulsar rotation rate . * a phenomenon of a sudden change in the sign of the second derivative @xmath0 . * in section 3.2 , we reported the detection of a new phenomenon that is associated with a change of pulsar rotation parameters @xmath1 , and @xmath0 .
these parameters change in such a way that the sign of @xmath0 is reversed , as is shown in figure [ modulff ] .
this phenomenon was detected in the timing behavior of two pulsars , b0823 + 26 and b1929 + 10 .
it is very likely that a similar rapid change of rotation parameters can occur in many other pulsars .
if such a process cycles in time , then it can be considered a modulation process .
a rough estimate shows that a change of rotation parameters can occur in the rotation rate of some pulsars about once every 100 yr . the sign of @xmath0 will change to the opposite sign with the same interval .
this process can be a candidate to represent the fourth type of rotational irregularities . in figure [ zavis27 ]
, we see that the pulsars that show cubic components in their timing residuals ( marked by solid circles ) can be thought of as being in the first stage of pulsar rotation evolution .
these pulsars have large values of @xmath0 with both signs and are the youngest pulsars in our sample .
the given sequence of pulsars also includes two pulsars , b0823 + 26 and b1929 + 10 ( marked by squares ) , whose timing behavior is determined by a rapid change of the rotation parameters and a change in the sign of @xmath0 . in figure [ cubic6 ] , we see that all these pulsars also have a quasi - periodic structure seen in their timing residuals after the removal of a cubic trend from the arrival times . at the first stage , there are also slow glitch events . as was discussed above , two pulsars from the first group , b1822@xmath609 and b0919
+ 06 , show a regular cyclical structure in their timing residuals which are due to slow glitches .
we conclude that rotational irregularities in the first stage of pulsar rotation evolution are produced by such phenomena as discrete glitches , slow glitches , quasi - periodic processes , and processes of a rapid change of pulsar rotation parameters .
figure [ zavis27 ] shows that the pulsars that stay in the second stage of pulsar rotation evolution ( marked by star symbols ) show timing residuals that are dominated by quasi - periodic components .
these pulsars are older pulsars , although their age partially overlaps with the ages of the pulsars remaining in the first stage .
these pulsars also have considerably smaller measured @xmath0 values .
the timing residuals of these pulsars exhibit a great variety of quasi - periodic structures , as is seen in figures [ resid27 ] and [ cyclic7 ] .
besides quasi - periodic processes , small discrete glitches and slow glitches can also occur in the rotation frequency of these pulsars .
for example , the cyclical timing residuals of psr b1642@xmath603 are due to the presence of slow glitches .
as discussed above , the pulsar b2217 + 47 experienced a small discrete glitch of magnitude @xmath62 that occurred in 2011 october .
it is very likely that in the second stage of rotation evolution , events such as large discrete glitches already will not occur .
the process of rapid changes of pulsar rotation parameters also disappears .
we conclude that rotational irregularities in the second stage of pulsar rotation evolution are produced by such phenomena as quasi - periodic processes , small discrete glitches , and slow glitches .
these processes are the cause of measurable values of @xmath0 for these pulsars .
the third group ( marked by triangle symbols ) includes pulsars whose timing residuals exhibit noise - like variations . in figure
[ zavis27 ] , we see that the ages of these pulsars almost completely overlaps with the ages of the pulsars remaining in the second stage .
this result indicates that pulsars of the same age can remain in different stages of rotation evolution .
the pulsars that belong to the third stage of pulsar rotation evolution have many properties in common ( see figure [ quiet10 ] ) .
the periods of these pulsars usually are more than one second .
the timing residuals mainly are due to measurement errors , although some pulsars of this group can show a weak remnant structure , as in the case of psr b1839 + 56 . in figure
[ zavis27 ] , we see that the measured @xmath0 values for these pulsars are on the boundary of the precision of our measurements , which is equal to @xmath129 s@xmath31 . the limited span of timing data does not allow us to measure the deterministic @xmath0 values , which for most pulsars from this group should be less than @xmath130 s@xmath31 . we conclude that rotational irregularities in the third stage of pulsar rotation evolution should gradually disappear . at this stage in the timing residuals ,
noise - like variations dominate .
it follows that in the future these pulsars will never be observed to glitch .
an analysis of the data , plotted in figure [ zavis27 ] , shows that the age boundaries between different evolutional stages are indistinct and diffusive .
this result implies that the rotation rates of pulsars of the same age can evolve along different paths . in figure [ zavis27 ] and table [ deriv27 ]
, we see that the middle - aged pulsar b0834 + 06 is approximately the same age as the pulsars b2217 + 47 and b1929 + 10 .
in contrast to these pulsars , b0834 + 06 stays in the third stage of pulsar rotation evolution .
this example illustrates that pulsars of the same age can have different evolution rotation histories . from figure [ zavis27 ]
, we may estimate the age range for each evolutionary stage .
the younger pulsars from our sample with characteristic ages between @xmath131 and @xmath132 yr belong to the first evolutionary stage .
the second stage includes the middle - aged pulsars with ages greater than @xmath133 yr .
the third evolutionary stage includes both the middle - aged and old pulsars but the middle - aged pulsars in the third stage exhibit much less timing noise than pulsars of the same age in the second evolutionary stage .
the uncertainty of the age boundaries between different evolutionary stages complicates the correlation between the amount of timing noise and pulsar age .
a detailed inspection of figure [ zavis27 ] shows that rotation rates of pulsars do not always pass through all three evolutionary stages .
some pulsars can stay in the first stage during their whole lives . on the contrary
, some pulsars can escape the first stage of rotation evolution if they never experience large glitches .
pulsars that have stable rotation over their whole lives can at once fall into the third evolutionary stage .
the last feature of rotation evolution implies that , despite the limited accuracy of our measurements , the deterministic values of @xmath0 can be measured for some middle - aged pulsars that stay in the third evolutionary stage . among these pulsars
, there will always be pulsars that did not experience any shocks in their life and consequently have stable rotation and a low level of timing noise .
for example , we examine the pulsar b0834 + 06 .
this middle - aged pulsar has a relatively low level of timing noise , as discussed above ( see figure [ resid27](c ) ) .
we may suppose that neither discrete glitches nor quasi - periodic oscillations disturbed the rotation of this pulsar during its life .
figure [ zavis27 ] shows that the measured @xmath0 value for this pulsar falls very close to the sloping line of the deterministic values of @xmath0 , indicating that slowdown of the rotation of this pulsar is closely related to a simple @xmath1 spin - down model .
we conclude that a measurement of the deterministic value of @xmath0 for this pulsar is possible in the near future rather than in hundreds of years as for the other older pulsars .
evolution of the magnetic field of a neutron star was discussed in detail in a series of papers @xcite .
we consider the evolutionary picture of the magnetic field for the middle - aged and old pulsars taking into account the signatures of the timing residuals observed in these pulsars .
figure [ zavb0](a ) shows the dependence of the surface magnetic field strength @xmath134 on the characteristic age @xmath127 for the 27 studied pulsars .
the parameters @xmath134 and @xmath127 are listed in table [ deriv27 ] .
the three plotted sequences indicate that the magnetic field values @xmath134 considerably differ for the three groups of pulsars and that these values are furthemore correlated with certain signatures of the timing residuals . here , a sequence of the @xmath134 values for the pulsars with cubic signatures of the timing residuals is marked by solid circles , pulsars with quasi - periodic signatures in their residuals are marked by star symbols , and pulsars with noise - like signatures in their residuals are marked by triangle symbols .
we see two surprising features in the observed correlation picture .
first , the @xmath134 values of the pulsars showing noise - like components in their timing residuals ( the @xmath135 values ) , is are greater than the @xmath134 values of the pulsars of the same age that show timing residuals dominated by cubic components ( the @xmath136 values ) or quasi - periodic components ( the @xmath137 values ) .
this result may mean that magnetic fields are stronger for pulsars whose rotation is more stable .
second , a sequence of @xmath137 values continues well into a sequence of the @xmath136 values .
this result indicates a common evolutionary path for these two groups of pulsars . for analysis ,
these two sequences are combined into one sequence @xmath138 , as is shown in figure [ zavb0](b ) .
we see that the two straight lines fit to the sequences of @xmath139 points and @xmath138 points have identical slopes . for pulsars of the same age , the @xmath139 values nearly four times greater than the @xmath138 values .
this result suggests that the magnetic fields are approximately four times stronger for the pulsars that have stable rotation .
the dependence of the @xmath134 values on different kinds of signatures of the timing residuals is a good indication of the validity of the supposition that the rotation rates of pulsars pass through different evolutionary stages and that some pulsars can have a special path of evolution that is characterized by a very stable pulsar rotational behavior during the whole pulsar s life .
we conclude that the observed correlation picture of the surface magnetic field strength @xmath134 with certain signatures of the timing residuals , presented in figure [ zavb0 ] , provides convincing evidence for the existence of the evolutionary scenario of the origin of rotational irregularities . in order to check the results obtained above for a large number of pulsars
, we analyzed the jbo timing residuals , presented in figure 3 of @xcite .
we tried to classify ordinary pulsars of middle and older age according to the signatures of their timing residuals .
the obtained results are shown in figure [ zavjbo ] .
figure [ zavjbo](a ) displays a relation between the measured values of @xmath0 and the characteristic age @xmath127 . here , 73 selected pulsars showing timing residuals dominated by cubic and quasi - periodic components are marked by squares and 64 pulsars showing noise - like components in their timing residuals are marked by open circles .
the plotted three sequences present the prao data , as in figure [ zavis27 ] .
we see that the distribution of the plotted points does not contradict the supposition about the existence of three stages of pulsar rotation evolution .
figure [ zavjbo](b ) displays the relation between the surface magnetic field strength @xmath134 and @xmath127 for the selected pulsars from the jbo sample .
the plotted two sequences present the prao data , as in figure [ zavb0](b ) . in this plot
, the two straight lines fit to two groups of jbo points show that magnetic fields are stronger for pulsars exhibiting noise - like signatures in their timing residuals .
this result confirms that pulsars that stay in the third evolutionary stage have a tendency to show stronger magnetic fields than pulsars in other stages .
we conclude that the evolutionary scenario of the occurrence of rotational irregularities is confirmed by the dependencies that are presented in figure [ zavjbo ] for a large number of pulsars .
we suppose that the large values of @xmath0 with both signs are the result of the existence of four types of rotational irregularities that have evolutionary characteristics and form three evolutionary stages in pulsar rotation rate .
rotational irregularities alone are responsible for the large values of the measured second derivative @xmath0 and the variable structure of the timing noise over a long time scale . as discussed above
, rotational irregularities occur in the form of discrete glitches , quasi - periodic oscillatory processes , slow glitches , and processes that change the pulsar rotation parameters @xmath1 , and @xmath0 in such a way that the sign of @xmath0 is reversed .
an analysis of timing noise showed that several types of rotational irregularities can simultaneously occur in the rotation rate of one pulsar .
for example , the rotation frequency of psr b0919 + 06 over the interval 19792009 underwent large changes that were due to three types of rotational irregularities including a series of 19 slow glitches , a long - term quasi - periodic process , and a large discrete glitch @xcite .
we conclude that rotational irregularities are the cause of the deviation of the timing behavior of most pulsars from a simple @xmath1 model .
this conclusion is in accordance with the statement by @xcite that the timing noise can not be explained by models that are based on random walks in the pulse phase , frequency , or spin - down rate .
such models are not consistent with the observations .
it is known that the @xmath0 values for the simple model of magnetic dipole braking should be positive . however ,
timing observations show that pulsars with characteristic ages more than @xmath140 yr have both the positive and negative @xmath0 values .
as pointed out by @xcite , approximately 52% of these pulsars have a positive @xmath0 value and 48% have negative @xmath0 .
these authors noted that this situation is not related to glitch recovery nor magnetic dipole radiation , but is a result of some other process .
we propose that we have found a process that can result in a change in the sign of @xmath0 in pulsars .
as was reported in section 3.2 , this process changes the pulsar rotation parameters together with a change in the sign of @xmath0 , as shown in figure [ modulff ] .
this process was detected in the timing behavior of two pulsars , b0823 + 26 and b1929 + 10 .
it is very likely that a similar rapid change of rotation parameters can occur in many other pulsars .
if so , then this process could explain the problem that approximately equal numbers of pulsars have timing residuals in the form of a cubic polynomial with positive ( 20% ) and negative ( 16% ) signs of @xmath0 @xcite . while glitch recovery produces cubic trends corresponding to the positive values of @xmath0 , this process could cause large values of @xmath0 with both signs .
it should be noted that such a process apparently can operate only in pulsars that stay in the first stage of pulsar rotation evolution . in the final stages of rotation evolution , negative signs of @xmath0
can result from the limited temporal baseline of data of a more long - term oscillatory process . in the literature ,
there are different interpretations of the origin of large @xmath0 values . in the paper by @xcite
, the large values of @xmath0 with both signs are attributed to small variations in the spin - down torque .
@xcite discussed this problem in the context of the pulsar spin - down model that involves the existence of a long - term cyclical process changing the observed rotation parameters .
this cyclical process is superimposed on the secular spin - down of pulsars and produces cyclical variations in @xmath0 with time scales of a few thousand years .
the second interpretation is in accordance with our experimental result describing the fourth type of rotational irregularities .
this type of irregularity produces a change in the sign of @xmath0 owing to a rapid change of pulsar rotation parameters . however , we explain the origin of the large @xmath0 values and the variable structure of the timing noise by the existence of four types of rotational irregularities that have evolutionary characteristics
. we propose an evolutionary scenario for the occurrence of rotational irregularities to explain the observed properties of timing noise in pulsars .
timing noise is common among pulsars because rotational irregularities occur in pulsars of all ages , but the strength of timing noise depends on the evolutionary stage of pulsar rotation , as is shown in figure [ zavis27 ] .
younger pulsars stay in the first evolutionary stage of pulsar rotation where the pulsar rotation frequency is subject to all four types of rotational irregularities discrete glitches , quasi - periodic oscillations , slow glitches , and processes that change pulsar rotation parameters , and , as a consequence , change the sign of @xmath0 .
these pulsars have the largest @xmath0 values and show strong timing noise . in their timing residuals a large cubic trend dominates that is the result of relaxation after a discrete glitch ( for @xmath118 ) or the result of a process of a change in the sign of @xmath0 ( for @xmath0 with both signs ) .
the middle - aged pulsars stay in the second evolutionary stage of pulsar rotation and their timing activity is related to three types of rotational irregularities quasi - periodic oscillations , slow glitches , and small discrete glitches .
the timing residuals of such pulsars exhibit mainly a quasi - periodic character .
the strength of the timing noise of old pulsars that stay in the third evolutionary stage weakens because rotational irregularities at this stage gradually disappear .
the timing residuals of old pulsars are mainly dominated by noise - like variations .
this presented evolutionary picture is in accordance with the results of @xcite , who discussed the features of timing residuals depending on the characteristic age of pulsars .
the age boundaries between different evolutionary stages are diffusive .
this feature of pulsar rotation evolution explains why two pulsars with similar rotation parameters exhibit different signatures in their timing residuals .
for example , two pulsars , b0943 + 10 and b1133 + 16 , have almost identical rotation parameters @xmath15 and @xmath20 , as follows from table [ param27 ] , but exhibit different structures in their timing and frequency residuals , as seen in figures [ resid27](c ) and [ resid27](d ) .
this result occurs because these pulsars have different evolutionary histories ; they evolve along different paths and therefore belong to different evolutionary stages of pulsar rotation . in figure [ zavis27
] , we see that the pulsar b0943 + 10 stays in the first evolutional stage , while the pulsar b1133 + 16 stays in the second stage of pulsar rotation evolution .
it was believed that all pulsars can experience glitches @xcite .
the evolutionary scenario for the origin of pulsar rotational irregularities is inconsistent with this statement .
pulsar glitches are not common phenomena experienced by pulsars of all ages .
figure [ zavis27 ] shows that pulsar glitches occur more frequently in younger pulsars as these pulsars belong to the first evolutionary stage of pulsar rotation .
pulsar glitches will never occur in old pulsars that belong to the third evolutionary stage of pulsar rotation .
pulsars in the final evolutional stage either already have experienced the stage of glitches in their rotation rate or have never experienced any shocks in their life and so immediately entered the final stage of pulsar rotation evolution .
as a consequence , one pulsar may regularly experience glitch events while another pulsar with similar properties has never been observed to glitch .
on the basis of a detailed analysis of the timing behavior of 27 pulsars over long data baselines spanning up to 43.5 yr , we have revealed the following important features of pulsar rotation .
we detected a new type of rotational irregularity that is the result of a rapid change of pulsar rotation parameters in such a way that the sign of @xmath0 is reversed .
this process was revealed in the timing behavior of two pulsars , b0823 + 26 and b1929 + 10 .
the detected phenomenon of changing the sign of @xmath0 can help to explain the problem why pulsars with characteristic ages older than @xmath140 yr have both positive and negative @xmath0 values .
we showed that a variable structure of timing noise over long time scales is produced by four types of rotational irregularities that occur in the form of discrete glitches , quasi - periodic oscillations , slow glitches , and a process changing pulsar rotation parameters together with a change in the sign of @xmath0 .
we found that all four types of observed rotational irregularities have evolutionary nature and form three evolutionary stages in pulsar rotation rate .
this result means that a certain type of rotational irregularity can occur only at a certain stage of pulsar rotation evolution . in the first stage of rotation evolution , including the younger pulsars with @xmath141 yr
, all four types of rotational irregularities occur discrete glitches , slow glitches , quasi - periodical oscillations , and a process of changing pulsar rotation parameters and the sign of @xmath0 . in the second stage of rotation evolution , including the middle - aged pulsars with @xmath142 yr , large glitches and the process that changes the sign of @xmath0 disappear .
rotational irregularities include only small discrete glitches , slow glitches , and quasi - periodical oscillations . in the third stage of rotation evolution , including both the middle - aged and old pulsars , rotational irregularities should gradually weaken and disappear entirely . in this stage ,
the timing residuals are mainly dominated by noise - like variations .
the evolutionary nature of rotation irregularities indicates that the deterministic values of @xmath0 and accordingly to the true values of the braking indices @xmath3 can be never measured for ordinary pulsars that stay in the first or second stages of pulsar rotation evolution .
we found that the surface magnetic field strength @xmath134 for pulsars is correlated with certain signatures of the timing residuals , as shown in figure [ zavb0 ] .
we revealed two surprising features in this correlation picture .
the first feature is that the magnetic fields are stronger for pulsars that show noise - like signatures in their timing residuals , that is , for the pulsars whose rotation has a more stable character .
the second feature is that the magnetic fields for two groups of the pulsars , those exhibiting timing residuals with cubic signatures and those exibiting timing residuals with quasi - periodic signatures , smoothly vary from one group to another .
this result indicates that these two groups of pulsars have a common evolutionary path .
the observed correlation of the magnetic field with certain signatures of the timing residuals confirms the evolutionary nature of rotational irregularities .
we found that the age boundaries between different evolutionary stages are indistinct and diffusive .
this result is because different pulsars with similar properties evolve along different paths and have different evolutionary histories .
the first corollary of this feature is that pulsar glitches are not a common event experienced by pulsars of all ages .
pulsar glitches will never occur in old pulsars that belong to the third evolutionary stage of pulsar rotation .
the second corollary of this feature is that the measurements of the deterministic values of @xmath0 are possible for some middle - aged pulsars that stay in the third evolutionary stage and show stable rotation and a low level of timing noise .
it is very likely that these pulsars did not experience any shocks over their lifetimes .
then , the measurements of the deterministic @xmath0 values for such pulsars will not require very long , in hundreds of years , data baselines . in summary
, we have studied the properties of the timing behavior of 27 pulsars whose characteristic ages range from @xmath131 to @xmath143 yr .
we propose that the evolutionary scenario of the occurrence of rotational irregularities can be generalized to all ordinary pulsars in the given age range .
very likely , the timing residuals of younger pulsars can also be well explained in the context of the evolutionary scenario of the occurrence of rotational irregularities .
we are grateful to r. d. dagkesamansky for discussions and valuable comments and the staff of the prao for their aid in carrying out the many - year pulsar observations on the bsa radiotelescope .
this work was supported by the russian foundation for basic research ( grant no .
09 - 02 - 00473 ) and the program of the russian academy of sciences physical division `` active processes in galactic and extragalactic objects '' .
the authors are grateful to the referee for helpful comments and suggestions .
alpar , m. a. , & baykal , a. 1994 , , 269 , 849 alpar , m. a. , & baykal , a. 2006 , , 372 , 489 arzoumanian , z. , nice , d. j. , taylor , j. h. , & thorsett , s. e. 1994 , , 422 , 671 backus , p. r. , taylor , j. h. , & damashek , m. 1982 , apjl , 255 , l63 baykal , a. , alpar , m. a. , boynton , p. e. , & deeter , j. e. 1999 , , 306 , 207 biryukov , a. , beskin , g. , & karpov , s. 2012 , , 420 , 103 boynton , p. e. , groth , e. j. , hutchinson , d. p. , et al .
1972 , , 175 , 217 chatterjee , s. , & cordes , j. m. 2004 , apjl , 600 , l51 cordes , j. m. , & downs , g. s. 1985 , , 59 , 343 cordes , j. m. , & helfand , d. j. 1980 , , 239 , 640 dalessandro , f. , mcculloch , p. m. , hamilton , p. a. , & deshpande , a. a. 1995 , , 277 , 1033 demianski , m. , & proszynski , m. 1979 , natur , 282 , 383 downs , g. s. , & krause - polstorff , j. 1986 , , 62 , 81 downs , g. s. , & reichley , p. e. 1983 , , 53 , 169 espinoza , c. m. , lyne , a. g. , stappers , b. w. , & kramer , m. 2011 , , 414 , 1679 gullahorn , g. e. , & rankin , j. m. 1982 , , 260 , 520 helfand , d. j. , taylor , j. h. , backus , p. r. , & cordes , j. m. 1980 , , 237 , 206 hobbs , g. , lyne , a. g. , & kramer , m. 2010 , , 402 , 1027 hobbs , g. , lyne , a. g. , kramer , m. , martin , c. e. , & jordan , c. 2004 , , 353 , 1311 janssen , g. h. , & stappers , b. w. 2006 , , 457 , 611 jones , d. i. 2012 , , 420 , 2325 konacki , m. , lewandowski , w. , wolszczan , a. , doroshenko , o. , & kramer , m. 1999 , apjl , 519 , l81 livingstone , m. a. , kaspi , v. m. , gavriil , f. p. , & manchester , r. n. 2005 , , 619 , 1046 lyne , a. , hobbs , g. , kramer , m. , stairs , i , & stappers , b. 2010 , sci , 329 , 408 lyne , a. g. 1987 , natur , 326 , 569 lyne , a. g. 1996 , in iau colloq .
160 , pulsars : problems and progress , ed .
s. johnston , m. a. walker , & m. bailes ( san francisco , ca : asp ) , 73 lyne , a. g. , pritchard , r. s. , & smith , f. g. 1988 , , 233 , 667 lyne , a. g. , shemar , s. l. , & smith , f. g. 2000 , , 315 , 534 manchester , r. n. , hobbs , g. b. , teoh , a. , & hobbs , m. 2005 , , 129 , 1993 manchester , r. n. , & taylor , j. h. 1974 , apjl , 191 , l63 manchester , r. n. , & taylor , j. h. 1977 , pulsars ( san francisco , ca : freeman ) melatos , a. , peralta , c. , & wyithe , j. s. b. 2008 , , 672 , 1103 middleditch , j. , marshall , f. e. , wang , q. d. , gotthelf , e. v. , & zhang , w. 2006 , , 652 , 1531 ruderman , m. 1991a , , 382 , 576 ruderman , m. 1991b , , 382 , 587 ruderman , m. 2004 , arxiv : astro - ph/0410607 ruderman , m. 2006 , arxiv : astro - ph/0610375 ruderman , m. , zhu , t. , & chen , k. 1998 , , 492 , 267 scott , d. m. , finger , m. h. , & wilson , c. a. 2003 , , 344 , 412 shabanova , t. v. 1990 , sva , 34 , 372 shabanova , t. v. 1995 , , 453 , 779 shabanova , t. v. 1998 , , 337 , 723 shabanova , t. v. 2005 , , 356 , 1435 shabanova , t. v. 2007 , , 308 , 591 shabanova , t. v. 2009a , arep . , 53 , 465 shabanova , t. v. 2009b , , 700 , 1009 shabanova , t. v. 2010 , , 721 , 251 shabanova , t. v. , lyne , a. g. , & urama , j. o. 2001 , , 552 , 321 shabanova , t. v. , & urama , j. o. 2000 , , 354 , 960 shemar , s. l. , & lyne , a. g. 1996 , , 282 , 677 taylor , j. h. , manchester , r. n. , & lyne , a. g. 1993 , , 88 , 529 urama , j. o. , link , b. , & weisberg , j. m. 2006 , , 370 , l76 wang , n. , manchester , r. n. , pace , r. t. , et al .
2000 , , 317 , 843 yu , m. , manchester , r. n. , hobbs , g. , et al . 2013 , , 429 , 688 yuan , j. p. , wang , n. , manchester , r. n. , & liu , z. y. 2010 , , 404 , 289 zou , w. z. , wang , n. , wang , h. x. , et al .
2004 , , 354 , 811 lcccccc j0141 + 6009 & b0138 + 59 & 45326 - 55957 & 1541 & 0.8 & 34.91 & * + j0323 + 3944 & b0320 + 39 & 44268 - 55973 & 2660 & 0.7 & 26.17 & * + j0332 + 5434 & b0329 + 54 & 43702 - 55973 & 3037 & 0.05 & 26.776 & 1 + j0358 + 5413 & b0355 + 54 & 45086 - 55973 & 2632 & 0.7 & 57.153 & 2 + j0814 + 7429 & b0809 + 74 & 44087 - 55973 & 2777 & 0.4 & 5.7513 & 1 + j0826 + 2637 & b0823 + 26 & 44807 - 55973 & 2706 & 0.2 & 19.454 & 2 + j0837 + 0610 & b0834 + 06 & 43701 - 55973 & 2211 & 0.1 & 12.8579 & 1 + j0922 + 0638 & b0919 + 06 & 45569 - 55974 & 2822 & 0.5 & 27.271 & 2 + j0946 + 0951 & b0943 + 10 & 45134 - 55957 & 1573 & 0.7 & 15.4 & 2 + j0953 + 0755 & b0950 + 08 & 43703 - 55958 & 917 & 0.2 & 2.958 & 2 + j1136 + 1551 & b1133 + 16 & 46750 - 55974 & 2191 & 0.2 & 4.8471 & 1 + j1509 + 5531 & b1508 + 55 & 49250 - 55972 & 2625 & 0.05 & 19.613 & 2 + j1543 + 0929 & b1541 + 09 & 45088 - 55972 & 2723 & 1.0 & 34.99 & 1 + j1645@xmath60317&b1642@xmath603 & 45962 - 55973 & 2706 & 0.3 & 35.737 & * + j1740 + 1311 & b1737 + 13 & 48285 - 55973 & 1510 & 1.1 & 48.673 & 2 + j1823 + 0550 & b1821 + 05 & 48284 - 55950 & 579 & 0.9 & 66.775 & 2 + j1825@xmath60935&b1822@xmath609 & 48333 - 55973 & 1713 & 1.0 & 19.383 & * + j1840 + 5640 & b1839 + 56 & 45398 - 55973 & 1743 & 0.2 & 26.764 & * + j1921 + 2153 & b1919 + 21 & 43715 - 55973 & 2187 & 0.2 & 12.4309 & 1 + j1932 + 1059 & b1929 + 10 & 51385 - 55957 & 946 & 0.4 & 3.180 & 2 + j2018 + 2839 & b2016 + 28 & 45724 - 55973 & 2327 & 0.2 & 14.19 & * + j2022 + 2854 & b2020 + 28 & 48270 - 55973 & 2061 & 0.4 & 24.66 & * + j2113 + 2754 & b2110 + 27 & 45724 - 55973 & 2447 & 0.5 & 25.113 & 2 + j2219 + 4754 & b2217 + 47 & 45073 - 55973 & 1491 & 0.06 & 43.519 & 2 + j2225 + 6535 & b2224 + 65 & 45144 - 55957 & 1281 & 0.8 & 36.226 & * + j2305 + 3100 & b2303 + 30 & 48329 - 55973 & 955 & 1.2 & 49.544 & 2 + j2317 + 2149 & b2315 + 21 & 48330 - 55971 & 1309 & 1.1 & 20.854 & * + llllcccc b0138 + 59 & 0.8176959968288(5 ) & -0.261461(3 ) & 0.00018(6 ) & 45326.7082 & 45326 - 55957 & 0.6 & 0.5 + b0320 + 39 & 0.329807506325(1 ) & -0.069141(3 ) & 0.00021(5 ) & 44268.6762 & 44268 - 55973 & 1.0 & 1.0 + b0329 + 54 & 1.399543746342(8 ) & -4.01374(3 ) & 0.0343(4 ) & 40105.2304 & 40105 - 55973 & 24 & 11 + b0355 + 54 & 6.39460122291(2 ) & -179.7892(2 ) & 1.671(4 ) & 48298.6660 & 42298 - 55973 & ... & 2.7 + b0809 + 74 & 0.773849236501(1 ) & -0.100671(4 ) & 0.00049(8 ) & 44087.3634 & 44087 - 55973 & 1.2 & 1.2 + b0823 + 26 & 1.88444719685(3 ) & -5.929(1 ) & -4.14(8 ) & 40264.4381 & 40264 - 42364 & 4.9 & 1.4 + b0823 + 26 & 1.88444611497(2 ) & -6.0976(1 ) & 0.646(2 ) & 42364.6405 & 42364 - 55973 & ... & 40.2 + b0834 + 06 & 0.7850739585879(2 ) & -4.190673(1 ) & 0.00113(3 ) & 43701.4411 & 43701 - 55973 & 0.3 & 0.2 + b0919 + 06 & 2.32226169326(9 ) & -74.0527(5 ) & 1.91(2 ) & 44210.5817 & 44210 - 55139 & ... & 57 + b0919 + 06 & 2.32219477708(9 ) & -74.037(5 ) & 18(1 ) & 55140.1604 & 54140 - 55974 & 1.9 & 2.2 + b0943 + 10 & 0.91099181885(4 ) & -2.8750(2 ) & -0.719(4 ) & 45134.5681 & 45134 - 55957 & ... & 48 + b0950 + 08 & 3.951553316192(9 ) & -3.58546(3 ) & -0.0644(6 ) & 40104.8560 & 40104 - 55958 & 20.7 & 5.1 + b1133 + 16 & 0.841813872293(3 ) & -2.64632(1 ) & 0.0109(2 ) & 40154.6631 & 40154 - 55974 & 16 & 8.6 + b1508 + 55 & 1.351932973645(13 ) & -9.13014(9 ) & -0.347(4 ) & 49250.5223 & 49250 - 55972 & 18 & 6 + b1541 + 09 & 1.33609766492(2 ) & -0.77321(6 ) & 0.009(2 ) & 45087.9517 & 45088 - 55972 & 8.9 & 8.8 + b1642@xmath603 & 2.57938868630(3 ) & -11.8466(2 ) & 0.013(3 ) & 40414.1297 & 40414 - 55973 & 23 & 23 + b1737 + 13 & 1.245252847252(4 ) & -2.25742(2 ) & 0.2121(3 ) & 44240.7561 & 44240 - 55973 & 52 & 4 + b1821 + 05 & 1.328186108116(4 ) & -0.40025(2 ) & 0.0022(6 ) & 48284.3024 & 48284 - 55950 & 0.9 & 0.9 + b1822@xmath609 & 1.30038133415(4 ) & -88.859(2 ) & 14.4(1 ) & 54116.3366 & 54116 - 55973 & 14.8 & 4.5 + b1839 + 56 & 0.6050113973913(7 ) & -0.547042(3 ) & -0.0057(2 ) & 45398.2170 & 45398 - 55973 & 2.4 & 1.0 + b1919 + 21 & 0.747774504487(1 ) & -0.753867(2 ) & -0.00050(5 ) & 43715.8588 & 43715 - 55973 & 0.9 & 0.8 + b1929 + 10 & 4.41467924848(3 ) & -22.5547(4 ) & -0.18(2 ) & 40401.2136 & 40401 - 45236 & 2.1 & 2.0 + b1929 + 10 & 4.41465783839(3 ) & -22.6674(3 ) & 4.42(1 ) & 51385.8682 & 51385 - 55957 & 20.5 & 2.3 + b2016 + 28 & 1.792264372162(6 ) & -0.47720(2 ) & 0.0269(2 ) & 40105.0491 & 40105 - 55973 & 12.0 & 4.3 + b2020 + 28 & 2.912039586312(7 ) & -16.06542(4 ) & 0.2926(8 ) & 48270.4234 & 48270 - 55973 & 9.6 & 1.8 + b2110 + 27 & 0.831358116226(1 ) & -1.813019(3 ) & 0.00429(7 ) & 45724.4286 & 45724 - 55973 & 1.3 & 0.7 + b2217 + 47 & 1.857122692273(8 ) & -9.53658(3 ) & -0.0073(3 ) & 40585.0654 & 40585 - 55854 & 7.1 & 6.1 + b2224 + 65 & 1.46512685285(5 ) & -20.7457(2 ) & 0.313(4 ) & 45144.0676 & 45144 - 55957 & 57 & 24 + b2303 + 30 & 0.634563570173(2 ) & -1.16455(2 ) & -0.0096(5 ) & 48329.3741 & 48329 - 55973 & 2.3 & 1.7 + b2315 + 21 & 0.692207715845(1 ) & -0.501662(8 ) & -0.0018(3 ) & 48330.3796 &
48330 - 55971 & 1.1 & 1.1 + lcccc 1 & 1994 & 49615(8 ) & 0.2(1 ) & -0.6(2 ) + 2 & 1 slow & 49857 & 12.8(2 ) & -7.0(2 ) + 3 & 1a slow & 50253 & 4.3(2 ) & -4.8(3 ) + 4 & 2 slow & 51060 & 28.7(6 ) & -24.2(4 ) + 5 & 3 slow & 51879 & 32.0(9 ) & -16.7(8 ) + 6 & 4 slow & 52700 & 2.5(3 ) & -2.9(3 ) + 7 & 2006 & 53745(2 ) & 6.7(4 ) & -0.1(6 ) + 8 & 2007 & 54115.0(2 ) & 121(1 ) & -0.2(4 ) + lc pre - glitch parameters & + mjd range & 5560055854 + @xmath15 ( hz ) & 1.85711010737(1 ) + @xmath20 ( @xmath144 s@xmath121 ) & -9.5599(7 ) + epoch ( mjd ) & 55858.0 + rms timing residual ( @xmath85s ) & 60 + post - glitch parameters & + mjd range & 5586655973 + @xmath15 ( hz ) & 1.85711010972(3 ) + @xmath20 ( @xmath144 s@xmath121 ) & -9.536(5 ) + epoch ( mjd ) & 55858.0 + rms timing residual ( @xmath85s ) & 60 + glitch parameters & + @xmath145 & 1.27(4 ) + @xmath146 & -2.5(5 ) + epoch ( mjd ) & 55858(1 ) + lcccccc b1822@xmath609 & 0.23 & 5.36 & -23.84 & 237 & -25.74 & 12.81 + b0919 + 06 & 0.50 & 5.70 & -24.70 & 80 & -26.15 & 12.39 + b0919 + 06 & 0.50 & 5.70 & -23.74 & 762 & -26.15 & 12.39 + b0355 + 54 & 0.56 & 5.75 & -24.78 & 33 & -25.82 & 11.92 + b2224 + 65 & 1.10 & 6.05 & -25.50 & 106 & -27.06 & 12.42 + b1508 + 55 & 2.30 & 6.37 & -25.46 & -562 & -27.73 & 12.29 + b2020 + 28 & 2.87 & 6.46 & -25.53 & 330 & -27.58 & 11.91 + b0834 + 06 & 2.95 & 6.47 & -27.96 & 4 & -28.17 & 12.47 + b2217 + 47 & 3.08 & 6.49 & -27.14 & -14 & -27.83 & 12.09 + b1929 + 10 & 3.10 & 6.49 & -25.74 & -156 & -27.46 & 11.72 + b1929 + 10 & 3.08 & 6.49 & -24.36 & 3797 & -27.46 & 11.72 + b1642@xmath603 & 3.45 & 6.54 & -26.89 & 23 & -27.79 & 11.92 + b0823 + 26 & 5.04 & 6.70 & -24.38 & -22188&-28.25 & 11.98 + b0823 + 26 & 4.90 & 6.69 & -25.19 & 3274 & -28.23 & 11.98 + b0943 + 10 & 5.02 & 6.70 & -25.14 & -7924 & -28.56 & 12.30 + b1133 + 16 & 5.04 & 6.70 & -26.96 & 131 & -28.60 & 12.33 + b0329 + 54 & 5.52 & 6.74 & -26.46 & 297 & -28.46 & 12.10 + b2110 + 27 & 7.26 & 6.86 & -27.37 & 108 & -28.93 & 12.26 + b2303 + 30 & 8.63 & 6.93 & -27.01 & -449 & -29.19 & 12.34 + b1737 + 13 & 8.74 & 6.94 & -25.67 & 5181 & -28.91 & 12.04 + b1919 + 21 & 15.71 & 7.20 & -28.30 & -65 & -29.64 & 12.13 + b0950 + 08 & 17.46 & 7.24 & -26.19 & -1979 & -29.01 & 11.39 + b1839 + 56 & 17.52 & 7.24 & -27.24 & -1152 & -29.83 & 12.20 + b2315 + 21 & 21.86 & 7.34 & -27.74 & -496 & -29.96 & 12.10 + b1541 + 09 & 27.37 & 7.44 & -27.05 & 2012 & -29.87 & 11.76 + b0138 + 59 & 49.55 & 7.69 & -28.74 & 215 & -30.60 & 11.84 + b1821 + 05 & 52.57 & 7.72 & -27.66 & 1825 & -30.44 & 11.62 + b2016 + 28 & 59.50 & 7.77 & -26.57 & 21186 & -30.42 & 11.46 + b0320 + 39 & 75.57 & 7.88 & -28.68 & 1455 & -31.36 & 12.15 + b0809 + 74 & 121.8 & 8.09 & -28.31 & 3717 & -31.41 & 11.67 +
|
we present results from timing observations of 27 pulsars made at the pushchino observatory over 33.5 yr between 1978 july and 2012 february .
we also analyze archival jet propulsion laboratory data of 10 pulsars to extend individual data span to 43.5 yr .
we detected a new phenomenon in the timing behavior of two pulsars , b0823 + 26 and b1929 + 10 , that demonstrates a rapid change of pulsar rotation parameters such that the sign of the second derivative @xmath0 is reversed .
an analysis of the @xmath0 changes showed that this process can be considered as a modulation process in @xmath0 .
we showed that the process of rapidly changing of pulsar rotation parameters represents a new type of rotational irregularity that , together with three other types of rotational irregularities ( discrete glitches , slow glitches and quasi - periodic oscillations ) , forms a large - scale structure of timing noise .
these effects are all the cause of the deviation of the timing behavior of most ordinary pulsars from a simple @xmath1 spin - down model .
we found that all four types of observed rotational irregularities have evolving nature .
irregularities in pulsar rotation rate pass through three evolutional stages that show that a certain type of rotational irregularity can occur only at a certain stage of pulsar rotation evolution
. the age boundaries between different evolutionary stages are indistinct and diffusive .
this fact is because different pulsars having similar properties evolve along different paths .
the evolutionary scenario of the occurrence of rotational irregularities explains well many of the observed properties of pulsar rotation .
|
in this appendix we obtain the scattering probability @xmath162 in eq .
( [ eq : w_kk ] ) that gives rise to anomalous transport properties . for simplicity
we consider point - like impurities whose scattering matrix elements are independent of momentum , @xmath163 .
skew scattering appears beyond the lowest born approximation as a result of a particular structure of the order parameter in @xmath0 superconductors .
we note that even weak impurities , @xmath164 , lead to the existence of bound states in @xmath0 superconductors with binding energies @xmath165 @xcite .
therefore low energy quasiparticles undergo resonant scattering .
we show below that this also results in energy - dependent skew scattering .
the scattering properties of a single impurity are described by the @xmath166-matrix . in a superconductor , for each initial and final momenta @xmath167 and @xmath47 the @xmath166-matrix acquires an additional @xmath168 structure in the nambu - gorkov space .
we denote these @xmath168 matrices by @xmath169 .
the quasiparticle scattering amplitude is given by the on - shell matrix element @xmath170 , with @xmath171 being the two component quasiparticle wave function .
using the fermi golden rule the scattering probability can be expressed as @xmath172 where @xmath61 is the impurity concentration .
the @xmath166-matrix obeys the lipmann - schwinger equation @xmath173 for point - like impurities , @xmath174 ( with @xmath175 being the pauli matrices in the nambu - gorkov space ) , the @xmath166-matrix depends only on the energy @xmath176 .
thus eq .
( [ tmatrix ] ) simplifies to @xmath177 the order parameter of @xmath0 superconductor can be expressed as @xmath178 where @xmath179 is the momentum - dependent phase such that @xmath180 and @xmath181 .
this leads to the bcs hamiltonian @xmath182 and the green function @xmath183 , @xmath184 replacing @xmath185 with @xmath186 , where @xmath32 is the volume of the sample and @xmath55 is the density of state at fermi surface , and using eq .
( [ tmatrixe ] ) and eq .
( [ green ] ) , we obtain for the @xmath166-matrix , @xmath187 the pole of the @xmath166-matrix describes the subgap bound state @xcite with the binding energy @xmath188 . for the energy in the continuum , @xmath189 , we obtain the @xmath166-matrix in the form @xmath190 where @xmath191 is given by eq .
( [ eq : phase_shift ] ) and @xmath192 the quasiparticle scattering amplitude is given by the on - shell matrix elements between two quasiparticle states with energy @xmath193 , @xmath194 in a @xmath195 superconductor the bogoliubov amplitudes carry a momentum - dependent phase @xmath196 , @xmath197 where @xmath198 and @xmath199 have the standard bcs form and are independent of the momentum direction , @xmath200 substituting eqs .
( [ eq : spinor ] ) and ( [ eq : uv ] ) into ( [ eq : t_amplitude ] ) we get @xmath201\big\}\label{tkkprimesquare}\\ & = & \frac{1}{2}f^2(\epsilon)\left(1+\frac{\xi_{\mathbf{p}}\xi_{\mathbf{p}^{\prime}}-\delta^2}{\epsilon_{\mathbf{p}}^2}\right)+ \nonumber \\ & & f^2(\epsilon)\frac{\delta^2}{2\epsilon_{\mathbf{p}}^2}\big[1-\cos\left(2\delta_\epsilon+\varphi_{\mathbf{p}}- \varphi_{\mathbf{p}^{\prime}}\right)\big ] \end{aligned}\ ] ] substituting this result into eq .
( [ scatproba ] ) , and recalling that for weak impurities , @xmath202 , the scattering rate in the normal state is @xmath60 we obtain the scattering probability in the form of eq .
( [ eq : w_kk ] ) and eq .
( [ eq : w_12 ] ) .
phys . * 75 * , 657 ( 2003 ) . , science , * 306 * , 1151 ( 2004 ) .
f. kidwingira , j. d. strand , d. j. van harlingen , and y. maeno , science , * 314 * , 1267 ( 2006 ) .
j. xia , y. maeno , p. t. beyersdorf , m. m. fejer , and a. kapitulnik , phys .
* 97 * , 167002 ( 2006 ) .
g. m. luke _
et al . _ ,
nature * 394 * , 558 ( 1998 ) .
y. maeno , s. kittaka , t. nomura , s. yonezawa , and k. ishida , j. phys .
soc . of japan ,
* 81 * , 011009 ( 2012 ) . , j.phys .
: condens .
matter * 7 * , l643 ( 1996 ) .
s. nishizaki , y. maeno , and z. mao , j. low temp . phys . * 117 * , 1581 ( 1999 ) . k. ishida , h. mukuda , y. kitaoka , z. q. mao , y. mori , and y. maeno , phys .
lett . * 84 * , 5387 ( 2000 ) .
p. g. bjornsson _
et al . _ , phys . rev .
b * 72 * , 012504 ( 2005 ) ; j. kirtley _ et al . _ ,
b * 76 * , 014526 ( 2007 ) ; c. hicks _ et al . _ , phys .
b * 81 * , 214501 ( 2010 ) .
z. q. mao , y. maeno , s. nishizaki , t. akima , and t. ishiguro , phys .
lett . * 84 * , 991 ( 2000 ) .
n. nagaosa , j. sinova , s. onoda , a. h. macdonald , and n. p. ong , rev .
phys . * 82 * , 1539 ( 2010 ) .
di xiao , ming - che chang , and qian niu , rev .
mod . phys .
* 82 * , 1959 ( 2010 ) .
a. g. aronov , yu . m. galperin , v. l. gurevich and v. i. kozub , advances in physics , 1981 , vol .
* 30 * , no . 4 , 539 - 592 .
a. g. aronov , yu . m. galperin , v. l. gurevich and v. i. kozub , in _ nonequilibrium superconductivity _ ,
d. n. langenberg and a. i. larkin ( elsevier , 1986 ) .
l. d. landau , and e. m. lifshitz , _ quantum mechanics , 3rd edition _ ,
151 , ( butterworth - heinemann , amsterdam , 2004 ) .
r. m. white and t. h. geballe , _ long range order in solids _ , ( academic , new york , 1979 ) .
r. m. lutchyn , p. nagornykh , and v. m. yakovenko , phys .
b * 77 * , 144516 ( 2008 ) .
e. tailor , c. kallin , phys .
lett . * 108 * , 157001 ( 2012 ) . c. kallin , rep .
75 * , 042501 ( 2012 ) .
a. keles , a. v. andreev , and b. z. spivak , phys .
b * 89 * , 014505 ( 2014 ) .
j. a. sauls , adv.phys .
* 43 * , 113 ( 1994 ) . j. d. strand _ et .
al _ , science * 328 * , 1368 ( 2010 ) .
j. d. strand _ et .
al _ , phys .
lett . * 103 * , 197002 ( 2009 ) .
e. r. schemm _ et .
al _ , phys .
b * 91 * , 140506 ( 2015 ) .
e. r. schemm , science , * 345 * , 190 ( 2014 ) . y. okuno , m. matsumoto , and m. sigrist , j. phys .
jpn . * 68 * , 3054 - 3061 ( 1999 ) .
|
spontaneous breaking of time - reversal symmetry in superconductors with the @xmath0 symmetry of the order parameter allows for a class of effects which are analogous to the anomalous hall effect in ferromagnets .
these effects exist below the critical temperature , @xmath1 .
we develop a kinetic theory of such effects . in particular , we consider anomalous hall thermal conductivity , the polar kerr effect , the anomalous hall effect , and the anomalous photo- and acousto - galvanic effects . _ introduction : _ one of the leading candidates for @xmath2-wave pairing in electronic systems is sr@xmath3ruo@xmath4 .
there are numerous pieces of experimental evidence that the superconducting state of sr@xmath3ruo@xmath4 has odd parity , breaks time reversal symmetry and is spin triplet @xcite .
an order parameter consistent with these experiments is given by the chiral @xmath2-wave state @xcite which is an analog of @xmath5he-@xmath6 .
it has the form @xmath7 where @xmath8 is the fourier transform of @xmath9 .
however , the observation of power laws in specific heat @xcite and nmr @xcite , the absence of electric currents along edges @xcite , and the absence of a split transition in the presence of an in - plane magnetic field @xcite are inconsistent with the theoretically expected properties of a simple chiral superconductor .
consideration of additional experimental manifestations of spontaneous breaking of time - reversal symmetry in @xmath10 superconductors may clarify the nature of superconducting state in sr@xmath3ruo@xmath4 .
due to spontaneous breaking of time - reversal symmetry , @xmath0 superconductors must exhibit anomalous transport phenomena similar to those which exist in metallic ferromagnets ( see refs .
@xcite and @xcite for a review ) . in this article
we develop a theory of several such effects in @xmath0 superconductors : the anomalous hall effect , polar kerr effect for microwave radiation , anomalous hall thermal conductivity , and anomalous photo- and acousto - galvanic effects .
it should be noted that @xmath2-wave superconductivity exists only in the clean regime , @xmath11 , where electron transport may be described semiclassically .
generally , in the semiclassical regime there are three contributions to anomalous transport phenomena : skew scattering , side jumps , and the intrinsic contribution .
the side jump contribution arises from the shift of the center of mass of electron wave packets during the scattering events , while the intrinsic contribution is related to the anomalous velocity due to berry curvature .
the magnitude of these contributions is independent of the mean free path .
in contrast , the magnitude of the skew scattering contribution is proportional to the quasiparticle mean free path @xmath12 . as a result , the skew scattering contribution exceeds the intrinsic and side jump contributions by a factor @xmath13 . here
@xmath14 is the fermi momentum .
therefore in this article we will take into consideration only the skew scattering contribution .
we focus on anomalous transport phenomena in the vicinity of the critical temperature , where quasiparticles play a major role .
_ kinetic scheme : _ transport theory in conventional time reversal invariant superconductors was developed long ago ( see for example reviews refs .
@xcite and @xcite ) .
below we generalize this approach to superconductors without time reversal symmetry , which exhibit anomalous transport phenomena . in the clean regime , @xmath15 , and at sufficiently low frequencies , @xmath16 , where @xmath17 is the modulus of the order parameter , the quasiparticle dynamics can be described by the boltzmann kinetic equation for quasiparticles @xmath18 where [ epsilon ] @xmath19 in eq .
( [ epsilon ] ) @xmath20 is the electron mass , while @xmath21 and @xmath22 , are given by @xmath23 where @xmath24 is the order parameter phase , and @xmath25 and @xmath26 are the scalar and vector potentials . from eq .
( [ eq : p_s_phi ] ) one obtains the equation for the acceleration of the condensate , @xmath27 equations ( [ kineq])-([eq : p_s_phi ] ) should be supplemented by the expression for the current density , @xmath28 and by the charge neutrality condition , @xmath29 that relates the gauge invariant scalar potential and the odd in @xmath30 part of the quasiparticle distribution function , the self - consistency equation for the order parameter .
here @xmath31 ( @xmath32 is the volume of the sample ) and @xmath33 .
we work in linear response to external perturbations , and neglect corrections to equilibrium value of @xmath17 . the collision integral @xmath34 in eq .
( [ kineq ] ) describes both elastic and inelastic scattering .
we will assume that @xmath35 , where @xmath36 and @xmath37 are inelastic and elastic mean free time respectively .
therefore the main contribution to the aforementioned anomalous effects comes from elastic scattering , which is described by the collision integral @xmath38 skew scattering of quasiparticles corresponds to the part of scattering probability in eq .
( [ eq : i ] ) that is associated with breaking of time reversal symmetry , @xmath39 .
thus , all the aforementioned effects are proportional to @xmath40 .
skew scattering arises beyond the lowest born approximation for the scattering amplitude .
below we consider point - like impurities . in the normal state
such impurities scatter electrons only in the @xmath41-wave channel and do not cause skew scattering .
therefore in the superconducting state skew scattering of quasiparticles is entirely due to the breaking of time reversal symmetry by the @xmath0 order parameter .
the elastic scattering probability for quasiparticles with energy @xmath42 can be characterized by @xmath43 , @xmath44 and the asimuthal angles @xmath45 , @xmath46 , which define the direction of @xmath47 and @xmath48 in the @xmath49-plane . for simplicity ,
we assume cylindrical fermi surface and obtain for the scattering probability ( see appedix for details ) .
@xmath50.\ ] ] here @xmath51 is the energy - dependent scattering phase shift .
it is related to the @xmath41-wave scattering phase shift @xmath52 in the normal state by @xmath53 we assume weak impurities , for which @xmath54 is small . here
@xmath55 is the density of states on the fermi level and @xmath56 is the impurity pseudo - potential @xcite . in this case @xmath57 and @xmath58 are given by [ eq : w_12 ] @xmath59 here @xmath60 , with @xmath61 being the impurity density , is the elastic scattering rate in the normal state .
the coefficient @xmath62 $ ] represents the enhancement factor of the quasiparticle scattering cross - section over the normal state value . the first term in eq .
( [ eq : w_kk ] ) , @xmath57 given by eq .
( [ eq : w_0 ] ) has the same structure as in @xmath41-wave superconductors .
it describes scattering only within the same ( particle - like , @xmath63 , or hole - like , @xmath64 ) branch and does not lead to branch imbalance relaxation . the second term , @xmath58 in eq .
( [ eq : w_kk ] ) is absent in @xmath41-wave superconductors .
it leads leads to both skew scattering and scattering between branches of quasiparticle spectrum with different signs of @xmath30 .
the skew scattering cross - section , described by the @xmath65 term in eq .
( [ eq : w_kk ] ) , is energy - dependent .
it follows from eqs .
( [ eq : w_kk ] ) , ( [ eq : phase_shift ] ) , and ( [ eq : w_1 ] ) that it changes sign when impurity potential @xmath66 changes from repulsive to attractive .
below we consider linear response to several external perturbations and look for the quasiparticle distribution function in the form @xmath67 , where @xmath68 is a locally equilibrium fermi distribution , and @xmath69 describes the deviation from equilibrium . noting that the collision integral ( [ eq : i ] ) is nullified by an arbitrary function @xmath70 we write the linearized boltzmann equation in the form @xmath71 where the specific form of the source @xmath72 depends on the type of perturbation .
_ anomalous hall thermal conductivity : _ we first consider the hall component of the thermal conductivity @xmath73 which describes the heat flux perpendicular to the direction ( @xmath74-axis ) of the temperature gradient . in this case
the source term in eq .
( [ eq : linear_response_general ] ) has the form , @xmath75 the expression for the heat flux is @xmath76 note that @xmath77 is the group velocity of the quasiparticles while @xmath78 is the bare velocity as in a normal metal , @xmath79 .
the solution of eqs .
( [ eq : linear_response_general ] ) , ( [ eq : s_thermal ] ) has the form @xmath80.\ ] ] the hall component of the thermal conductivity tensor , @xmath73 , is determined by @xmath81 in the above expression , which is given by @xmath82 $ ] , with [ eq : ab_def ] @xmath83 for weak impurities , @xmath84 , we obtain , close to @xmath85 @xmath86 where @xmath87 ( with @xmath88 being the diffusion constant ) is the normal state thermal conductivity . _
polar kerr effect : _ next we consider a linearly polarized electromagnetic wave at normal incidence to the @xmath49 surface of @xmath0 superconductor . the reflected wave is elliptically polarized with the major axis rotated with respect to the incident one by the polar kerr angle @xcite @xmath89 where @xmath90 and @xmath91 are , respectively , the real and imaginary part of the refraction index and @xmath92 where @xmath93 is the complex ac conductivity . in this case
the electric field is uniform in the direction parallel to the surface of the sample , @xmath94 , and the value of @xmath95 is determined by eq .
( [ eq : condensate_acceleration ] ) .
the diagonal part of the conductivity is given by @xcite @xmath96 where @xmath97 is the temperature dependent superfluid density and @xmath98 is the drude conductivity .
in contrast to the thermal conductivity consideration , in the present case @xmath99/t+1)$ ] gives a nonvanishing contribution to the current response ( via @xmath97 ) because the superfluid momentum depends on the electric field .
the kerr angle @xmath100 is determined by the value of the hall component of conductivity @xmath93 . to find @xmath93 we seek the solution of eq .
( [ kineq ] ) in the form @xmath101 .
the source in eq .
( [ eq : linear_response_general ] ) becomes @xmath102 where the external electric field @xmath103 is along @xmath74-direction .
the nonequilibrium distribution @xmath69 has the form @xmath104.\ ] ] the hall conductivity depends only on the function @xmath105 , which is given by @xmath106 ^ 2+a^2(\epsilon)\}$ ] , with @xmath107 and @xmath108 being defined in eq .
( [ eq : ab_def ] ) . substituting eq .
( [ ansatzn1 ] ) into eq .
( [ current ] ) we obtain the hall conductivity in the weak impurity limit , @xmath109 , in the form @xmath110 where @xmath111 . at temperature close to @xmath85 and at low frequencies ,
@xmath112 , this expression yields @xmath113 this result was derived assuming @xmath114 symmetry of the order parameter . in the @xmath115 state
the hall conductivity @xmath93 has opposite sign .
it also changes sign if the impurity potential @xmath56 changes from repulsive , @xmath116 , to attractive , @xmath117 , in agreement with ref .
@xcite .
note that our result for the low frequency hall conductivity , eq .
( [ eq : sigma_xy_dc ] ) , is proportional to the elastic mean free time @xmath37 and to the density of quasiparticles .
there is another contribution to @xmath93 associated with the existence of the transverse component of the superfluid velocity @xmath118 , which is proportional to the condensate acceleration in the @xmath74-direction .
it may not be obtained within the present formalism that is based on the boltzmann kinetic equation for the quasiparticles . at @xmath119
this contribution is smaller than the quasiparticle contribution , eq .
( [ eq : sigma_xy_dc ] ) .
however at @xmath120 when the quasiparticle contribution becomes exponentially small in eq .
( [ eq : sigmaxyintegralanytemperature ] ) it becomes the dominant contribution . the requirement for this contribution to exist is violation of galilean invariance in the system . thus it should exist in any crystalline superconductors with @xmath0 symmetry @xcite .
it can also be caused by electron - impurity scattering . in this case
this contribution is inversely proportional to the electron mean free time @xcite .
_ hall effect for normal current injection : _ let us now consider a normal metal/@xmath0-superconductor junction , through which a steady current is flowing . at @xmath121 this situation was considered in ref . @xcite . in this regime conversion of normal current to supercurrent
is mediated by multiple andreev reflections .
here we work near the critical temperature and consider a setup , in which the normal current is injected into the superconductor in the @xmath74- direction , as shown in the inset in fig .
[ fig : f_plot ] . in this case
the conversion of quasiparticle current to the supercurrent occurs in the superconductor . just as in the case of @xmath41-wave superconductor , near @xmath85 , the electric field penetrates into superconductor to a large distance @xmath122 , which is determined by the relaxation of imbalance between the populations of quasiparticles in electron - like , @xmath63 , and hole - like , @xmath64 branches of the spectrum ( see for example ref .
@xcite and references therein ) .
the new feature of normal current injection that appears in @xmath0 superconductors is that skew scattering of quasiparticles generates nonequilibrium current that is perpendicular to the electric field .
another aspect is that , in contrast to s - wave superconductors , impurity scattering leads to branch imbalance relaxation even if the magnitude of the order parameter @xmath17 is isotropic in the fermi surface .
below we assume that the inelastic scattering rate is smaller than @xmath123 and thus impurity scattering gives the dominant contribution to branch imbalance relaxation . in linear response
we write the quasiparticle distribution function in the superconductor in the form @xmath124 .
this yields the source term in eq .
( [ eq : linear_response_general ] ) @xmath125 at length scales in excess of the mean free path we may employ the diffusive approximation . with the aid of eq . ( [ current ] )
the hall current @xmath126 can be expressed in the form @xmath127 where @xmath128 is the antisymmetric in @xmath30 part of the distribution function averaged over the momentum directions .
the latter satisfies the diffusion equation with relaxation @xmath129 with energy dependent relaxation rate @xmath130 .
the solution of eq .
( [ eq : diffusionnabar ] ) is @xmath131,\ ] ] where @xmath132 is the energy - dependent branch imbalance relaxation length .
we work in the vicinity of the critical temperature @xmath85 , where for typical thermal quasiparticles @xmath133 , ( @xmath134 ) , the relaxation lengths are long , @xmath135 .
these quasiparticles diffuse into the bulk of the superconductor and contribute to the gauge - invariant potential @xmath22 given by eq .
( [ phi ] ) .
the boundary value of the nonequilibrium quasiparticle population , @xmath136 in eq .
( [ eq : nabar ] ) is obtained by matching the solution of eq .
( [ eq : diffusionnabar ] ) with the solution of diffusion equation with energy relaxation for the electrons in the normal metal .
the result depends on both the inelastic mean free path in the normal metal , @xmath137 , and the branch imbalance relaxation length @xmath138 in the superconductor . for @xmath139 the boundary condition is @xmath140 where @xmath141 is the electric field in the normal metal generating the steady current .
here we used the fact that in the stationary case @xmath142 , which follows from eq .
( [ eq : condensate_acceleration ] ) . using this relation and substituting eqs .
( [ barnaxix ] ) , ( [ eq : nabar ] ) into eqs .
( [ phi ] ) and ( [ eq : current_grad_n ] ) we obtain the spatial distributions of the electric field @xmath143 and the hall current @xmath144 in the superconductor , @xmath145 where @xmath146 and the functions @xmath147 , are defined as @xmath148 and are plotted in fig .
[ fig : f_plot ] .
the spatial distributions of the hall current @xmath144 and the electric field @xmath149 are drastically different , and can not be related by a local hall conductivity @xmath93 . at relatively short distances , @xmath150 , we see from eq .
( [ eq : j_y_x ] ) that @xmath151 , so that the hall current is @xmath152 ( solid line ) and @xmath153 ( dashed line ) in eq .
( [ eq : f_def ] ) .
the inset shows a schematic setup of the normal current injection experiment .
electric current is injected into the superconductor @xmath154 from the normal metal @xmath155 along along the @xmath74-axis .
skew scattering of quasiparticles generates an anomalous hall current in the @xmath156 direction.,title="fig:",width=302 ] + _ anomalous photo- and acousto - galvanic effects : _ when an electromagnetic or an acoustic wave propagates through a conductor it generates an anisotropic in momentum @xmath157 distribution function .
the induced current density is proportional to the rate of the the momentum transfer from the wave to the electron system @xcite , @xmath158 here @xmath159 is the rate of momentum density transfer due to the wave adsorption , and @xmath74 is the direction of the wave propagation . in @xmath0 superconductors an anomalous current in the @xmath156 direction is generated .
considerations similar to those leading to eq .
( [ eq : sigma_xy_dc ] ) near @xmath160 yield @xmath161 finally , we note that all anomalous transport phenomena discussed above are driven by the underlying symmetry of the superconducting state
. therefore they should exist in any superconductor whose order parameter breaks time reversal symmetry , see for example @xcite .
although our consideration focused on @xmath0 materials we believe our approach is applicable to other superconductors with broken time - reversal invariance .
s. l. and a. a. were supported by the u. s. department of energy office of science , basic energy sciences under award number de - fg02 - 07er46452 .
b.s .
thanks the international institute of physics ( natal , brazil ) for hospitality .
|
cycling can play an important role in creating sustainable and equitable transport systems .
cycling already provides reliable , healthy , affordable , and convenient mobility to millions of people each day ( komanoff , 2004 ) and is one of the fastest growing modes of transport in cities such as london , new york and barcelona ( fishman , 2016 ) .
there is mounting evidence about the external costs of car - dominated transport systems ( e.g. han and hayashi , 2008 ; shergold et al . , 2012 ) , and the benefits of cycling ( de nazelle et al .
, 2011 ; oja et al . , 2011 ; tainio et al . ,
2016 ) . in this context
there is growing interest , and in some cases substantial investment , in cycling infrastructure , including in countries with historically low rates of cycling .
providing high - quality infrastructure can play a key role in promoting cycling uptake ( parkin , 2012 ) .
off - road cycle paths , for example , have been found to be associated with an uptake of cycling for commuting ( heinen et al . , 2015 ) .
overall there is growing evidence linking cycling infrastructure to higher rates of cycling ( buehler and dill , 2016 ) .
but where should this infrastructure be built ?
this paper seeks to demonstrate the potential of online , evidence - based tools to help answer this question , with reference to the propensity to cycle tool ( pct ) .
the pct is an online planning support system funded by the uk s department for transport to map cycling potential ( department for transport , 2015 ) .
the pct was developed alongside two branches of academic research : a ) methodological developments for estimating cycling potential and b ) planning support systems ( pss ) .
the subsequent overview of this policy and academic landscape places the pct in its wider context and explains its key features .
a number of factors influence the attractiveness of cycling for everyday trips ( parkin , 2015 ; pucher et al . , 2010 )
. however , the intervention that has received the most attention has been the construction of new cycle paths . in the uk context , devolved transport budgets mean that local authorities have some control over the design and implementation of cycling networks .
planning new cycle paths requires many decisions to be made , including in relation to the width ( pikora et al . , 2002 ;
wegman , 1979 ) , quality ( heath et al . , 2006 ) ,
directness ( crow , 2007 ) and geographic location of the paths . yet
while much guidance has been produced regarding physical design ( e.g. transport for london , 2015 ; welsh government , 2014 ) , little work has explicitly tackled the question of where high quality infrastructure should be built ( aultman - hall et al . , 1997 ; larsen et al . , 2013 ; minikel , 2012 ) . within this policy context
, the pct focuses explicitly on the question of _ where _ to build rather than _ what _ to build , although it does provide evidence on potential capacity requirements across the route network .
there is an emerging literature exploring cycling potential .
this links to the question of ` where to build ' because areas and routes with the highest potential are likely to be cost - effective places for investment from health and emissions perspectives . with the notable exceptions of larsen et al .
( 2013 ) and zhang et al .
( 2014 ) , this body of research has not provided systematic or quantitative evidence for transport planners .
the methods broadly fit into three categories depending on the level of the input data used : * area - based measures are based primarily on data at the level of administrative zones .
outputs from these measures can assist with the location of site - specific transport infrastructure such as cycle parking . *
individual - based measures are based on survey data , typically a household travel survey .
these are not always geographically specific and tend to be used to identify and categorise demographic groups in relation to cycling , such as near - market or as warranting tailored interventions , such as targeted cycle training schemes .
* route - based measures use origin - destination data which can be used to create ` desire lines ' and ( using route allocation ) estimates of existing and potential demand at each point on the road network .
this work is reviewed in relation to the pct below and summarised in table 1 .
parkin et al .
( 2008 ) presented an area - based measure of cycling potential using regression model to estimate the proportion of commuter trips cycled across wards in england and wales .
factors associated with lower levels of cycling included road defects , high rainfall , hills and a higher proportion of ethnic minority and low - income inhabitants .
parkin et al .
concluded that policy makers must engage with a mixture of physical and social barriers to promote cycling effectively , with the implication that some areas have lower barriers to cycling and hence higher propensity to cycle than others .
zhang et al . ( 2014 ) created an individual - based model of cycling potential to prioritise where to build cycle paths to `` achieve maximum impacts early on '' .
the outputs of this model were aggregated to the level of 67 statistical zones in the study area of belo horizonte , brazil , and used to generate a ` usage intensity index ' for potential cycle paths .
this , combined with survey data on cyclists stated preferences on whether people would cycle were infrastructure provided along particular routes and origin - destination data on travel to work , was used to rank key routes in the city in terms of their cycling potential . while the methods presented by parkin et al .
( 2008 ) and zhang et al .
( 2014 ) were developed in an academic context , albeit closely related to policy needs and interests , the analysis of cycling potential ( acp ) tool was developed by practitioners ( transport for london , 2010 ) .
the acp combined area and individual - level data to produce a heat map estimating cycling potential across london , uk , for all trip purposes .
the underlying model examined which types of trips are most likely to be cycled , based on the characteristics of observed cycle trips ( e.g. time of day , characteristics of the traveller , distance ) .
the results of the acp have informed local cycling schemes , such as where to build new cycle hire stations .
the acp does not use origin - destination data directly or route allocation . again working within academia but
also closely focussed on local planning and policy issues , larsen et al .
( 2013 ) created an area - based ` prioritization index ' , for montreal , canada .
this was based on four variables : the area s current level of cycling , its cycling potential ( estimated based on the shortest path between the origin and destination of short car trips from a travel survey ) , the number of injuries to cyclists , and locations prioritised by current cyclists for improvement ( larsen et al .
these four were aggregated to the level of evenly spread cells covering the study area .
the resulting heat map was used to recommend the construction or upgrade of cycle paths on specific roads .
a more localised approach is the permeability assessment tool ( pat ) , which was developed by a transport consultancy payne ( 2014 ) .
the pat is based on the concept of ` filtered permeability ' , which means providing a more direct route to people cycling than driving ( melia , 2015 ) .
the pat works by combining geographical data , including the location of popular destinations and existing transport infrastructure , with on - site audit data of areas that have been short - listed . unlike the prioritisation index of larsen et al .
( 2013 ) , which is primarily aimed at informing a city - wide strategic cycling network , the results of the pat are designed to guide smaller , site specific interventions such as ` contraflow ' paths and cyclist priority traffic signals .
p3cm|p2cmp2cmxxp1.8 cm tool / method & scale & accessibility & levels of input data & levels of output & software licence + propensity to cycle tool & national : england & online map - based tool & area , od , route , individual & area , od , route , route network & r : open source ( agpl ) + permeability assessment tool ( payne 2014 ) & local : dublin , ireland & gis - based & area , od , route & od , route & arcgis : proprietary + usage intensity index ( zhang et al . 2014 ) & local : belo horizonte , brazil & gis - based & area , od , route , individual & area , individual , route & arcgis : proprietary + prioritization index ( larsen et al .
2013 ) & local : montreal , canada & gis - based & area , route , point & area & arcgis : proprietary + cycling potential tool ( tfl 2010 ) & local : london , uk & static results & area , individual & area , population segment & unknown + bicycle share model ( parkin et al .
2008 ) & national : england , wales & static results & area , route & area & unknown + the methods and tools for estimating cycling potential outlined in table 1 were generally created with only a single study region in mind .
the benefit of this is that they can respond context - specific to practitioner and policy needs .
however , the pct aims to provide a _ generalisable _ and _ scalable _ tool , in the tradition of planning support systems ( pss ) .
pss were developed to encourage evidence - based policy in land - use planning ( klosterman , 1999 ) .
the application of pss to transport planning has been more recent , with a goal of `` systematically [ introducing ] relevant ( spatial ) information to a specific process of related planning actions '' ( brmmelstroet and bertolini , 2008 ) .
the pct is systematic in its use of national data for all parts of the study region ( in this case england ) and relates to a specific planning process the creation of new and enhancement of existing cycle infrastructure .
pss typically work by presenting evidence about the characteristics and needs of the study region in an interactive map .
a central objective is to visualise alternative scenarios of cycling uptake and explore their potential impacts .
the results of traditional scenario - based models are usually not locally specific ( lovelace et al . , 2011 ;
mccollum and yang , 2009 ; woodcock et al . , 2009 ) .
online pss can overcome this issue by using interactive maps ( pettit et al .
the emergence of libraries for web mapping ( haklay et al . , 2008 ) has facilitated online pss , offering the potential for public access to the planning process .
transparency is further enhanced by making pss open source , in - line with a growing trend in transport modelling ( borning et al . , 2008 ;
novosel et al . , 2015 ; tamminga et al . ,
2012 ) . in these ways
, pss can make evidence for transport planning more widely available , and tackle the issue that transport models are often seen as ` black boxes ' , closed to public scrutiny ( golub et al . , 2013 ) .
in addition to the international policy and academic context , the pct was influenced by the national context .
it was commissioned by the uk s department for transport to identify `` parts of [ england ] with the greatest propensity to cycle '' ( department for transport , 2015 ) .
thus the aim was not to produce a full transport demand or land use model , but to provide an evidence base to prioritise where to create and improve cycling infrastructure based on scenarios of change .
local and national cycling targets are often based on a target mode share by a given date .. http://cycling.tfgm.com/documents/ccag2-executive-summary.pdf[manchester ]
( 10% by 2025 ) , https://www.derbyshire.gov.uk/images/2015-07-07%20item%207l%20cycle%20plan_tcm44-267216.pdf[derbyshire ] ( to double the number of people cycling by 2025 ) and http://content.tfl.gov.uk/gla-mayors-cycle-vision-2013.pdf[london ] ( to ` double cycling ' by 2025 ) provide further examples of local ambitious time - bound cycling targets .
] however , there is little evidence about what this might mean for cycling volumes along specific routes .
the pct tackles this issue by estimating rate of cycling locally under different scenarios and presenting the results on an interactive map .
its key features include : * estimation of cycling potential at area , ` desire line ' and route network levels . * route - allocation of od ( origin - destination ) pairs by a routing algorithm specifically developed for cycling .
this was done by cyclestreets.net , a routing service developed by cyclists , for cyclists .
* visualisation of outputs at multiple geographic levels .
the interactive map enables users to examine cycling potential at a very local level ( e.g. just a few streets ) or at a more regional level ( e.g. across a large metropolitan area ) . *
public accessibility of results and code .
the tool is freely available online and developers are encouraged to modify the pct ( e.g. to create alternative scenarios ) by provision of the source code underlying the pct under the open source agp license . *
the presentation of estimated health economic and carbon impacts under future scenarios , providing an evidence base that could be used in business cases for investment . as with any tool , the pct s utility depends on people knowing how to use it . for that reason training materials and
a user manual are being developed to show how the tool can be used ( see the ` manual ' tab in figure 3 and http://pct.bike/manual.html[pct.bike/manual.html ] ) .
this section describes the data and methods that generate the input data for the pct .
this is summarised in figure 1 and described in detail in the appendix .
central to the pct approach is origin - destination ( od ) data recording the travel flow between administrative zones .
combined with geographical data on the coordinates of the population - weighted centroid of each zones , these can be represented as straight ` desire lines ' or as routes allocated to the transport network .
the central input dataset was a table of origin - destination ( od ) pairs from the 2011 census .
this was loaded from open access file ` wu03ew_v2.csv ` , provided by the http://wicid.ukdataservice.ac.uk/[uk data service ] .
this captures the number of commuters travelling between middle super output area zones ( msoas , average commuter population : 3300 ) , by mode of travel ( see table 2 ) .
this dataset was derived from responses to the following questions in the http://www.ons.gov.uk/census/[english 2011 census ] : `` in your main job , what is the address of your workplace ? ''
( question 40 ) and `` how do you usually travel to work ? ( tick one box only , for the longest part , by distance , of your usual journey to work ) '' ( question 41 ) .
this dataset was enhanced by merging in information on the gender composition of cyclists in each od pair ( dataset 1 in figure 1 ) ; data at the area level on the background mortality rate ( dataset 2 ) ; and data at the od pair - level on route distance ( km ) and hilliness ( average gradient , as a percentage ) ( dataset 3 ) .
od data was assigned to the transport network using the r package stplanr ( lovelace et al . , 2016 ) .
see the appendix for further details .
.sample of the od ( origin - destination ) input dataset , representing the number of people who commute from locations within and between administrative zones ( msoas ) .
note ` car ' refers to people who drive as their main mode of travel per od pair , rather than people who travel to work as a passenger in a car .
[ cols= " < , < , < , < , < , < , < " , ] the starting point for generating our scenario - based ` cycling futures ' was to model baseline data on cycle commuting in england .
we did this using od data from the 2011 census , and modelling cycling commuting as a function of route distance and route hilliness .
we did so using logistic regression applied at the individual level , including squared and square - root terms to capture ` distance decay ' the non - linear impact of distance on the likelihood of cycling ( iacono et al .
, 2008 ) and including terms to capture the interaction between distance and hilliness .
model fit is illustrated in figure 2 ; see the appendix for details and for the underlying equations .
we also developed equations to estimate commuting mode share among groups not represented in the between - zone ( ` interzonal ' ) od data , e.g. those commuting within a specific msoa ( this is within - zone or ` intrazonal ' travel ) , or those with no fixed workplace .
this model of baseline propensity to cycle formed the basis of three of the four scenarios ( government target , go dutch and ebikes ) , as described in more detail in the next section .
four scenarios were developed to explore cycling futures in england . these can be framed in terms of the removal of different infrastructural , cultural and technological barriers that currently prevent cycling being the natural mode of choice for trips of short to medium distances .
they are not predictions of the future .
they are snapshots indicating how the spatial distribution of cycling may shift as cycling grows based on current travel patterns . at a national level , the first
two could be seen as shorter - term and the second two more ambitious .
the choice of scenarios was informed by a government target to double the number of cycle trips and evidence from england overseas about which trips _ could _ be made by cycling .
summaries of the four scenarios are as follows ( see the appendix for full details ) : * government target .
this scenario represents a doubling of the level of cycling in england ( department for transport , 2014 ) .
although substantial in relative terms , the rate of cycling under this scenario ( rising from 3% to 6% of commuters ) remains low compared with countries such as the netherlands and denmark .
this scenario was generated by adding together a ) the observed number of cyclists in each od pair in the 2011 census , and b ) the modelled number of cyclists , as estimated using the baseline propensity to cycle equations described in the previous section .
the result is that cycling overall doubles at the national level , but at the local level this growth is not uniform , in absolute or relative terms .
areas with many short , flat trips and a below - average current rate of cycling are projected to more than double .
conversely , areas with above - average levels of cycling and many long - distance hilly commuter routes will experience less than a doubling . * gender equality .
this scenario illustrates the increase in cycling that would result if women were as likely as men to cycle a given trip .
specifically , the scenario sets the proportion of female cycle commuters to be equal to the current proportion of males in each od pair .
the scenario is based on the observation that in places where cycling accounts for a high proportion of personal travel , women cycle at least as much as men ( aldred et al .
, 2016 ; pucher et al . ,
this scenario has the greatest relative impact in areas where the rate of cycling is highly gender - unequal .
* go dutch . while the government target and gender equality scenarios model relatively modest increases in cycle commuting ,
go dutch represents what would happen if english people were as likely as dutch people to cycle a trip of a given distance and level of hilliness .
this scenario thereby captures the proportion of commuters that would be expected to cycle if all areas of england had the same infrastructure and cycling culture as the netherlands ( but retained their hilliness and commute distance patterns ) .
this scenario was generated by taking baseline propensity to cycle ( see previous section ) , and applying dutch scaling factors , generated through analysis of the english and dutch national travel surveys , in which we estimated how much more likely it was that a given trip would be cycled in the netherlands than in england .
we parameterised the dutch scaling factors as a main effect plus an interaction with trip distance , to take into account the fact that the `` dutch multiplier '' is greater for shorter trips compared to longer trips .
the scenario level of cycling under go dutch is not affected by the current level of cycling .
this scenario models the additional increase in cycling that would be achieved through the widespread uptake of electric cycles ( ` ebikes ' ) .
this scenario was generated by taking baseline propensity to cycle , applying the dutch scaling factors described above , and then additionally applying ebike scaling factors .
the ebikes scenario is thus currently implemented as an extension of go dutch but could be implemented as an extension of other scenarios .
the ebike scaling factors were generated through analysis of the english , dutch and swiss national travel surveys , in which we estimated how much more likely it was that a given commute trip would be cycled by ebike owners versus cyclists in general .
we parameterised the ebike scaling factors as interactions with trip distance and with hilliness , to take account of the fact that electric cycles enable longer journeys and reduce the barrier of hills .
additional scenarios could be developed ( see discussion ) .
if deployed in other settings , the pct will likely benefit from scenarios that relate to both the current policy context and long - term aspirations . because the cost effectiveness of cycling investments are influenced by wider social impacts ,
estimated health economic and emissions impacts are presented in the pct .
an approach based on the world health organization s health economic assessment tool ( http://www.heatwalkingcycling.org/[heat ] ) was used to estimate the number of premature deaths avoided due to increased physical activity ( kahlmeier et al . , 2014 ) .
to allow for the fact that cycling would in some cases replace walking trips , heat estimates of the increase in premature deaths due to the reduction in walking were also included .
the change in walking was estimated based on the assumption that , within a given od pair , all modes were equally likely to be replaced by cycling .
thus all the non - cycling modes shown in table 2 experienced the same relative decrease .
trip duration was estimated as a function of the ` fast ' route distance and average speed . for walking and cycling
we applied the standard heat approach .
ebikes are not specifically covered in heat cycling but enable faster travel and require less energy from the rider than traditional bikes .
thus we estimated new speeds and intensity values for this mode , giving a smaller benefit for every minute spent using ebikes than conventional cycles . for more details
see the appendix .
the risk of death varies by gender and increases rapidly with age .
this was accounted for using age and sex - specific mortality rates for each local authority in england .
for the baseline and government target scenario the age distribution of cyclists recorded in the 2011 census was used .
new cyclists under go dutch and ebikes were assumed to have the age - gender profile of commuter cyclists in the netherlands .
the inclusion of age specific parameters and mode shift from walking shows how the heat approach can generate nuanced health impact estimates using publicly available data .
the net change in the number of deaths avoided for each od pair was estimated as the number of deaths avoided due to cycle commuting minus the number of additional deaths due to reduced walking .
note that this approach means that for some od pairs where walking made up a high proportion of trips , additional deaths were incurred .
the monetary value of the mortality impact was calculated by drawing on the standard ` value of a statistical life ' used by the department for transport .
we also estimated the reduction in transport carbon emissions resulting from decreased car driving in each scenario .
this again relied on the assumption that all modes were equally likely to be replaced by cycling .
the average co^2 ^ -equivalent emission per kilometre of car driving was taken as 0.186 kg , the 2015 value of an ` average ' car ( defra , 2015 ) .
the data analysis and preparation stages described in the previous sections were conducted using the national od dataset for england as a whole . by contrast , the stages described in this section were conducted using a region - by - region approach .
transport decisions tend to be made at local and regional levels ( gaffron , 2003 ) , hence the decision to display results on a per region basis .
figure 3 shows the output : ` desire lines ' lines with attributes for each od pair aggregated in both directions ( chan and suja , 2003 ; tobler , 1987 ) , and visualised as centroid to centroid ` flows ' ( rae , 2009 ; wood et al . , 2010 ) .
desire lines allocated to the route network are illustrated in figure 4 .
this shows two route options : the ` fast ' route , which represents an estimate of the route taken by cyclists to minimise travel time and the ` quiet ' route that preferentially selects smaller , quieter roads and off road paths .
routes generated by cyclestreets.net do not necessarily represent the paths that cyclists currently take ; route choice models based on gps data have been developed for this purpose ( broach et al .
, 2012 ; ehrgott et al . ,
2012 ) . of the available routes ( see https://www.cyclestreets.net/journey/help/howtouse/[cyclestreets.net/journey/help ] for more information ) , the ` fastest ' option was used .
this decision was informed by recommendations from crow ( 2007 ) , building on evidence of cyclists preferance direct routes .
the spatial distribution of cycling potential can be explored interactively by selecting the ` top n ' routes with the highest estimated cycling demand ( see the slider entitled `` n. lines ( most cycled ) '' in figures 3 and 4 ) .
information about the _ aggregate cycling potential _ on the road network is shown in the route network layer .
because the layer is the result of aggregating overlapping ` fast ' routes , and summing the level of cycling for each scenario ( see figure 5 ) , it relates to the _ capacity _ that infrastructure may need to handle .
cycling along otley road ( highlighted in figure 5 ) , under the go dutch scenario , rises from 73 to 296 commuters along a single route , but from 301 to 1133 in the route network .
note that more confidence can be placed in the relative rather than the absolute size of these numbers : the route network layer excludes within - zone commuters , commuters with no fixed workplace , and commuters working in a different region ( see figure 1 ) .
route network values also omit routes due to the adjustable selection criteria : maximum distance and minimum total numbers of all - mode commuters per od pair . at the time of writing
these were set to 20 km euclidean distance and 10 commuters respectively .
nationally , the route network layer under these settings accounts for around two thirds of cycle commuters .
this section describes and illustrates some outputs from the pct , alongside discussion of how these outputs could be used in transport planning .
note that some details of the graphics in the online version may evolve as the pct develops .
tabs are panels within the pct that reveal new information when clicked ( see the top of figure 3 ) . of these , the first four provide region - specific information : * * map * : this interactive map is the main component of the pct , and is the default tab presented to users .
it shows cycling potential at area , desire - line , route and route network levels under different scenarios of change , as described throughout this paper .
` popups ' appear when zones , desire lines or segments on the route network are clicked , presenting quantitative information about the selected element . * * lines * : when lines are displayed on the interactive map , this tab provides raw data on a sample of the variables as a table at the od pair level . * * areas * : this tab is the equivalent of the ` lines ' tab , but with data at the area level . * * model output * : this tab provides key statistics , diagnostic plots and model results for each region .
the document is produced by a ` dynamic document ' which runs embedded code for each regional dataset .
diagnostic plots include the distribution of cycling by trip distance under each scenario ( see figure 6 ) , providing insight into local travel patterns and how they relate to cycling potential in the region overall .
figure 6 shows how the proportion of trips made by cycling varies as a function of distance in two regions currently , and under the pct s four scenarios of change .
the frequency of all mode trips by distance band ( the red lines ) illustrates the two regions have different spatial structures .
oxfordshire has a high proportion short ( under 5 km ) trips , helping to explain the relatively high level of cycling there .
west yorkshire ( figure 6 , left ) , by contrast , has a higher proportion of longer distance commutes and a lower level of cycling than oxford .
note that under go dutch and ebikes scenarios , regional differences in the rate of cycling diminish , however , illustrating that these scenarios are not influenced by the current level of cycling .
the spatial distribution of cycling potential differs markedly between scenarios , as illustrated in figure 7 for the city of leeds , west yorkshire .
the top 6 od pairs ( a low number was used to focus on the city centre ) in leeds under government target are strongly influenced by the current distribution of cycling trips , concentrated in the north of the city ( see figure 3 for comparison with the baseline ) . under the go dutch scenario , by contrast , the pattern of cycling shifts substantially to the south .
the cycling patterns under the go dutch scenario are more representative of short - distance trips across the city overall . in both cases
the desire lines are focussed on leeds city centre : the region has a mono - centric regional economy , making commute trips beyond around 5 km from the centre much less likely to be made by cycling .
the same scenario is illustrated in figure 8 with the route network layer .
this shows how the number of commuter cyclist using different road segments could be expected to change .
the number using york road , highlighted in figure 8 , for example more than triples ( from 71 to 236 ) under government target and increases more than 10 fold under go dutch ( from 71 to 966 ) .
this contrasts with otley road ( highlighted in figure 5 ) , which ` only ' triples under go dutch .
these outputs suggest that the geographical distribution of cycling may shift if the proportion of trips cycled increases in the city .
the results also suggest that cycle paths built to help achieve ambitious targets , as represented by the go dutch scenario , should be of sufficient width to accommodate the estimated flows .
another potentially useful output is the difference between ` fast ' and ` quiet ' routes .
figure 9 illustrates this by showing routes in manchester with the highest cycling potential under the government target scenario .
the ` quiet ' route is longer : 2.6 km ( as shown by clicking on the line ) .
the ` fast ' route is more direct ( with a route distance of 2.3 km ) but passes along a busy dual carriage way .
the euclidean distance associated with this od pair is 1.6 km ( this can be seen by clicking on a line illustrated from the ` straight lines ' layer in the pct s interface ) , resulting in ` circuity ' ( see iacono et al . , 2008 ) , values of 1.6 and 1.4 for ` quiet ' and ` fast ' routes respectively .
dutch guidance suggests that circuity values `` for cycle provision should be 1.2 '' ( crow , 2007 ) .
evidence indicates that women and older people have a greater preference for off - road and shorter routes ( garrard et al . , 2008 ; woodcock et al . , 2016
this suggests the ` fast route ' option , if built to a high standard , may be favourable from an equity perspective in this context .
three basemap options are worth highlighting in addition to the grey default basemap .
these were selected to provide insight into how the geographical distribution of latent demand for cycling relates to current cycle infrastructure and socio - demographics : ` opencyclemap ' indicates where cycle provision is ( and is not ) currently ; ` index of deprivation ' illustrates the spatial distribution of social inequalities ; and the ` satellite ' basemap can help identify opportunities for re - allocating space away from roads and other land uses for cycle and walking paths by providing visual information on road widths and land uses along desire lines .
the health benefits of cycling do not necessarily rise in direct proportion to the number of people cycling .
longer trips lead to a greater health benefit than short ones and older people benefit more from increased physical activity .
further , health benefits along desire lines with a low pedestrian mode share can be expected to be greater .
this is demonstrated in the data presented by the pct .
the economic value of health benefits reported for the 4.1 km route in figure 5 is estimated to be 70785 .
when health benefits are the main criteria for policy evaluation , od pairs with low current rates of walking would be favoured for intervention . when emissions are the main criteria , od pairs with a high baseline level of car use are also favoured .
the exploration of these considerations is facilitated in the pct by allowing users to select the top routes ranked by health and carbon benefits .
we have outlined a method for modelling and visualising the spatial distribution of cycling flows , currently and under various scenarios of ` cycling futures ' .
inspired by previous approaches to estimating cycling potential ( larsen et al .
, 2013 ; zhang et al . , 2014 ) and by online , interactive planning support systems ( pss ) ( pettit et al . , 2013 ) , the pct tackles the issue of how to generate an evidence base to decide where new cycle paths and other localised pro - cycling interventions should be prioritised . by showing potential health - related benefits
the tool provides various metrics for transport planners , going beyond the number of additional trips .
illustrative uses of the pct demonstrated the potential utility of the tool , for example by showing settings in which the spatial distribution of cycling demand is likely to shift as cycling grows .
in addition to creating an evidence base for planning specific routes and area - based interventions , the long - term go dutch and ebikes scenarios could be used for envisioning different transport futures ( hickman et al .
the pct could also : help translate national targets into local aspirations ( as illustrated by the government target scenario ) ; inform local targets ( e.g. by indicating what the potential in one region is relative to neighbouring regions ) ; support business cases ( by showing that there is high cycling potential along proposed routes ) ; and help plan for cycle capacity along the route network via the network analysis layer .
ongoing case study work with stakeholders will be needed to establish and develop these uses .
future developments will be facilitated by the open source codebase underlying the pct ( see https://github.com/npct[github.com/npct ] ) ( lima et al . , 2014 ) .
as with any modelling tool , the approach presented in this paper has limitations : the reliance on census origin - destination ( od ) data from 2011 means that the results are limited to commuting and may not encapsulate recent shifts in travel behaviour , and the user interface is constrained to a few , discrete , scenarios .
these limitations suggest directions for future work , most notably the use of new sources of od data .
there is often a tension between transparency and complexity in the design of tools for transport planning .
excessive complexity can result in tools that are ` black boxes ' ( saujot et al . , 2016 ) . in a context of limited time , expertise , and resources ,
saujot et al .
caution against investing in ever more complex models . instead , they suggest models should be more user focussed
. the pct s open source , freely available nature will , we believe , facilitate the future development of the pct organically to meet the needs of its various users .
for instance , we envision stakeholders in local government modifying scenarios for their own purposes , and that academics in relevant fields may add new features and develop new use cases of the pct .
such enhancements could include : * additional scenarios to illustrate a wider range of ` cycling futures ' , including medium - term and local targets .
* use of individual level data to estimate cycling potential and impacts .
the use of synthetic ` spatial microdata ' , for example , could enable specification of scenarios and analysis of outcomes by a much wider range of predictors ( lovelace et al . , 2014 ) . * additional purposes of trips in the model .
an ` education layer ' would enable prioritisation of ` safe routes to school ' , building on methods analysing ` school commute ' data ( singleton , 2014 ) .
other data sources to include more trip types include mobile telephone providers ( alexander et al .
, 2015 ) and outputs from transport models .
* deployment of the pct for new cities , regions or countries .
this depends on the availability of appropriate od data , perhaps from sources mentioned in the previous point .
such work could also facilitate international comparisons of cycling potential .
transport planning is a complex and contested field ( banister , 2008 ) .
when it comes to sustainable mobility , policy , politics , leadership and vision are key ingredients that computer models alone can not supply ( melia , 2015 ) .
the approach described here can , however , assist in this wider context by providing new tools for exploring the evidence at high geographical resolution and envisioning transformational change in travel behaviours . by providing transport authorities , campaign groups and the public with access to the same evidence base
, we hypothesise that tools such as the pct can encourage informed and rigorous debate , as advocated by golub et al . (
in conclusion , the pct provides an accessible evidence base to inform the question of where to prioritise interventions for active travel and raises more fundamental questions about how models should be used in transport planning .
the pct was built as a collaborative effort .
the principal investigator of the project was jw , and the initial concept for the project came from jw and rl in response to a call from the department of transport .
ag led the creation of the model underlying the pct , and the generation of estimates of cycling levels , health gains and carbon impacts in each scenario .
jw and ag led the development of the methods for calculating health impacts .
jw , rl , ag , and ra contributed to the development of the cycling uptake rules .
rl led the processing of these modelled estimates for spatial visualisation in the online tool and coordinated the development of the online tool .
ra led the coordination of policy implementation and collection of practitioner feedback .
aa and nb led on the user interface , with contributions from all authors
. nb led the deployment of the pct on a public facing server .
ag led the writing of the appendix .
rl led the writing of this manuscript , with input from all authors .
we would like to thank the following people for comments on earlier versions of the manuscript and the development of the pct : roger geffen ( cycling uk ) , tom gutowski ( sustrans ) , helen bowkett ( welsh government ) , john parkin ( university of the west of england ) , nike groot and phil tate .
thanks to simon nutall and martin lucas - smith for access to and instructions on the use of the cyclestreets.net api .
thanks to barry rowlingson from the university of lancaster for developing functionality in the * stplanr * package that enabled the creation of the route network layer .
we are grateful to matthew tranter at dft for merging additional geographic data into the uk national travel survey , and to eva heinen , rick prins and thomas gtschi for their help in analysing dutch and swiss travel survey data .
thanks to alvaro ullrich for contributions to the code .
thanks to developers of open source software we have been able to make the pct free and open to the world .
we would also like to thank brook lyndhurst for assistance with the user testing , and all participants in the user testing sessions .
we would also like to thank shane snow and other staff at the dft for specifying the project s aims and providing feedback on early versions of the tool .
the work presented was funded by the department for transport ( contract no .
rm5019so7766 : `` provision of research programme into cycling : propensity to cycle '' ) , with contract facilitation and project management by brook lyndhurst in phase 1 , and by atkins in phase 2 .
rl s contribution was supported by the consumer data research centre ( esrc grant number es / l011891/1 ) .
jw s contribution was supported by an mrc population health scientist fellowship .
jw s and aa s contributions were supported by the centre for diet and activity research ( cedar ) , a ukcrc public health research centre of excellence funded by the british heart foundation , cancer research uk , economic and social research council , medical research council , the national institute for health research ( nihr ) , and the wellcome trust .
ag s contribution was supported by an nihr post - doctoral fellowship .
the views reported in this paper are those of the authors and do not necessarily represent those of the dft , brook lyndhurst , atkins the nihr , the nhs or the department for health .
aldred , r. , woodcock , j. , goodman , a. , 2016 .
does more cycling mean more diversity in cycling ?
transport reviews 36 , 2844 .
doi : https://doi.org/10.1080/01441647.2015.1014451[10.1080/01441647.2015.1014451 ] alexander , l. , jiang , s. , murga , m. , gonz , m.c .
validation of origin - destination trips by purpose and time of day inferred from mobile phone data .
transportation research part b : methodological 120 .
doi : https://doi.org/10.1016/j.trc.2015.02.018[10.1016/j.trc.2015.02.018 ] aultman - hall , l. , hall , f. , baetz , b. , 1997 .
analysis of bicycle commuter routes using geographic information systems : implications for bicycle planning .
transportation research record : journal of the transportation research board 1578 , 102110 .
doi : https://doi.org/10.3141/1578 - 13[10.3141/1578 - 13 ] broach , j. , dill , j. , gliebe , j. , 2012 . where do cyclists ride ?
a route choice model developed with revealed preference gps data .
transportation research part a : policy and practice 46 , 17301740 .
doi : https://doi.org/10.1016/j.tra.2012.07.005[10.1016/j.tra.2012.07.005 ] brmmelstroet , m. te , bertolini , l. , 2008 . developing land use and transport pss : meaningful information through a dialogue between modelers and planners .
transport policy 15 , 251259 .
doi : https://doi.org/10.1016/j.tranpol.2008.06.001[10.1016/j.tranpol.2008.06.001 ] de nazelle , a. , nieuwenhuijsen , m.j . ,
ant , j.m . , brauer , m. , briggs , d. , braun - fahrlander , c. , cavill , n. , cooper , a.r .
, desqueyroux , h. , fruin , s. , 2011 . improving health through policies that promote active travel : a review of evidence to support integrated health impact assessment .
environment international 37 , 766777 .
ehrgott , m. , wang , j.y.t . ,
raith , a. , houtte , c. van , 2012 . a bi - objective cyclist route choice model .
transportation research part a : policy and practice 46 , 652663 .
doi : https://doi.org/10.1016/j.tra.2011.11.015[10.1016/j.tra.2011.11.015 ] gaffron , p. , 2003 .
the implementation of walking and cycling policies in british local authorities .
transport policy 10 , 235244 .
doi : https://doi.org/10.1016/s0967 - 070x(03)00024 - 6[10.1016/s0967 - 070x(03)00024 - 6 ] garrard , j. , rose , g. , lo , s.k . , 2008 . promoting transportation cycling for women : the role of bicycle infrastructure .
preventive medicine 46 , 5559 .
doi : https://doi.org/10.1016/j.ypmed.2007.07.010[10.1016/j.ypmed.2007.07.010 ] golub , a. , robinson , g. , nee , b. , 2013 .
making accessibility analyses accessible : a tool to facilitate the public review of the effects of regional transportation plans on accessibility .
journal of transport and land use 6 , 1728 .
doi : https://doi.org/10.5198/jtlu.v6i3.352[10.5198/jtlu.v6i3.352 ] haklay , m. , singleton , a. , parker , c. , 2008 .
web mapping 2.0 : the neogeography of the geoweb .
geography compass 2 , 20112039 .
doi : https://doi.org/10.1111/j.1749 - 8198.2008.00167.x[10.1111/j.1749 - 8198.2008.00167.x ] heath , g.w . ,
brownson , r.c . ,
kruger , j. , miles , r. , powell , k.e . ,
ramsey , l.t . , services , t.f . on c.p . , 2006 . the effectiveness of urban design and land use and transport policies and practices to increase physical activity : a systematic review .
journal of physical activity & health 3 , s55 .
heinen , e. , panter , j. , mackett , r. , ogilvie , d. , 2015 .
changes in mode of travel to work : a natural experimental study of new transport infrastructure
. international journal of behavioral nutrition and physical activity 12 , 81 .
doi : https://doi.org/10.1186/s12966 - 015 - 0239 - 8[10.1186/s12966 - 015 - 0239 - 8 ] hickman , r. , ashiru , o. , banister , d. , 2011 .
transitions to low carbon transport futures : strategic conversations from london and delhi .
journal of transport geography , special section on alternative travel futures 19 , 15531562 .
doi : https://doi.org/10.1016/j.jtrangeo.2011.03.013[10.1016/j.jtrangeo.2011.03.013 ] kahlmeier , s. , kelly , p. , foster , c. , gtschi , t. , cavill , n. , dinsdale , h. , woodcock , j. , schweizer , c. , rutter , h. , lieb , c. , 2014 .
health economic assessment tools ( heat ) for walking and for cycling , methods and user guide .
world health organization regional office for europe , copenhagen , denmark 2014 .
larsen , j. , patterson , z. , el - geneidy , a. , 2013 .
but where ?
the use of geographic information systems in identifying locations for new cycling infrastructure .
international journal of sustainable transportation 7 , 299317 .
lovelace , r. , ballas , d. , watson , m. , 2014 . a spatial microsimulation approach for the analysis of commuter patterns : from individual to regional levels .
journal of transport geography 34 , 282296 .
doi : https://doi.org / http://dx.doi.org/10.1016/j.jtrangeo.2013.07.008[http://dx.doi.org/10.1016/j.jtrangeo.2013.07.008 ] lovelace , r. , beck , s. , watson , m. , wild , a. , 2011 .
assessing the energy implications of replacing car trips with bicycle trips in sheffield , uk .
energy policy 39 , 20752087 .
doi : https://doi.org/10.1016/j.enpol.2011.01.051[10.1016/j.enpol.2011.01.051 ] mccollum , d. , yang , c. , 2009 .
achieving deep reductions in us transport greenhouse gas emissions : scenario analysis and policy implications .
energy policy 37 , 55805596 .
doi : https://doi.org/10.1016/j.enpol.2009.08.038[10.1016/j.enpol.2009.08.038 ] oja , p. , titze , s. , bauman , a. , degeus , b. , krenn , p. , reger - nash , b. , kohlberger , t. , 2011 .
health benefits of cycling : a systematic review .
scandinavian journal of medicine & science in sports 21 , 496509 .
doi : https://doi.org/10.1111/j.1600 - 0838.2011.01299.x[10.1111/j.1600 - 0838.2011.01299.x ] parkin , j. , wardman , m. , page , m. , 2008 .
estimation of the determinants of bicycle mode share for the journey to work using census data .
transportation 35 , 93109 .
doi : https://doi.org/10.1007/s11116 - 007 - 9137 - 5[10.1007/s11116 - 007 - 9137 - 5 ] pettit , c.j .
, klosterman , r.e .
, nino - ruiz , m. , widjaja , i. , russo , p. , tomko , m. , sinnott , r. , stimson , r. , 2013 .
the online what if ?
planning support system , in : planning support systems for sustainable urban development .
springer , pp .
349362 .
pikora , t.j . , bull , f.c.l . ,
jamrozik , k. , knuiman , m. , giles - corti , b. , donovan , r.j . ,
2002 . developing a reliable audit instrument to measure the physical environment for physical activity .
american journal of preventive medicine 23 , 187194 .
doi : https://doi.org/10.1016/s0749 - 3797(02)00498 - 1[10.1016/s0749 - 3797(02)00498 - 1 ] pucher , j. , dill , j. , handy , s. , 2010 .
infrastructure , programs , and policies to increase bicycling : an international review .
preventive medicine 50 , s106s125 .
doi : https://doi.org / doi:%2010.1016/j.ypmed.2009.07.028[doi : 10.1016/j.ypmed.2009.07.028 ] rae , a. , 2009 . from spatial interaction data to spatial interaction information ?
geovisualisation and spatial structures of migration from the 2001 uk census .
computers , environment and urban systems 33 , 161178 .
doi : https://doi.org/10.1016/j.compenvurbsys.2009.01.007[10.1016/j.compenvurbsys.2009.01.007 ] saujot , m. , lapparent , m. de , arnaud , e. , prados , e. , 2016 . making land use
transport models operational tools for planning : from a top - down to an end - user approach .
transport policy 49 , 2029 .
doi : https://doi.org/10.1016/j.tranpol.2016.03.005[10.1016/j.tranpol.2016.03.005 ] singleton , a. , 2014 .
a gis approach to modelling co2 emissions associated with the pupil - school commute .
international journal of geographical information science 28 , 256273 .
doi : https://doi.org/10.1080/13658816.2013.832765[10.1080/13658816.2013.832765 ] tainio , m. , nazelle , a.j . de , gtschi , t. , kahlmeier , s. , rojas - rueda , d. , nieuwenhuijsen , m.j . , s , t.h . de , kelly , p. , woodcock , j. , 2016 .
can air pollution negate the health benefits of cycling and walking ?
preventive medicine .
doi : https://doi.org/10.1016/j.ypmed.2016.02.002[10.1016/j.ypmed.2016.02.002 ] tamminga , g. , miska , m. , santos , e. , lint , h. van , nakasone , a. , prendinger , h. , hoogendoorn , s. , 2012 .
design of open source framework for traffic and travel simulation .
transportation research record : journal of the transportation research board 2291 , 4452 .
doi : https://doi.org/10.3141/2291 - 06[10.3141/2291 - 06 ] woodcock , j. , edwards , p. , tonne , c. , armstrong , b.g .
, ashiru , o. , banister , d. , beevers , s. , chalabi , z. , chowdhury , z. , cohen , a. , franco , o.h . ,
haines , a. , hickman , r. , lindsay , g. , mittal , i. , mohan , d. , tiwari , g. , woodward , a. , roberts , i. , 2009 .
public health benefits of strategies to reduce greenhouse - gas emissions : urban land transport .
the lancet 374 , 19301943 .
doi : https://doi.org/10.1016/s0140 - 6736(09)61714 - 1[10.1016/s0140 - 6736(09)61714 - 1 ] zhang , d. , magalhaes , d.j.a.v . ,
wang , x. ( . ,
prioritizing bicycle paths in belo horizonte city , brazil : analysis based on user preferences and willingness considering individual heterogeneity .
transportation research part a : policy and practice 67 , 268278 .
doi : https://doi.org/10.1016/j.tra.2014.07.010[10.1016/j.tra.2014.07.010 ]
|
getting people cycling is an increasingly common objective in transport planning institutions worldwide .
a growing evidence base indicates that high quality infrastructure can boost local cycling rates . yet for infrastructure and other cycling measures to be effective , it is important to intervene in the right places , such as along ` desire lines ' of high latent demand .
this creates the need for tools and methods to help answer the question ` where to build ? ' .
following a brief review of the policy and research context related to this question , this paper describes the design , features and potential applications of such a tool .
the propensity to cycle tool ( pct ) is an online , interactive planning support system which was initially developed to explore and map cycling potential across england ( see http://pct.bike/[www.pct.bike ] ) .
based on origin - destination data , it models and visualises cycling levels at area , desire line , route and route network levels , for current levels of cycling , and for scenario - based ` cycling futures ' .
four scenarios are presented , including ` go dutch ' and ` ebikes ' , which explore what would happen if english people cycled as much as dutch people and the potential impact of electric cycles on cycling uptake .
the cost effectiveness of investment depends not only on the number of additional trips cycled , but on wider impacts such as health and carbon benefits .
the pct reports these at area , desire line , and route level for each scenario .
the pct is open source , facilitating the creation of additional scenarios and its deployment in new contexts .
we conclude that the pct illustrates the potential of online tools to inform transport decisions and raises the wider issue of how models should be used in transport planning .
|
the radiation from the innermost regions of active galactic nuclei ( agn ) is primarily emitted in the x - ray band , making x - ray spectroscopy a prime tool for studying this region .
moreover , the variability of the radiation can provide additional diagnostics .
flux and spectral variations in agn are usually nothing out of the ordinary .
the drivers and nature of this variability are known to depend on the timescale of the variability .
for example , changes in the accretion rate govern the longest timescales , and lead the continuum to become softer when brighter in the x - ray band @xcite .
absorption changes can drive strong variability , with timescales and spectral signatures that depend on the type of absorber .
in general , all timescales depend on the black hole mass ( see * ? ? ? * ; * ? ? ? * and references therein ) , where the timescales are longer with increasing black hole mass . in this light
, the x - ray spectral variability of fairall 9 stands out .
this luminous seyfert 1 galaxy ( @xmath0 ) , which has been well - studied throughout the years , is renowned for its lack of a warm absorber @xcite and its simple x - ray spectrum @xcite .
the spectrum is well described by a powerlaw continuum with cold and ionized reflection @xcite , possibly with an additional soft x - ray continuum component @xcite . in particular , from a multi - epoch spectral analysis including _ xmm _ data from 2009 and _ suzaku _ data from 2007 and 2010 , @xcite conclude that the spectral modeling of the soft excess is ambiguous .
both a multi - zone ionized reflection model and a model with an additional comptonization component with a single zone ionized reflector describe the data .
the accretion disk parameters are dependent on the model chosen . for example
the black hole spin is determined to be @xmath1 for a reflection - only soft excess and @xmath2 for a comptonization soft excess .
the high black hole mass of @xmath3 @xcite suggests low amplitude short - term variability , and indeed the source does not display strong intra - observation variability on hour to day timescales @xcite .
the discovery of surprisingly rapid x - ray flux dips in 10years of _ rxte _ monitoring however has put this picture of low amplitude variations on short timescales into question @xcite . during a dip the 2 - 10kev flux decreases by a factor of two or more .
these flux dips happen on a timescale of a few days , so slightly longer than the length of a typical long x - ray observation . while it was not possible to determine the reason for the flux dips from the _ rxte _ monitoring , several possibilities
have been suggested : inner disk or coronal instabilities , absorption from compton - thick clumps in the torus , or even a failed jet cycle .
the aim of this paper is to study the broad - band spectral shape and the nature of the soft x - ray excess in fairall 9 .
to do so we have obtained new high quality _ xmm _ ccd observations combined with _
nustar _ hard x - ray coverage targeted at answering these questions . in this paper
we present a spectral analysis of these data .
a future paper will discuss the variability and timing properties .
the paper is organized in the following way ; in [ reduction ] we discuss the data reduction followed by a description of the results from the spectral analysis in [ results ] . before finally discussing the results ( [ discuss ] )
, we briefly show in [ previous ] how our new insights are consistent with the spectra already presented in @xcite .
to learn more about the x - ray flux dips , we designed an observing campaign to provide a detailed spectral and temporal study of the source .
it consists of three parts , starting with a _ swift _ @xcite monitoring campaign to observe fairall 9 in the optical , uv , and x - ray bands for several months in 2013 and 2014 with an average cadence of 2-days to search for dips .
the earliest part of this monitoring which ranged from april to june 2013 was presented in @xcite . in that study we focused on the connection of the x - ray variability to the uv variability and found correlated optical / uv variability on all time scales sampled .
we reduced the newly added data in a similar fashion to that from the pilot study .
the resulting x - ray light curve is shown in fig .
[ light_f9 ] .
the dips are immediately apparent in a visual analysis of the x - ray light curve ( fig .
[ light_f9 ] ) . based on this monitoring ,
we triggered two _ xmm _ observations , one in a dip and one out of a dip constituting the second part of our campaign .
these observations are indicated as vertical red bars in the light curve .
the third part of our campaign ( indicated by the third red bar in the light curve ) corresponds to a joint _ xmm_-_nustar _ observation meant to reveal the exact broad - band spectral shape of fairall 9 .
this observation was also taken at high flux , although not as high as the non - dip observation .
the flux variability is also apparent from the unfolded _ xmm _
spectra shown in fig .
[ spectra_unfolded ] . with three observations taken at different flux states , we have a dataset that is well - suited to study possible flux - related spectral changes .
erg / s/@xmath4 indicating the flux state of the three newest _ xmm _ observations ( xmmc , xmmd , xmme ) as red bars .
it is apparent that xmmc was taken at very low flux , while xmmd and xmme were at high flux . ]
fairall 9 is a well - studied agn ; an overview of the existing data from _ xmm _ , _ suzaku _ and _ nustar _ is given in table [ nu_obs ] . throughout the paper , reference names are used for all the pointings .
this work focuses on the newest three _ xmm _ observations ( xmmc , xmmd , xmme ) and the _ nustar _ observation ( taken in the form of two immediately subsequent pointings ) simultaneous with xmme , i.e. parts two and three of our observing campaign described in the previous section .
part one will be presented together with an analysis of the temporal properties in a furthcoming paper . [ cols="^,^,^,^,^,^,^,^ " , ]
the results presented thus far are based on observations taken over a period of less than one year ( table [ nu_obs ] ) , while the observations considered in previous works spanned a period from 2000 to 2010 . changes to the geometry of the source are plausible over a several year period .
these changes might cause the flux ratios to the continuum to vary .
further , we also note the different sensitivities of the different satellites leads to a different average power law fitted to even the same band , which would affect the data / model residual plots as well .
therefore , instead of trying to make a comparable plot for the old data we simply try to fit the suza , suzb , and xmmb observations jointly in a similar fashion to the xmmc , xmmd , and xmme / nu observations to test whether the data have similar properties . for ease of fitting , we combine the two front - illuminated _ suzaku _ xis detectors .
we include the power law soft excess in our model , as it was already seen as a possibility for the soft excess when the same observations were analyzed in @xcite .
however it was already noted then that the soft x - ray energies of _ suzaku_-xis might not be very reliable because of contamination on the detector . to avoid this affecting our general conclusion about the spectra , we exclude all _ suzaku _ data below 1kev for the spectral modeling . the high energy cut - off is kept fixed at 1000kev , as it can not be determined from _ suzaku_-pin .
we find that a model similar to that described in section [ soft ] describes the old data reasonably well ( @xmath5 for 5386 dof ) , see fig .
[ residuals_previous ] .
in particular such a fit returns very similar best fit parameters ( @xmath6,@xmath7 ) to the fit to the newer data .
the spin and inclination measurements are in agreement with those found in section [ soft ] .
this implies that the source behavior remained mostly unchanged over the last 7 years
. found to fit the new _ xmm _ and _ nustar _ data .
the residuals below 1kev are consistent with the calibration uncertainity of _ xmm _ epic - pn . ]
in this paper we have studied three _ xmm _ observations of fairall 9 taken in 2013/2014 and one _ nustar _ observation in the same time period .
our main findings are : * we are able to confirm the absence of excess ionized or neutral absorption in the soft x - ray band .
we detect emission lines in the soft band , as well as a single absorption line at 21.7 that can not be associated with fairall 9 . *
we discover fractional residuals to a simple power law in the _ xmm _ band , which remain constant even as the source flux changes significantly .
this implies a constant fractional contribution from the different model components . *
we clearly detect blurred ionized reflection in the spectrum and are able to constrain the accretion disk parameters . * a soft excess additional to that due to blurred ionized reflection , described by an additional power law component is present .
in section [ rgs_data ] we noted that there is a noticeable absorption line at 21.7 , which agrees most closely with the ovii resonance line at 21.6 .
the hot phase of the interstellar medium is known to produce prominent ovii absorption lines ( see , e.g. , * ? ? ? * and references therein ) . however , assuming the line is produced by ovii absorption , it would have a velocity shift of @xmath8kms@xmath9 , which is beyond any velocity range in the galaxy , including high velocity clouds ( @xmath10kms@xmath9 ) , or uncertainties due to errors in the rgs absolute wavelength scale ( typically 0.004 or @xmath11kms@xmath9 ) .
the most likely and appealing origin of this feature is therefore the intergalactic medium lying in the line of sight towards fairall9 . assuming that the absorption feature detected at @xmath12 at 21.7 is produced by ovii
, this would correspond to @xmath13@xmath14 at @xmath15 . in general ,
the best indicator for igm absorption is the redshifted hi ly@xmath16 absorption line ( 1215.67 at rest ) , which is commonly found in fuv spectra of agn , including fairall9 ( see , e.g. , * ? ? ?
the redshift of our putative ovii feature is consistent with the hi absorption line redshifted to 1221 clearly seen in the hi ly@xmath16 profile of fairall9 ( see * ? ? ?
* fig.10 ) .
the 21.7 line is therefore most likely of intergalactic origin . from the rgs spectra ,
we confirm the absence of any significant absorption other than galactic , as noted previously by @xcite and @xcite .
in fact , we see a set of low energy emission lines including those consistent with o vii , o viii , and n vii . plotting the flux of these lines against the ionization state ( figure [ fig : xi_vs_flux ] ) indicates that the ionization state of most of the gas is around @xmath17 .
such low ionization emission lines are consistent with a warm absorber out of our line of sight . within the context of the unified agn model ,
this agrees well with our finding that the inclination of the inner accretion disk , as measured from the multi - epoch spectra , is very low and the accretion disk is almost face - on .
previous measurements of the inclination angle have found it to be more intermediate @xcite .
this discrepancy could could be due to that earlier analyses have used an angle averaged reflection model , which the analysis presented here uses the new angle - resolved ` relxill ` model @xcite .
for very low inclinations and intermediate ionization states such as found here , the differences between an angle - averaged and angle - resolved model are especially pronounced ( see figure 2 of * ? ? ?
* ) . as we showed in section[previous
] , a reanalysis of the old data with the new ` relxill ` model now also finds a low inclination .
a second complication of the inclination measurement already discussed in @xcite , is that for intermediate inclinations ( @xmath1845deg ) the cut - off of the blue wing of the iron line , which provides the best constraint on the inclination of the accretion disk , coincides with the fe k@xmath19 line , possibly leading to a false association of this energy as the end of the blue wing of the line .
the exact modeling of the highly ionized emission lines ( fe xxv , fexxvi ) adds additional complexity to this region of the spectrum .
future high - resolution observations with , for example , the microcalorimeter on board _ astro - h _ , have the potential to answer this question once and for all . in our initial exploration of the x - ray spectrum ( section [ res_ccd ] ) we discovered that the data / model residuals of an absorbed power law , with the absorption fixed at the galactic value , are very similar for all three _ xmm _ pointings ( fig . [ resid_powerlaw ] ) .
a similar observation has been made by @xcite in mrk335 where the soft excess component at 0.6kev was found to scale with the flux , yielding a constant flux ratio
. however , the persistence of the fractional flux contributions is much more apparent in fairall 9 .
this is most likely driven by the high black hole mass of fairall 9 and its aborption - free spectrum , simplifying a detection significantly .
the observed behavior has several implications for our understanding the x - ray spectra and spectral variability .
our analysis demonstrates that the relative contribution of each individual model component at a given energy remains constant .
the number and nature of the components also remains the same .
this implies that , the geometry of the x - ray emitting region can not have changed within the time spanned by the observations .
indirectly , the accretion rate must remain approximately constant as drastic changes in the accretion rate would cause changes in the geometry . to gain further insight into the geometry , we model the simultaneous observations xmme and nu above 2kev ( to avoid the complications of the soft excess ) with a lamp - post geometry and @xmath20 were kept fixed at their multi - epoch best fit values . ] , where the corona is approximated by a point source above the disk at heigh @xmath21 .
we find that the corona would be at a height of @xmath22 above the accretion disk .
the time difference ( @xmath1810days ) between the two observations xmmc and xmmd combined with the fact that we do not see a significant lag of the cold reflection flux behind the continuum flux , allows us to place an upper limit on the distance to the cold reflector of 690@xmath23 .
this indicates that the origin of the cold reflection is the outer part of the accretion disk and not a more distant structure such as the torus .
a radius of @xmath24 lies well within the dust sublimation radius of fairall 9 ( @xmath25 ) and our finding therefore agrees well the claim of @xcite that the dust sublimation radius represents an upper limit to iron emitting region .
having set out in this paper to find the driver of the relatively rapid variability seen in the x - ray monitorings of fairall 9 , the question now is : what is causing the variability we observe in the x - ray spectra of this agn ? from our results , it is clear that the variability of the individual components is driven by the variability of the continuum ( power law ) .
both the cold and the ionized reflection components respond to these changes in the continuum , as does the soft excess .
the question of the nature of the variability is therefore transferred to the question of what causes the continuum variability .
current models connect the continuum variability to changes of the accretion flow @xcite , for example through inward propagating fluctuations of the accretion rate @xcite .
but to fully understand how these accretion flow changes are connected to changes in the coronal emission will require a better understanding of the detailed physics governing the corona . our detailed spectral analysis of the multi - epoch data revealed an additional continuum - like component in the spectra above the blurred reflection . in section
[ soft ] we found that a simple , steep power law describes this component well .
this component varies together with the bulk of the continuum emission . its strength with respect to the primary continuum does not change , similar to what is seen for the reflection components .
the spectral shape and variability behavior ( it must vary on relatively fast time scales , which are shorter than the time difference between the observations ) suggest that comptonization could give rise to the observed power law .
one possibility is that the primary corona is non - uniform in its temperature or optical depth and this gives rise to two seemingly distinct but coordinated varying components . one plausible explanation for this would be in - homogeneous heating of the corona as , for example , caused by different magnetic field strengths at different radii .
another possibility is that this comptonization occurs in an area that is spatially separate from the primary comptonization , e.g. on the surface of the innermost part of the accretion disk .
the need for an additional comptonization component to model the broad - band _ xmm _ and _ nustar _ spectra has also been found in ark120 @xcite , and ngc5548 @xcite .
both analyses assume the two comptonization components are coupled , with the first corona upscattering the accretion disk photons and the second corona comptonizing the already comptonized photons leaving the first corona .
such a scenario was outlined in @xcite .
two spatially separate but connected comptonization components , however would be problematic for the case of fairall 9 . to match the observed constancy of the flux ratios , for example , a doubling in the flux output of the first corona would need to lead to a doubling of the flux output of the second corona as well .
a doubling in the seed photon flux however does usually not lead to a doubling of the coronal emission ; in fact it would lead to a significant cooling of the second corona .
the only way to ensure the flux ratios remain constant , would be a fine - tuned heating of the second / primary corona .
therefore for fairall 9 a non - uniform corona is more likely .
it is important to note here that a separate comptonization component in the soft x - ray would naturally also extend into the uv / optical bands .
this provides another means of verifying the presence of an additional component , as a single temperature comptonization component extending from the optical to the soft x - rays would predict a significant correlation of the bands at zero lag .
this is called into question by the discovery of notable lags between the x - ray emission and the uv / optical bands @xcite .
notably these lags are seen between all x - ray energies and the uv / optical emission , not just the soft x - rays , with no strong correlation seen at zero lag .
usually these lags are associated with the thermal reprocessing of x - ray coronal emission on the optical / uv emitting regions of the accretion disk .
possibly , multi - zone comptonization , for example on the outer layers of the accretion disk ( with multiple temperatures ) could mimic a similar effect . for fairall 9
previous work also suggests the presence of thermal reprocessing in uv / optical @xcite . we will discuss the timing and uv - x - ray insights from the new data in a forthcoming paper
we thank the referee for their comments that have helped to improve the clarity of the paper .
al and acf acknowledge support from erc advanced grant feedback .
csr thanks the simons foundation fellows program ( us ) and the sackler fellowship program ( cambridge ) for support .
this work made use of data from the _ nustar _ mission , a project led by the california institute of technology , managed by the jet propulsion laboratory , and funded by the national aeronautics and space administration .
this research has made use of the _ nustar _ data analysis software ( nustardas ) jointly developed by the asi science data center ( asdc , italy ) and the california institute of technology ( usa ) .
this research has made use of data obtained from the _ suzaku _ satellite , a collaborative mission between the space agencies of japan ( jaxa ) and the usa ( nasa ) .
|
we present a multi - epoch x - ray spectral analysis of the seyfert 1 galaxy fairall 9 .
our analysis shows that fairall 9 displays unique spectral variability in that its ratio residuals to a simple absorbed power law in the 0.5 - 10kev band remain constant with time in spite of large variations in flux .
this behavior implies an unchanging source geometry and the same emission processes continuously at work at the timescale probed . with the constraints from _
nustar _ on the broad - band spectral shape , it is clear that the soft excess in this source is a superposition of two different processes , one being blurred ionized reflection in the innermost parts of the accretion disk , and the other a continuum component such as spatially distinct comptonizing region . alternatively , a more complex primary comptonization component together with blurred ionized reflection could be responsible .
|
theoretical interest in spherical gravitational wave ( gw ) antennas dates back to the 1970s .
spherical antennas have a greater cross - section than a bar detectors of similar dimensions .
more importantly , they have both omnidirectional and omnipolarization sensitivity , and also the potential to detect the direction of wave provenance .
interest in experimental research into resonant spheres has increased over the past 15 years , and today spherical antennas are recognized to be the new generation of gravitational resonant detectors , to complement existing cylindrical antennas .
two experiments are under way : minigrail in leiden ( holland ) @xcite and the graviton project in sau paulo ( brasil ) @xcite .
one of the main problems today in detecting gravitational waves with resonant spheres is to find the `` best '' location for amplifiers on the detector surface in order to optimize sensitivity . to date
, no theoretical basis has been developed to provide a definitive solution to this problem .
nevertheless many suggestions have recently been proposed , including the uncoupled transducer configuration by zhou and michelson in 1994 , and the phc configuration by lobo and serrano in 1996 ; both are five - resonator configurations , the first including radial and tangential motion of the resonators , the second only radial motion . in addition
, the tiga configuration , proposed by merkowitz and johnson ( 1995 ) , consists of six transducers moving radially , and it enjoys a peculiar symmetry which greatly simplifies equations .
the general theoretical aspects of an ideal simple sphere interacting only with gravitational radiation have already been treated in the literature ( see for example or @xcite ) . in the first part of this paper we will work out briefly some of them and obtain the energies stored in the five fundamental sphere quadrupole modes , as well as their dependence on the incident wave direction and polarization .
+ taking into account of earth s rotation and the detector s location , we will translate the direction dependence into galactic longitude and sidereal time dependence for a source lying in the galactic plane .
we will review the general problem of the spherical detector with radial moving resonators on its surface , again in the case of high signal - to - noise ratio ( snr ) .
we then focus our attention on the tiga configuration , finding the oscillation amplitudes for the six transducers as a function of the incident gw direction .
+ with the same method used in the case of the simple sphere without amplifiers , we will find the resonators amplitude dependence on sidereal time and on galactic longitude for a source lying in the galactic plane which emits randomly polarized gw radiation .
the paper is organized as follows . in section 2
we briefly sketch some general aspects of the physics of gw resonant detectors without any assumption on the shape of the antenna . in section 3
we focus our attention on the spherical detector , and we present the solutions to the equations of motion followed by the five modes angular and sidereal time sensitivity . in section 4 results of the previous sections
are generalized to the case of the spherical detector coupled to a set of radial transducers .
section 5 contains the summary and conclusions .
we will work with a solid whose density is @xmath0 , where @xmath1 is the position of an infinitesimal mass element relative to the center of the solid and we call the lam coefficients of the material @xmath2 and @xmath3 we suppose that a gravitational wave hits our detector at the time @xmath4 .
if we call @xmath5 the small displacement from the equilibrium position at the point @xmath6 and time @xmath7 , the equation of motion turns out to be @xcite [ eom ] -^2(,t)-(+)((,t))=(,t ) , where @xmath8 is the @xmath9 component of the tidal force density attributable to the gw @xcite . as is well known for a resonant detector the spatial dependence of the riemann tensor can be neglected . moreover , as all our summation indices are spatial and we reasonably assume the background metric to be flat on the earth s surface ( @xmath10 ) , we can write them all as lowered indices . thus , in the first order of @xmath11 , the riemann tensor can be written @xmath12 which is a rank 3 traceless symmetric tensor .
it represents then a particle of spin 2 , the graviton .
+ such a tensor can be decomposed into a base of 5 matrices . in order to do that
, we follow the procedure used by lobo @xcite , working with a basis of real matrices instead of imaginary matrices , as is usually done : @xmath13 @xmath14 [ matrix ] m^0=. noting @xmath15 the identity matrix , the above matrices have the following properties : m^_ij m^_ij=_,i_ij m^_ij=0 , + i_iji_kl+ _
m^_ijm^_kl=(_ik_jl+_il_jk ) , with @xmath16 + then we can rewrite the @xmath17 component of the force density : [ fg ] & & f_j(,t ) = r_j0k0(t)r_k + & & = r_kr_l0m0(_mj_lk+_lj_mk ) + & & = _
s m^_jk r_k m_lm^r_l0m0= _
f^_j ^(t ) .
repeated indices are summed .
eq.([fg ] ) introduces the functions @xmath18 and @xmath19 .
+ direct computation gives for the @xmath20 : ^1c()= & ^1s()= + ^2c()= & ^2s()=&.[f ] + ^0()= we now want to write down the explicit expressions for the functions @xmath21 given a gw propagating along a generic direction with respect to the detector coordinates system . + in the frame where the gw propagates along the @xmath22 axis we can pass into the tt gauge in order to rewrite the riemann tensor as r_i0j0=_ij .
we call @xmath23 the angle from the @xmath22 axis and @xmath24 the angle between @xmath25 and the projection into the @xmath26 plane .
@xmath27 is the detector frame basis . for a generic arrival direction given by the angle @xmath28
, we have to perform a rotation in order to obtain the riemann in the antenna frame : r_i0j0=(^t)_ij with = .
we now are able to compute the components @xmath21 which are [ g ] ^0(t)&=&^2()_+(t ) + ^1c(t)&=&(-()()()_+(t)+()()_(t ) ) + ^1s(t)&= & ( ( ) ( ) ( ) _ + ( t)+()()_(t ) ) + ^2c(t)&= & ( ( 1+^2())(^2()-)_+(t)-2()()()_(t ) ) + ^2s(t)&=&((1+^2())()()_+(t)+()(^2()-^2())_(t ) ) .
+ note that we have defined the @xmath29 and @xmath30 polarization with respect to the detector frame . in the case where the source has an intrinsic polarization @xmath31 we have to make the substitution [ pol ] h_+&=&(2)e_++(2)e _
+ h_&=&-(2)e_++(2)e _ where @xmath32 describes the polarization angle .
this becomes useful if we want to take the average of the polarization .
the solution of eq.([eom ] ) can be expressed formally by means of the green function integral @xcite , getting : [ solgenerale ] ( , t)=__n_n^-1_n^_n^(t)_n ( ) where [ fgfrak ] ^_n&=&m^-1_solid_n()^ ( ) d^3 r,[fn_alpha ] + ^_n(t)&=&_0^t ^(t)_n(t - t ) dt and @xmath33 are the normalized ( i.e. they verifies @xmath34 ) eigenfunctions solutions of the corresponding homogeneous equation [ sta ] _
n^2_n+^2_n+(+)(_n)=0 [ equ_hom ] with the appropriate boundary conditions .
@xmath35 indicates a collective quantum number denoting the quantum state . putting the above result in a different form , i.e. writing the solution as [ b ] ( , t)=_n b_n(t)_n ( ) , it is obvious , as expected from @xcite that the quantity @xmath36 satisfies equation [ eomb ] & _
n(t)&+_n^2 b_n(t)=f_n(t)= _ ^_n^(t ) + & = & _ m^-1_solid_n()^ ( ) ^(t)d^3 r + & = & _ solidm^-1 ( ) r_j0k0(t)r_k _ n , j ( ) .
each mode is then formally equivalent to a one - dimensional harmonic oscillator with frequency @xmath37 , driven by a force per unit mass @xmath38 for such a system , the expression for the energy per mass unit adsorbed from the driving force is given by @xcite : [ nrj ] e_s=_-^+f_n(t)e^-i_ntdt^2 . before moving on to the case of the spherical detector , it would be intersting to use this general method on a cylindrical detector .
while we omit this exercise here , it can easily be verified that this general method reproduces the results for the eigenfunctions and the energy stored in oscillations in the well - known case of the one - dimensional homogeneous bar @xcite .
the normal eigenfunctions of eq.([sta ] ) for the sphere can be found in the literature ( see , for example @xcite ) .
note that they are usually given in terms of imaginary spherical harmonics . here
we will put all results in terms of real spherical harmonics , but the shape of the solutions is the same . + let us briefly recall the main points .
there are two kinds of solutions with boundary conditions @xmath39 at @xmath40 , with @xmath41 the stress tensor , conventionally defined as @xmath42 , with @xmath43 these solutions are called the @xmath44 ones and the @xmath45 ones : _
nl^t()&=&t_nl(r ) i y_l ( , ) + _
nl^s()&=&a_nl(r)y_l(,)-b_nl(r)i y_l ( , ) where each mode @xmath46 corresponds to the generic @xmath35 in eq.([sta ] ) , @xmath47 is a positive integer which represents the energy level for a fixed angular momentum @xmath48 and @xmath49 are the real spherical harmonics with kinetic momentum @xmath48 obtained by the imaginary ones @xmath50 in the following way : [ sphehar ] y_l,0&=&y_l,0 + y_l , mc&=&(y_l ,- m+(-1)^m
y_l,+m ) + y_l , ms&=&(y_l ,-
m+(-1)^m+1y_l,+m ) + with @xmath51 so defined , all the spherical harmonics ( [ sphehar ] ) are real quantities . for @xmath52
this gives : y_2,0&=&(^2- ) + y_2,1c&=&2 + y_2,1s&=&2 + y_2,2c&=&^22 + y_2,2s&=&^22 .
also , given a generic unity vector in spherical coordinates , @xmath53 we can see that y_2,(,)=m_ij^n_in_j where @xmath54 are the matrices defined in eq.([matrix ] ) .
@xmath55 @xmath56 and @xmath57 are scalar functions of @xmath58 : t_nl(r)&=&c(n , l)j_l(k_nlr ) , + a_nl(r)&=&c(n , l ) , + b_nl(r)&=&c(n , l)where the @xmath59 are spherical bessel functions and _ 0(x)&=&,_1(x)=,_2(x)=j_l(x ) , + _
3(x)&=&_2(x)+@xmath60_0(x ) . the @xmath61 and the @xmath62 are related to the quantized @xmath63 of eq.([sta ] ) by k_nl^2=q_nl^2=. we start by computing @xmath64 of eq.([fgfrak ] ) for our spherical detector for both toroidal and spheroidal eigenfunctions : [ an ] ^s_nl & = & m^-1_sphere^s_nl()^()d^3r = a_n_l,2_, + ^t_nl & = & m^-1_sphere^t_nl()^()d^3r=0 , with [ a_n ] a_n=-_0^rr^3(a_n2(r)+3b_n2(r))dr .
it is clear from eq.([an ] ) that toroidal modes do not enter into play , so we will work from now on only with spheroidal modes and we will drop the @xmath65 for simplicity . once the riemann tensor and the @xmath21 are known for a gravitational wave coming from a generic direction @xmath28 , it is easy to compute @xmath66 using eq.([fgfrak ] ) .
the solution of eq.([eom ] ) can be written : ( , t)=_n=1^_n(,t)=_n=1^__n2()^_n2(t ) .
for spheroidal modes , @xmath67 as defined in the previous paragraph . in this case ,
@xmath68 and its equation of motion ( [ eomb ] ) is [ beomsf ] _
n , l,+_n2 ^ 2b_n , l,=_l,2a_n_(t ) .
the energy stored in the mode @xmath69 at frequency @xmath70 is ( from eq.([nrj ] ) ) : [ nrjs ] e_s(n , l,)=_-^+f_nl(t ) e^-i_nltdt^2 . using eqs.([eomb ] ) and ( [ an ] ) , eq.([nrjs ] ) becomes e_s(n , l,)=a_n^2_l,2 _
-^+^()(t)e^-i_n2tdt ^2 .
the only contributions have , as expected , @xmath71 for these , we fix the value of @xmath47 and replace @xmath72 using eq.([g ] ) . for the @xmath73 mode
, we have _ -^+^(0)(t)&&e^-i_n2tdt = _ -^+^2_+(t)e^-i_n2tdt + & = & ^2 ^ 2_n2_-^+h_+(t)e^-i_n2tdt + & = & ^2 ^ 2_n2_+(_n2 ) , thus e_s(n,2,0)=a_n^2 ^ 4 ^ 4_n2_+(_n2)^2 .
similarly , for the other modes , we get : [ ex ] e_s(n,2,1c)&=&a_n^2 ^ 4_n2 [ ^2 ^ 2 ^ 2_+^2+^2 ^ 2_^2 -^22 ( _ + ^ * _ ) ] , + e_s(n,2,1s)&=&a_n^2 ^ 4_n2 [ ^2 ^ 2 ^ 2_+^2+^2 ^ 2_^2- ^22 ( _ + ^ * _ ) ] , + e_s(n,2,2c)&=&a_n^2 ^ 4_n2 [ ( 1+^2)^2(^2-)^2_+^2 + 4 ^ 2 ^ 2 ^ 2_^2 + & -&4(1+^2)(^2- ) ( _ + ^ * _ ) ] , + e_s(n,2,2s)&=&a_n^2 ^ 4_n2 [ ( 1+^2)^2 ^ 2 ^ 2_+^2+^2(^2-^2)^2_^2 + & + & 2(1+^2)(^2-^2 ) ( _ + ^ * _ ) ] .
+ we have assumed that @xmath74 and @xmath75 decrease fast enough at @xmath76 to allow the integration by parts , @xmath77 and @xmath78 stand for the fourier transform of @xmath79 and @xmath80 in this paper we use the notation @xmath81 for the fourier transform and @xmath82 for the inverse . in eq.([ex ] ) all fourier transforms are evaluated at the frequency @xmath83 .
if the source emits randomly polarized radiation , we can rewrite @xmath84 and @xmath85 in terms of one possible intrinsic polarization of the source ( eq.([pol ] ) ) and average over @xmath32 .
we finally obtain [ nrj ] & & |e_s(n,2,0)=a_n^2 ^ 4_n2(_+^2+_^2)^4 , + + & & |e_s(n,2,1c)=a_n^2 ^ 4_n2(_+^2+_^2)^2(^2 ^ 2+^2 ) , + & & |e_s(n,2,1s)=a_n^2 ^ 4_n2(_+^2+_^2)^2(^2 ^ 2+^2 ) , + & & |e_s(n,2,2c)=a_n^2 ^ 4_n2(_+^2+_^2)[(1+^2)^2(^2-)^2 + 4 ^ 2 ^ 2 ^ 2 ] , + & & |e_s(n,2,2s)=a_n^2 ^ 4_n2(_+^2+_^2)[(1+^2)^2 ^ 2 ^ 2+^2(^2-^2)^2 ] .
+ as before , all fourier transforms are evaluated at the frequency @xmath86 it is important to stress at this point that if we sum the energies of eq.([nrj ] ) we obtain a constant value : sensitivity of our detector is always the same , independently of the gw direction .
this is to be expected because of the symmetry of the detector , and it is one of the most important properties differentiating spherical detectors from cylindrical ones . together with its greater sensitivity ( keeping the same dimensions ) , this feature makes the new generation of resonant detectors much more advantageous than bars .
figure 1 shows , for each mode @xmath47 , how the energy stored in each @xmath69 varies as a function of @xmath87 and @xmath88 the angles that give the direction of the incoming unpolarized gw .
the rate @xmath89 is zero where the graph is blue , and maximal ( @xmath90 ) where the graph is red .
[ fig1 ] the ratio @xmath91 is zero where the graph is blue , and maximal ( @xmath90 ) where the graph is red ( color online).,width=170,height=170 ] @xmath92 the ratio @xmath91 is zero where the graph is blue , and maximal ( @xmath90 ) where the graph is red ( color online).,width=170,height=170 ] @xmath93 the ratio @xmath91 is zero where the graph is blue , and maximal ( @xmath90 ) where the graph is red ( color online).,width=170,height=170 ] @xmath94 the ratio @xmath91 is zero where the graph is blue , and maximal ( @xmath90 ) where the graph is red ( color online).,width=170,height=170 ] @xmath95 the ratio @xmath91 is zero where the graph is blue , and maximal ( @xmath90 ) where the graph is red ( color online).,width=170,height=170 ] @xmath96 [ fig2 ] and of sidereal time in sidereal hours , ( @xmath7 ) , for a sphere near leiden , nl ( latitude @xmath97 , longitude @xmath98 ) . as in figure 1 ,
red zones indicate maximal values ( color online ) .
, width=170,height=170 ] @xmath99 and of sidereal time in sidereal hours , ( @xmath7 ) , for a sphere near leiden , nl ( latitude @xmath97 , longitude @xmath98 ) . as in figure 1 ,
red zones indicate maximal values ( color online ) .
, width=170,height=170 ] @xmath100 and of sidereal time in sidereal hours , ( @xmath7 ) , for a sphere near leiden , nl ( latitude @xmath97 , longitude @xmath98 ) . as in figure 1 ,
red zones indicate maximal values ( color online ) .
, width=170,height=170 ] @xmath101 and of sidereal time in sidereal hours , ( @xmath7 ) , for a sphere near leiden , nl ( latitude @xmath97 , longitude @xmath98 ) . as in figure 1
, red zones indicate maximal values ( color online ) . ,
width=170,height=170 ] @xmath102 and of sidereal time in sidereal hours , ( @xmath7 ) , for a sphere near leiden , nl ( latitude @xmath97 , longitude @xmath98 ) . as in figure 1 ,
red zones indicate maximal values ( color online ) .
, width=170,height=170 ] @xmath103 [ fig3 ] of figure 2 as a function of @xmath7 in sidereal hours for b=0 ( source near the galactic center).,width=207,height=207 ] the total sensitivity is the sum of contributions coming from the five modes @xmath69 and the sensitivity of each mode depends on the orientation of the source with respect to the chosen frame .
this means that for different angular positions of the source with respect to the detector reference frame , there will be different distributions among the five modes of the total energy transfered from the gw to the sphere .
in contrast , if the source is fixed in space , and the sphere moves because of the earth s rotation , the energy ratio stored in each of the five modes will be a periodic function of the arrival time , with period one sidereal day , while the sum of the five contributions will be time - independent .
the expected sources for gw resonant detectors are traditionally distinguished into : 1 ) burst of relatively great intensity , given by cataclysmic events as binary cohoaleshing , or supernovae explosions ; 2 ) sources which emit gw bursts occasionally but repeatedly in time and with peculiar statistic features , the so called `` gw - bursters '' @xcite ; 3 ) periodic sources , of longer duration and smaller spread in frequency , as pulsars or inspiraling binaries .
+ such astrophysical objets are present in galaxies , and as the energy hitting a detector decreases as the distance of the source squared @xcite , it is then reasonable to expect a gw signal coming from our galaxy to be more easily detected .
once the position of the source is known in galactic coordinates @xmath104 a set of space rotations gives the position of the source in the detector frame @xmath105 these rotations are derived and explained in detail in the appendix of @xcite . for a given @xmath106 one can then easily find the precise position @xmath107 of the source with respect to the detector frame , with sidereal time @xmath7 varying from 0 to 24 .
as example , figure 2 displays the energy rates @xmath91 stored in the five modes @xmath69 of a sphere near leiden , holland ( latitude @xmath97 , longitude @xmath98 , as functions of greenwich sidereal time ( in sidereal hours ) and of galactic longitude @xmath108 for a randomly polarized source lying in the galactic plane ( @xmath109 ) , where stars population is more important ( and then their remnants population also ) . as before , red zones indicate maximal values .
+ finally , the example of a section ( at @xmath110 , i.e. for a source near the gc ) for the @xmath73 mode is represented in figure 3 .
these graphics can be specially interesting for sources of type 2 and 3 mentioned above , for which a signal from the same location persists in time .
until now we have looked at the spherical detector as a unique object interacting with gw .
nevertheless , because of the smallness of @xmath11 , the magnitude displacement of the detector as a result of this interaction is tiny , of the order of @xmath111 , @xmath112 being the detector size .
for this reason small resonators of equal mass @xmath113 and elastic constant @xmath114 are put on the surface of the gw resonant detector .
the small resonators are tuned to the fundamental frequency of the detector , in order to amplify displacements of its surface at that chosen frequency , which for the sphere turns out to be @xmath115 . in this way
, frequencies other than the fundamental mode and nearby frequencies will not be amplified by the system .
imagine having @xmath116 such resonators on the surface of the sphere .
we are interested in what happens when a gw interacts with such a system , in the ideal case of high snr ( i.e. where noises are negligible ) . + in line with previous work , in deriving the equations of motion for such a system we will use the formalism and notations of , where the problem has already elegantly been posed , and a proposal for a six - detector configuration , the tiga configuration , is presented .
the equation of motion ( [ eom ] ) for small displacements now becomes [ eomres ] -^2(,t)-(+)((,t))=_^()()g^()(t ) where @xmath117 is a parameter that takes the values @xmath118 it is clear that the first components of this force density , _
gw=_^()g^(t)=_^()h^(t ) , are the usual ones ( see eq.([fg ] ) ) attributable to gw , while [ fi ]
_ res=_i=1^k^(i ) ( ) g^(i)(t ) are new contributions given by the elastic terms of resonators .
let us focus now on one resonator , which we assume to have only radial motion , linked with a tuned spring to the surface of the sphere , say at the position @xmath119 .
we define the scalar - time dependent quantity [ z ] z_i(t)=(_i , t)_i as the radial projection of the surface displacement at this point .
+ we also define @xmath120 to be the relative distance between the resonator and the sphere surface .
note that @xmath120 is not an inertial coordinate , but @xmath121 is .
then the equation of motion for this last quantity is [ z+qeom ] m_r(_i(t)+_i(t))=-k_r q_i(t)+f_i^gw(t ) , where @xmath122 is the radial component of the gw force density acting on the resonator , which will be neglected as we know that @xmath113 . since g^i(t)=q_i(t),^i()=k_r^3(-_i)_i
, the force density component ( [ fi ] ) can be written : [ fi ] _
res=_i k_r q_i(t)^3(-_i)_i .
generalizing the same method used above ( in particular , see eqs.([solgenerale ] ) , ( [ fgfrak ] ) and ( [ b ] ) ) , the solution of eq.([eomres ] ) can be expressed as : [ balpha ] ( , t)&=&__n , l,_n , l,^-1_n , l,^_n , l,^(t)_n , l , ( ) + & = & _ n , l , b_n , l,(t)_n , l , ( ) with _ n , l,^&=&_sphere_n , l,()^()d^3r + _ n , l,^(t)&=&_0^t g^(t)(_n , l,(t - t))dt. the last equality of eq.([balpha ] ) defines b_n , l,(t)=__n , l,^-1_n , l,^_n , l,^(t ) . in our case
we get ^_n , l,=\ { ll a_n , l_, & + a_n , l(r)y_n , l,(_i,_i ) & .
it is then straightforward to derive the equations of motion for the sphere modes @xmath123 by inserting eq.([balpha ] ) into eq.([eomres ] ) : [ beomot ] & & _ n , l,(t)+_n , l,^2b_n , l,(t)=__n , l,^g^(t ) + & & = a_n , l^(t)+_i=1^ka_n , l(r)y_n , l,(_i,_i)q_i(t ) . + when noises are negligible , the motion is dominated by quadrupole modes ( @xmath124 ) , the only ones interacting with gw .
we limit ourselves to the @xmath125 modes , for which the cross - section is the most significant @xcite .
note that for @xmath52 , @xmath126 of eq.([a_n ] ) .
+ from now on , we simplify the notation , writing b_1,2,(t ) = b_(t),a_1,2 & = & a , a_1,2(r)=a + y_1,2,(_i,_i ) & = & y_(_i,_i)=y_i .
this last equality defines the @xmath127 matrix @xmath128 multiplying for the sphere mass @xmath129 , and calling the five equal quantities @xmath130 eq.([beomot ] ) can be rewritten [ beom ] m_(t)+k_sb_(t)=ma^(t)-_ik_ray_iq_i(t ) .
defining now the vectors & & \{q_i(t)}=(t ) , \{ma^(t)}=\{^(t)}=(t ) , + & & \{b_(t)}=(t ) , we can again simplify the expression for these equations : [ matrix1 ] m_5(t)+k_s_5(t)-k_r a(t)=(t ) .
it is evident that @xmath131 and @xmath132 are 5-component vectors , @xmath133 is a @xmath116-component vector and @xmath134 the identity matrix @xmath135 thus eq.([matrix1 ] ) is a system of 5 equations .
+ in the same way we can rewrite eq.([z+qeom ] ) in matrix form . to do that
, we need to express @xmath136 as a function of @xmath137 . using the approximation that only the @xmath125 modes of the sphere are excited at high snr , from eq.([z ] )
we have : z_i(t)=(_i , t)_i_n=1,l=2,b_(t)ay_i . defining the vector @xmath138 , [ vecz ] ( t)=a^t(t ) .
then eq.([z+qeom ] ) becomes [ matrix2 ] m_r((t)+a^t(t))+k_r(t)=. the @xmath116 equations for @xmath133 ( [ matrix2 ] ) and the 5 equations for @xmath131 ( [ matrix1 ] ) can be rewritten as a system of @xmath139 equations [ star ] & & @xmath140 @xmath141 ( t ) + & + & @xmath142 @xmath143 ( t)=@xmath144 .
these are the equations of motion for a system composed of a homogeneous sphere with @xmath116 resonators moving radially on its surface , all interacting with a gw . in the remainder of this paper , we will solve these equations once a configuration for the positions of the resonators is given .
we first look at what happens in the case with only one resonator ( @xmath145 ) , in order to test the equations by looking at a simple example .
then we move on to the tiga configuration ( @xmath146 ) proposed by merkowitz and johnson in .
+ in this particular case we choose to put only one resonator on the surface of the sphere .
obviously each position is equivalent , but for simplicity let us put the resonator at the north pole of our coordinates system ( or , equivalently , at the south pole because of the symmetry of the problem ) . in this case , the only nonvanishing component of the @xmath148 matrix @xmath149 is @xmath150 then eq.([star ] ) can be written : @xmath151 & & @xmath152 + + @xmath153 & & @xmath154 = @xmath155.[star1 ] + it is clear from eq.([star1 ] ) that the only mode of the sphere which couples with our resonator is @xmath156 , while all the others behave as if the resonator was not present .
their equations of motion are the same as given above in eq.([beomsf ] ) .
+ so we can reduce our system of equations to a system of 2 coupled equations with 2 unknowns , taking only the first and the last lines of ( [ star1 ] ) : [ eomqb ] _ 0(t)+_0 ^ 2b_0&=&_0 ^ 2ayq+a^0(t ) + ( t)+_0 ^ 2q(t)&=&-ay_0(t ) . in eq.([eomqb ] ) we consider that the resonator frequency is perfectly tuned with the fundamental frequency of the sphere , i.e. that @xmath157 and we call @xmath158 the ( small ) ratio between the resonator mass and the sphere mass . in order to solve eq.([eomqb ] ) we move to fourier coordinates , giving : [ qbzero ] ( ) & = & + _ 0()&=&a^2 ^ 0 ( ) . as expected , the resonant frequencies @xmath159 for both the coupled quantities are the same , and they are the splitting of the original frequency @xmath160 . + they can be found setting the denominators of eq.([qbzero ] ) equal to zero : _ ^2&=&_0 ^ 2 ( 1ay+)= + & = & _ 0 ^ 2(1 + + o(^2 ) ) .
+ at the frequency @xmath160 we have @xmath161 and @xmath162 is proportional to the ratio between the sphere mass and the resonator mass , so we will have the greatest oscillations if we make the resonator mass as small as possible . note
that here we neglected the damping of both the coupled oscillators .
this is because we are interested in burst signals , where the time - scale for damping an oscillation of the system is much longer than the time duration of a signal .
taking account of damping means changing eq.([qbzero ] ) to cancel divergences which arise at @xmath163 and make them smoother . in particular , @xmath164 will no longer vanish .
computations adding damping terms are analog and easily done .
figure 4 shows a plot of @xmath165 with a sphere damping of @xmath166 , a resonator damping of @xmath167 , @xmath168 , @xmath169 and @xmath170 [ fig4 ] with a sphere damping of @xmath166 , a resonator damping of @xmath167 , @xmath168 , @xmath169 and @xmath170 , width=207,height=207 ] the quantity we are really interested in is the one we can measure , @xmath171 or its squared absolute value . rewriting @xmath172 to show its dependence on @xmath173 , @xmath174 and on the wave direction ( see eq.([g ] ) )
we get : |()|^2&=&^4|_+()|^2 + & = & ( ) ^4|_+()|^2 , or , assuming that the gw is unpolarized , and averaging over polarizations , = ^4 ( ) , with [ beta ] ( ) = .
this turns out to be the same angular sensitivity obtained for the cylindrical bar @xcite , for which one finds that sensitivity is optimal for a plane orthogonal to the bar and drops with a factor @xmath175 too .
+ having only one transducer on the sphere surface is a simple example showing that the way resonators are positioned can result in the loss of the main initial properties given by the spherical symmetry of the detector . in order to keep these properties as far as possible
, we have to put more resonators on the sphere in appropriate positions . with the hope that , in future , we will be able to place resonators moving radially on the sphere surface , and that we will be able to measure experimentally the displacements @xmath133 of these masses , and assuming that all other interactions with the sphere - resonators system are negligible with the exception of gravitational waves , in this section we find the response of @xmath176 in the tiga configuration to a gravitational wave coming from a generic direction @xmath177 in order to do so , we have to solve the equations of motion ( [ star ] ) in fourier space . in the tiga configuration
there are a total of six resonators : three of them have azimuth angle @xmath178 and @xmath179 and the other three have @xmath180 and @xmath181 .
the angles @xmath182 and @xmath183 are between @xmath184 and @xmath185 and they are solutions of the equation @xmath186 with this information , the @xmath149 matrix for this configuration is easily found : @xmath187 it is important to recall a very usefull property of this matrix : @xmath188 is proportional to the identity @xmath189 matrix . nevertheless , the inverse is not true : @xmath190 has all diagonal values equal to @xmath191 and all non diagonal values equal to @xmath192 it is easily seen that sums over lines or columns are always zero . actually , @xmath190 is a @xmath193 matrix obtained from two @xmath194 matrices , so it is clear that it can not be invertible .
using this property it is straightforward to solve the equations for the so - called @xmath195 @xcite [ p ] ( ) =( ) , [ modch ] _ ( ) = , and for @xmath196 _ ( ) = .
the denominators vanish for the two split frequencies ^2_=_0 ^ 2 , so one can rewrite eq.([modch ] ) as _ ( ) & = & - + & = & -^2 a^ , with _ + = , _ -= .
we then have an expression giving a linear combination of resonator displacements as a function of the spherical components of a gravitational wave , or we have each spherical component of @xmath197 as a function of different combinations of displacements .
nevertheless we are also interested in the inverse relation : we want to write each @xmath198 as a combination of the @xmath199 in order to do that we have to express 6 quantities as a function of 5 , which means inverting the matrix @xmath200 a @xmath194 matrix .
we then need a supplementary condition for @xmath133 .
the condition to be added is true for the tiga configuration only , in the case of a high snr .
in that case , the sum of all the six displacements is zero , and this can be seen by looking at the only non - trivial solution of @xmath201 this means that we can add a sixth line to the @xmath149 matrix , a line made up of six unities , and a sixth null component to the @xmath202 vector in eq.([p ] ) .
we will call the new @xmath193 matrix @xmath203 and the new 6-vector @xmath204 @xmath203 is invertible , and we can easily find @xmath133 using @xmath205 getting : _ 1()&=&_0()+(+1)_1c()+(-1)_2c ( ) ; + _ 2()&=&_0()-(+1)_1c()+(+1)_1s()-(-1)_2c ( ) + & -&(-1)_2s ( ) ; + _ 3()&=&_0()-(+1)_1c()-(+1)_1s()-(-1)_2c ( ) + & + & ( -1)_2s ( ) ; + _ 4()&=&-_0()+(-1)_1c()+(-1)_1s()-(+1)_2c ( ) + & + & ( + 1)_2s ( ) ; + _ 5()&=&-_0()-(-1)_1c()+(+1)_2c ( ) ; + _ 6()&=&-_0()+(-1)_1c()-(-1)_1s()-(+1)_2c ( ) + & -&(+1)_2s ( ) . taking the squared absolute value , and rewriting @xmath206 as a function of @xmath207 and of the direction @xmath28 ( see eq.([g ] ) )
we can find the explicit behavior of the amplitude of the oscillations for each resonator as a function of the gw direction and polarization .
supposing the source to be randomly polarized , we can replace @xmath208 with the expressions in eq.([pol ] ) , and average over polarizations ( as we have already done for the sphere without resonators ) .
this is equivalent to replacing |_+|^2 & & + the ratios @xmath209 with @xmath210 , give the sensitivities of the six resonators as a function of the direction of gw provenance .
+ the explicit expression for one of the @xmath211 previously defined is : & & a_2=^2 + & & + ^2 .
the other five @xmath211 have analogous expressions .
their angular dependence is shown in figure [ fig5 ] , while figure 6 shows the angular dependence of their sum .
note that the sum of the six sensitivities is no longer a constant .
this is because , as we noted before , putting resonators on the surface of the sphere means destroying the original spherical symmetry . nevertheless , the total sensitivity is never zero , so for the direction of a gw from any provenance , there is at least one resonator which will be excited significantly .
omnidirectional sensitivity is then preserved in the tiga configuration , even if omnidirectional symmetry is broken . in figure [ fig7 ]
we take into account the earth s rotation , as explained in the appendix of @xcite , for a fixed source near the galactic center .
angular sensitivities for the six resonators and for their sum are shown as a function of sidereal hour and galactic longitude .
as before , we take as an example the sphere near leiden , holland ( latitude @xmath97 , longitude @xmath98 ) .
in this work , the general problem of the interaction of a spherical detector with gws has been analyzed , in both the cases of a simple elastic sphere and of a system sphere + transducers .
the five responses of the quadrupole fundamental modes and of resonators motions for the tiga configuration have been found , explicitly identifying angular dependence and , for gw radiation coming from a source in the galactic plane , sidereal time dependence as well .
we find the expected omnidirectional constant sensitivity for the ideal case of a simple sphere . this omnidirectional sensitivity is shown to be preserved in the case of the sphere equipped with tiga transducers , nevertheless it is interesting to note that it is no longer a constant .
this confirms the fact that putting resonators on the surface of the sphere means choosing some favorite directions , breaking the omnidirectional initial symmetry . for current experiments ,
the results of this paper allow one to predict exactly how the resonators should behave once the position of a candidate source is known .
vice versa , relations between the resonators amplitude and @xmath107 can be inversed and whenever there is a statistically significative reason for thinking that a gravitational wave signal has been received , one can estimate the source direction by looking at the ratio between resonators excitations and at the gw arrival time .
nevertheless , all this work is based on an ideal sphere and ideal radial moving resonators .
deviations and errors on the above predictions exist in a real spherical detector experiment .
in particular , the asymmetry caused by the hole drilled through the center for the suspension system induces a degeneracy of the quadrupole five fundamental modes , which in is estimated to be less than 1@xmath212 .
we have not derived the consequences of this degeneracy in this paper , but it would be interesting to do that in order to better match theoretical predictions with experiments .
60 www.minigrail.nl .
www.das.inpe.br/ graviton . c. z. zhou and p. f. michelson , phys .
d * 51 * , 2517 ( 1994 ) .
j. a. lobo and m. a. serrano , europhys .
35 * , 253 ( 1996 ) . s. m. merkowitz and w. w. johnson , phys . rev .
d * 51 * , 2546 ( 1995 ) .
s. m. merkowitz and w. w. johnson , phys .
d * 56 * , 7513 ( 1997 ) .
r. v. wagoner and h. j. paik , _ gravitazione sperimentale _ , accademia nazionale dei lincei , rome , 1977 . n. ashby and j. dreitlein , phys . rev .
d * 12 * , 336 ( 1975 ) .
l. landau , e. lifshitz , _ thorie de llasticit _ , moscou , 1960 . c. w. misner , k. s. thorne , j. a. wheeler , _ gravitation _
, w. h. freeman @xmath220 company ( 1970 ) .
j. a. lobo _ what can we learn about gravitational wave physics with an elastic spherical antenna ? _ , phys .
d * 52 * , 592 ( 1995 ) .
l. landau , e. lifchitz , vol.1 _ mcanique _ , moscou , 1960 .
e. coccia and j. a. lobo and j. a. ortega , phys .
rev d * 52 * , 3735 ( 1995 ) .
k. s. thorne , in _ 300 years of gravitation _ , edited by s.hawking and w. israel ( cambridge university press , cambridge , england , 1987 ) .
s. m. merkowitz , ph.d .
thesis , louisiana state university ( 1995 ) .
f. dubath , s. foffa , m. a. gasparini , m. maggiore and r. sturani _ experimental signature of gravitational wave bursters _ phys .
d * 71 * , 124003 ( 2005 ) .
e. coccia , f. dubath , m. maggiore _
possible sources of gravitational waves bursts detectable today _ phys .
d * 70 * , 084010 ( 2004 ) .
i would like to thank florian dubath and stefano foffa for many helpful discussions .
+ a special thank to angela haden , danielle chevalier , andreas malaspinas , marti ruiz - altaba and fabio dubath .
+ this work is partially supported by the swiss national fund ( fns ) .
|
experimental projects using spherical antennas to detect gravitational waves are nowdays a concrete reality .
the main purpose of this paper is to give a possible way of interpreting output data from such a system .
responses of the five fundamental quadrupole modes and of the six resonators in tiga collocations are shown as a function of the incoming direction of the incident wave .
then , for a source lying in the galactic plane , sidereal time and galactic longitude dependence is given .
thus , once a candidate source of gravitational waves is considered , we can exactly predict the resonators response as a function of time .
|
quantum key distribution ( qkd ) @xcite allows two remote parties to share an unconditional secret key , which has been proven in theory @xcite and demonstrated in experiment @xcite .
however , the imperfections of practical devices will compromise the security of qkd systems @xcite .
so far , three main approaches have been proposed to bridge the gap between theory and practice .
the first one is to close specific loopholes of devices with security patches @xcite , but it could not close potential and unnoticed loopholes .
the second one is device - independent ( di- ) qkd @xcite . by testing bells inequality in a loophole - free setting
, security could be obtained without detailed information about the implementation devices .
but di - qkd is impractical because an almost perfect single photon detector ( spd ) is required , and even so the secret key rate is limited @xcite .
the third approach is to remove as many device loopholes and assumptions as possible by either modifying the qkd protocol or refining the security proof .
one of the best results with this approach is measurement - device - independent ( mdi- ) qkd @xcite , which can remove all detector loopholes . since the detection system is widely regarded as the achilles heel of qkd @xcite , mdi - qkd is of great importance .
indeed , recently , mdi - qkd has been demonstrated both in the laboratory and in the field @xcite .
based on the framework of mdi - qkd , the source becomes the final battlefield for the legitimate parties and eve .
and the major flaw of the source is that a semiconductor laser diode ( s - ld ) , which generates a weak coherent state , is normally used as a single photon source in most commercial and research qkd systems @xcite .
the security of mdi - qkd as well as bb84 based on s - ld has been proven with decoy state @xcite . hence , it has been convinced that if the source can be well characterized ( for example the source flaws could be taken care of with the loss - tolerant qkd protocol @xcite ) , perfect security can still be obtained . generally speaking ,
there are two main classes of qkd protocols , one is discrete - variable ( dv- ) qkd ( including bb84 , decoy state bb84 , mdi - qkd , scarani - acin - ribordy - gisin ( sarg04 ) @xcite , differential phase shift ( dps ) @xcite , and so on ) , and the other one is continuous - variable ( cv- ) qkd @xcite . in most dv- quantum communication protocols ( e.g. , dv - qkd , quantum coin tossing ( qct ) @xcite , weak coherent state blind quantum computing ( bqc ) @xcite ) , the phase randomization is a cornerstone assumption . by assuming that the overall phase is uniformly distributed from 0 to @xmath0 ( in fact ,
discrete randomization with finite points , e.g. , 10 , is sufficient to guarantee qkd security @xcite ) , a coherent state with intensity @xmath1 is reduced into a classical mixture state , that is @xmath2 then it allows one to apply classical statistics theory to analyze quantum mechanics .
note that although the security of qkd with nonrandom phase had been proven @xcite , the performance is very limited in distance and key rate . in this paper , however , we demonstrate a simple quantum hacking strategy , with both a commercial and homemade pulsed laser based on s - ld , that allows eve to actively violate the phase randomization assumption , without leaving a trace afterwards .
thus it is effective for most of dv- quantum communication protocols .
our attack may also be effective for cv - qkd , since other parameters of the source ( e.g. , intensity ) could also be changed .
for example , it had been proven that the local oscillator fluctuation will compromise the security of cv - qkd @xcite . since s - lds
are widely used in most quantum information protocols ( e.g. , dv - qkd , cv - qkd , qct , bqc , and so on ) , and the security of these protocols is closely related to s - ld s parameters @xcite , our work constitutes an important step towards secure quantum information processing .
our attack differs from previous attacks @xcite .
first , in our attack , eve actively violate some basic assumptions required in the security proof by tampering with an initial perfect source .
second , unlike the laser damage attack @xcite in which eve also actively creates loopholes for a perfect spd , the loopholes created by our attack are temporary , this makes our attack impossible for alice and bob to detect during the off - time of the qkd system .
third , our attack also differs from the trojan - horse attack @xcite . in our attack , eve directly break some basic assumptions of qkd protocols , whereas in the trojan horse attack , back - reflected light is measured to analyze alice s information . and as the best we know , the trojan horse attack is invalid for alice with multi - lasers @xcite , but our attack remains applicable to such systems .
fourth and most importantly , our attack targets the source instead of spd .
this makes our attack a serious threat for most quantum information protocols ( not only qkd , but also qct and bqc ) .
here we emphasize that the phase randomization is a cornerstone assumption in the security of many quantum communication protocols including qkd , qct and bqc .
it is important for not only weak coherent pulse based protocols , but also , for instance , parametric down conversion based protocols @xcite .
and continuous or discrete phase randomization is also crucial for the loss - tolerant protocol @xcite .
in fact , without the phase randomization , the performance of a quantum communication protocol will be dramatically reduced in distance and key rate @xcite . however , we demonstrate experimentally in a clear manner how easy it is for eve to violate such a fundamental assumption in a practical setting .
thus our work is very generality for most of quantum information processing protocols .
it works for most dv - qkd , with various encoding schemes ( polarization , phase and time - bin ) and various kinds of lasers ( pulsed laser and continuous wave ( cw ) laser ) .
it is also possibly a serious threat for cv - qkd and other quantum information processing protocols ( such as qct and bqc ) .
the basic principle of our attack is as follows . in the inter - driven mode ,
the semiconductor medium of the s - ld is excited from loss to gain by each driving current pulse .
a laser pulse is generated from _
seed _ photons originating from spontaneous emission .
the phase of the laser pulse is determined by the seed photons .
since the phase of the seed photons is random , the phase of each laser pulse is random inherently @xcite .
however , if a certain number of photons are injected from an external source into the semiconductor medium , these photons will also be amplified to generate laser pulses .
consequently , the seed photons consist of two parts : one from spontaneous emission and the other part from the external source
. both parts will affect the phase of the resulting laser pulse .
if the injected photons greatly outnumber the photons from spontaneous emission , the phase of the output laser pulse is largely determined by the phase of the injected photons .
therefore , eve can control the phase of alice s signal laser by illuminating the s - ld from an external ` control source ' , and successfully violate the phase randomization assumption .
figure [ fig : scheme ] shows the schematic setup of our experiment .
we test four sample s - lds operating in inter - driven mode , two id300 pulsed lasers from idquantique @xcite ( numbers id300 - 1 and id300 - 2 ) , and two homemade pulsed lasers with s - lds from sunstar communication technology coltd ( model : sdlp55hmbifpn , numbers hm-1 and hm-2 ) . to measure the phase relationship between adjacent pulses ,
an unbalanced mach - zehnder interferometer is used ( see fig.[fig : scheme](b ) ) .
the repetition rate of the signal laser is set to be 206.34mhz to match the delay of the interferometer .
the output light is detected by a photodiode ( @xmath3 ) with a bandwidth of 1 ghz , and the voltage of each pulse is recorded using an oscilloscope with bandwidth 33 ghz and sample rate 80 ghz ( agilent , model : dsox93304q ) .
because the central frequency ( with a finite linewidth ) and polarization of the signal laser are unstable in experiment , eve needs to carefully modulate the frequency and polarization of her control laser to match her control laser with alice s signal laser . in our experiment , a tuning laser module ( model : 81600b-201 , agilent )
is used as eve s control laser .
furthermore , in fig.1 of the main text , we consider eve s control laser working at cw mode .
however , at the end of this paper , we consider the possibility that eve modulates her control laser into short photon pulses .
this can reduce alice s ability to detect eve s attack . in theory , the output voltage after @xmath3 is @xmath4/2 $ ] , where @xmath5 is the phase difference between adjacent pulses , and @xmath6 is the inherent phase difference between the two paths of the interferometer . by passively controlling the interferometer with temperature controller and vibration isolator , we can stabilize the interferometer within about 2 minute . in the test we set the number of pulses to be 25791 in each experimental point of fig.[fig : pro ] ( in each experimental point of fig.[fig : pro ] , we collect and store 10 m data . note that the repetition rate of the laser is 206.34 mhz and the sample rate of oscilloscope is 80ghz .
the number of data is about ( 1/206.34mhz)/(1/80ghz)@xmath7 388 in each pulse cycle .
thus the number of pulses is about @xmath8 . ) , and the time interval is about 0.125 ms ( @xmath9 , which is much lower than the time scale of the interferometer .
thus we could set @xmath10/2 $ ] for @xmath11 .
a uniform distribution of @xmath5 from 0 to @xmath0 will produce a u - type intensity distribution , owing to the fact that the mapping from phase to intensity is non - linear , @xmath12 .
indeed when eve is absent , the same distributions ( solid lines of fig.[fig : pro ] ) are obtained in experiments with both id-300 and the homemade pulsed laser . however , a bright light from eve could correlate the phase of each pulse and violate the phase randomization assumption ( dashed lines of fig.[fig : pro ] ) . in fact ,
when photons are injected into alice s signal laser , the intensity distribution of @xmath13 for both id300 and the homemade signal laser becomes gaussian .
consequently , various quantum hacking strategies can be applied to spy on the final key @xcite .
figure [ fig : attack](a ) shows a schematic setup to attack a complete qkd system . theoretically speaking , eve can perfectly control the phase of alice s source , and then the intensity distribution should be a sharp line .
however , owing to the following two main reasons , the measured intensity distribution in fig.[fig : pro ] of the main text follows gaussian distribution : ( 1 ) there exists phase noise in eve s controlling laser , which follows gaussian distribution .
the measured intensity is the interference of adjacent pulses ( the interval of adjacent pulses is about 5ns ) , thus the experimental results depends on the phase noise of eve s control laser at different time .
( 2 ) the interference is imperfect , including the loss of two paths of the interferometer , the time jitter of optical pulse , and so on .
therefore , a practical eve ca nt perfectly control the phase of alice s source , and the phase noise decides how much information will be leaked to eve .
furthermore , although the security of the bb84 protocol had been proven based on uniformed random phase from 0 to @xmath0 @xcite and nonrandom phase @xcite , but the key rate ( or mutual information between alice and eve ) is still unknown , if the phase of source follows gaussian distribution or a general probability distribution , which will be studied in future .
furthermore , we note that when the ld is operated in inter driven mode , the emitted pulses have random phase , and such phase noise had been used as a quantum random number generator by many groups @xcite .
however , fig.[fig : pro](e ) of the main text does not prove that the phase of each pulse follows uniform distribution from @xmath14 to @xmath0 .
in fact , if the phase is uniformly distributed from 0 to @xmath15 , the same probability distribution could also be obtained .
thus , the phase randomization assumption must be carefully evaluated , particularly for a high - speed qkd system @xcite .
active phase randomization @xcite is a good countermeasure to guarantee the phase randomization assumption .
figure [ fig : attack](b ) shows a possible countermeasure for alice to monitor our attack .
it includes three main devices , an isolator ( iso . ) , a filter and a photodetector .
but these devices could not defeat our attack completely , if they are not carefully configured ( see appendix [ appendix_a ] for details ) .
( 1 ) the isolator could not entirely stop eve s photons due to its finite isolation ( see fig.[fig : pro_isolator ] ) , and other imperfections of practical isolators have been found in a recent paper @xcite . ( 2 ) since the wavelength of eve s control laser is the same as that of alice s signal laser in our attack , an optical frequency filter is also ineffective .
( 3 ) both optical power meter and classical photodetector could be foiled by eve , so that they could not accurately show the power of light from the channel .
for example , a short pulse light might reduce the average power of eve s light , and the finite bandwidth of these monitor devices might worsen the monitoring results .
furthermore , a recent paper also shows other imperfections of a practical monitoring photodetector @xcite .
an active phase randomization ( fig.[fig : attack](c ) ) @xcite , or the cw laser followed by an external intensity modulator and an active phase randomization scheme , is another important choice for practical qkd systems , especially when the qkd system works in a high repetition rate @xcite
. then phase randomization assumption is automatically guaranteed .
but such a countermeasure may not remove our attack entirely , since eve can tamper with other parameters ( e.g. , intensity and shape , see fig.[fig : waveform ] ) to compromise the security of such systems . for example , the key rate of both cv - qkd and dv - qkd depends on the intensity of the signal pules @xcite .
but the stability of s - ld ( no matter whether it works on pulsed mode or cw mode ) could be damaged by bright light , so that the intensity of alice s laser is unstable . therefore , in this sense , our attack is also effective for the qkd system with a cw laser and an active phase randomization scheme .
another countermeasure is to use a protocol ( or security proof ) with an unrandom phase , but the performance of such a protocol is dramatically reduced in distance and key rate@xcite .
fig.[fig : waveform ] shows that the pulse shape would also be changed by eve s bright light .
these changed parameters are also helpful . for example , the signal pulse is emitted earlier than that without eve @xcite , and the time shift is different for each s - lds . furthermore , in the absence of an external field , the first oscillation is much stronger than the following oscillation , and a few oscillations appear @xcite .
but when eve is present , more oscillations are observed , and different laser diodes have different oscillation waveform .
thus it is possible for eve to compromise the security of qkd systems with multi - lasers @xcite by measuring the characters of signal pules ( e.g. , time - shift , pulse width , optical frequency ) .
here we remark that , generally speaking , the changes of pulse shape are helpful for both eve and alice . although more imperfection could be exploited by eve , more parameters could be monitored by alice to discover the existence of eve .
in fact , both eve and alice must be very careful in the cat - and - mouse game ( see appendix [ appendix_b ] for details ) .
first , if alice wants to completely monitor the changes of pulse shape , some advanced devices with high speed and bandwidth are required , which may dramatically increase the technology challenge and cost of a practical alice .
second , eve could carefully configure her attack to ensure that her attack could not increase the error rate and the changes of pulse shape could not be discovered by alice . third , generally speaking , the changed shape may actually benefit eve more than alice and bob .
this is because eve could well be a spy or national security agency such as the nsa and so eve has a much larger power and budget than alice and bob .
thus eve is probably at a better position to exploit the imperfections that she has introduced in the quantum signal .
furthermore , note that even a tiny violation of the phase randomization assumption or other parameters of the source will undermine the very foundation of security proofs in qkd and it will no longer be fair for alice and bob to claim unconditional security . finally , in addition to using a laser , eve can also attack the qkd system by using temperature , microwave radiation , and so on . at the same time , although most quantum hackers focus on the optical devices of the legitimate parties , eve can also exploit imperfections in the electrical devices of the qkd system .
for example , if the electromagnetic shielding of devices of alice and bob is imperfect , eve could use microwave radiation from outside to control the parameters of these devices .
these are the subjects of future investigations .
in summary , phase randomization is a cornerstone assumption for many quantum communication protocols , and a tiny violation of such an assumption is fatal to the security of such protocols .
however , here we demonstrate experimentally , with both commercial and homemade pulsed lasers , how easy it is for eve to violate such a fundamental assumption in a practical setting . additionally , besides the random phase , other parameters ( e.g. , intensity ) of the source could also be changed .
our attack works for most dv - qkd protocols , and possibly for cv - qkd and other quantum information processing protocols ( e.g. , qct and bqc ) .
thus our work constitutes an important step towards secure quantum information processing .
we thank z. yuan and v. makarov for helpful discussions .
this work is supported by the national natural science foundation of china , grant no .
l.m.l is supported by the ncet program .
lo is supported by nserc .
f. xu is supported by the office of naval research ( onr ) and the air force office of scientific research ( afosr ) .
now we show that alice s countermeasure , shown in fig.[fig : attack](b ) , of the main text ( include an isolator , an optical filter , and a photodetector ) , ca nt remove our attack entirely . _
( i ) isolator- _ in general , an optical isolator serves to prevent back - reflected photons from returning to alice s lab .
however , owing to the finite isolation of practical isolators , this approach only reduces the probability that photons infuse into alice s zone , but can not eliminate this probability entirely .
we perform a proof - of - principle experiment , by inserting a 25db isolator after the output port of the signal laser id300 - 1 .
the experimental results of fig.[fig : pro_isolator ] of the main text show that the intensity distribution is still gaussian - type but not u - type when eve uses a cw laser with a power of @xmath16 .
thus the phase of adjacent pulses can be still correlated .
although isolation of some commercial isolators reaches 50db ( or alice can use two or more isolators in series to increase the isolation ) , it can not totally foil our attack , because eve can always increase the power of her control laser .
furthermore , other imperfections of the practical isolator have been found in a recent paper @xcite . _
( ii ) filter- _ an optical frequency filter is often used by alice to remove any wavelength - dependent flaws . by doing so
, only the light within a narrow band of frequencies can enter alice s lab .
however , the wavelength of eve s control laser is the same as that of alice s signal laser in our attack .
thus , an optical frequency filter is not an effective countermeasure against our attack . _
( iii ) photodetector- _ alice can use both an optical power meter and photodetector to monitor the intensity of light from a quantum channel , but the optical power meter measures the average power of light .
thus it could be foiled by eve who uses a pulsed laser .
for example , fig.4 of the main text shows that a cw laser with an optical power @xmath16 is sufficient to correlate the phase of alice s signal pulse . now , suppose that the repetition rate of the qkd system is 10mhz , and eve uses a pulsed control laser with width of 100ps .
then the duty circle of eve s pulse is 100ps/10ns=0.001 .
thus the average optical power is reduced to @xmath17 .
a classical photodetector with a discrimination voltage can be used to monitor the intensity of pulsed light .
however , the classical photodetector could also be cheated due to the following two reasons . first , the classical photodetector can be damaged by bright light so that it may not work as expected .
there are two kinds of classical photodetectors : one based on the pin , and the other one based on the apd .
both can be damaged by bright light @xcite .
for example , the detector based on ingaas - apd from thorlabs has a maximal input power 10mw ( model : apd310 ) and 1mw ( model : apd110c ) .
the maximal input power for the detector based on ingaas - pin from thorlabs ( model : pda8gs ) is about 1mw for cw and 20mw for 60ms @xcite .
second , the finite bandwidth of the classical photodetector may worsen the monitoring results .
we experimentally measure the amplitude of an electrical signal using an oscilloscope with various bandwidth ( fig.[fig : pulsewidth](a ) ) .
furthermore , the theoretical amplitudes of an ideal gaussian pulse which passes a linear time - invariant ideal low - pass filter are also shown in fig.[fig : pulsewidth](b)-(c ) . generally speaking ,
when a signal pulse , @xmath18 , passes a linear time - invariant device , its amplitude function becomes @xmath19e^{i\omega t}d\omega,\ ] ] where @xmath20 $ ] is the fourier transformation , and @xmath21 is the frequency response function of device .
it clearly shows that devices with finite bandwidth will filter high - frequency signals , and reduce the amplitude of a signal pulse . for
simply , we assume that the signal is gaussian pulse and the device is an ideal low - pass filter , that is , @xmath22,\\ g(\omega)&=\begin{cases } 1 & |\omega|\leq\omega_0\\ 0 & |\omega| > \omega_0\end{cases}. \end{split}\ ] ] here @xmath23 is the standard deviation of a signal pulse @xmath24 .
if the 3db width of @xmath24 is noted as @xmath25 , it is easy to check that @xmath26 .
@xmath27 is the maximal bandwidth of the ideal low - pass filter .
the theoretical amplitude of @xmath28 is shown in fig.5(b)-(c ) of the main text .
the results clearly show that monitoring devices with finite bandwidth could not faithfully characterize the factual amplitude of the input signal , and eve could foil the monitoring devices with a sharp pulsed signal .
although the test of fig.[fig : pulsewidth ] is performed for an electrical signal , the results can be directly applied to the photodetector with finite bandwidth .
for example , suppose that the gain and discrimination voltage of the photodetector are @xmath29 v / w and 0.2 v , and that eve uses a pulsed control light with a 3db width of 100ps and a peak power of @xmath30 .
then the expected output voltage of the photodetector should be 1v , which is much larger than the discrimination voltage , 0.2v .
fig.[fig : waveform ] of the main text also shows that if the bandwidth of alice s photodetector is high enough ( e.g. , @xmath31 ) , eve can be discovered .
( note that generally speaking , the gain of photodetector will be decreased when the bandwidth is increased .
but here we simply assume the gain is independent of the bandwidth .
) however , if the bandwidth of photodetector is limited ( e.g. , 1ghz ) , the factual output voltage is lower than the discrimination voltage , 0.2v .
alice can not discover the existence of eve .
note that fig.[fig : pro ] of the main text has shown that @xmath30 is sufficient for eve to break the phase randomization assumption .
furthermore , a recent paper also shows other imperfections in a practical monitoring photodetector @xcite .
therefore , the possible countermeasure of fig.[fig : attack](b ) of the main text could be cheated by eve , if the devices are not carefully configured .
furthermore , illumination by a bright light changes not only the phase but also the pulse waveform , including its width , amplitude , and shape . although we still do not know how eve can obtain more information by exploiting such a modified waveform , it remains possible for eve to attack the qkd system .
fig.[fig : waveform ] of the main text clearly shows that when the signal laser is illuminated by bright light , the pulse shape would also be changed . generally speaking , the additional changes are helpful for both eve and the legitimate parties .
more imperfections can be exploited by eve to spy the final key , and more parameters can be monitored by alice to discover the existence of eve .
but it is still possible for eve to perform our attack . theoretically speaking
, eve could perform a suitable attack to ensure that the modification of the pulse shape would not increase the error rate between alice and bob .
in fact , eve can perform the intercept - and - resend attack , and ensure that the error rate is lower than a reasonable value .
for example , in the system with multi - laser diodes , she first measures the time - shift of each laser diode to determine alice s state .
then she can resend a faked state to bob according to her measurement results . in this case , if the time - shift is distinguishable for each laser diode ( it is possible according to fig.[fig : waveform ] ) , eve could know the state sent by alice
. then she can resend a perfect faked state to bob according to her measurement .
thus no additional error will be introduced , and the legitimate parties could not discover the existence of eve by monitoring the error rate . for alice , she may discover the existence of eve by carefully monitoring the parameters of the signal laser .
but since the change is tiny in some parameters , some advanced devices with high speed and bandwidth ( e.g. , photodetectors , analog - digital convertors , or time - amplitude convertors , and so on ) are required for alice , which may dramatically increase the technology challenge and cost of a practical alice .
for example , the time - shift for id300 lasers is about 100ps ; thus if alice wanted to characterize the time - shift of her pulses , the bandwidth and sample rate of alice s analog - digital convertor should be larger than 40ghz ( generally speaking , at least four points are needed to recover a pulse ) .
furthermore , the bandwidth and sample rate should be increased for homemade lasers ( see fig.[fig : waveform ] of the main text for hm-1 and hm-2 ) , since much smaller changes are introduced . for eve s part
, she should carefully configure her attack to foil alice s monitor devices . ( 1 ) eve may carefully stable her controlling laser , and match the optical frequency of her controlling laser with that of alice s signal laser , so that , excepting the random phase , many tiny changes will be introduced on the pulse shape . taking the homemade lasers ( hm-1 and hm-2 ) as an example , eve s
light will correlate the phase of each of the pulses ( see fig.[fig : pro ] ( c ) and ( d ) of the main text ) , but fig.[fig : waveform ] ( c ) and ( d ) of the main text show that the changes of pulse shape are very tiny ( at least , compared with id300 - 1 and id300 - 2 , we do not find any obvious changes in the pulse shape using a photodetector with 40ghz bandwidth , an oscilloscope with 33ghz bandwidth and a sample rate of 80ghz , thus if alice wants to discover the changed shape of hm-1 and hm-2 , advanced devices with higher bandwidth and sample rate are required .
( 2 ) eve may reduce the risk of being discovered by spying parts ( not all ) of final key .
for example , it has been proven that a small fluctuation of intensity will dramatically reduce the secret key rate of decoy state bb84 protocol @xcite .
thus she still could obtain parts of final key by trivially changing the intensity of alice s signal laser .
in fact , it had been shown that , if the intensity of alice s signal pulses fluctuates 1% , 2% and 3% , the final key rate will be reduced by 11.86% , 23.91% and 36.17% @xcite ( the simulation was performed based on the experimental parameters of ref.@xcite(b ) ) . furthermore ,
generally speaking , the ability for eve to change other parameters in an optical signal may actually benefit eve more than alice and bob .
this is because eve could well be a spy or work for a national security agency such as the nsa , and so eve has a much larger budget than alice and bob , and thus is probably in a better position to exploit the imperfections that she has introduced in the quantum signal .
s. wang , w. chen , j. f. guo , z. q. yin , h. w. li , z. zhou , g. c. guo , and z. f. han , _ opt .
lett . _ * 37 * , 1008 - 1010 ( 2012 ) ; y. liu , t. y. chen , j. wang , w. q. cai , x. wan , l. k. chen , j. h. wang , s. b. liu , h. liang , l. yang , _ et al .
express _ * 18 * , 8587 - 8594 ( 2010 ) ; z. l. yuan , a. r. dixon , j. f. dynes , a. w. sharpe , and a. j. shields , _ appl .
lett . _ * 92 * , 201104 ( 2008 ) .
z. l. yuan , j. f. dynes , and a. j. shields .
_ , * 4 * , 800 ( 2010 ) ; t. f. da silva , g. b. xavier , g. p. termorao , and j. p. von der weid . _
express _ , * 20 * , 18911 ( 2012 ) ; t. f. da silva , g. c. do amaral , g. b. xavier , g. p. temporao , and j. p. von der weid .
_ ieee journal of selected topics in quant .
* 21 * , issue . * 3 * , 6600309 ( 2015 ) ; c. c. w. lim , n. walenta , m. legr , n. gisin , and h. zbinden . _ ibid _ , vol . * 21 * , issue . * 3 * , 6601305 ( 2015 ) .
a. rubenok , j. a. slater , p. chan , i. lucio - martinez , w. tittel .
_ , * 111 * , 130501 ( 2013 ) ; y. liu , t. y. chen , l. j. wang , h. liang , g. l. shentu , j. wang , k. cui , h. l. yin , n. l. liu , l. li , _ et al .
_ _ ibid . _
* 111 * , 130502 ( 2013 ) ; z. y. tang , z. f. liao , f. h. xu , b. qi , l. qian , and h. k. lo .
* 112 * , 190503 ( 2014 ) ; y. l. tang , h. l. yin , s. j. chen , y. liu , w. j. zhang , x. jiang , l. zhang , j. wang , l. x. you , j. y. guan , _ et al .
_ _ ibid . _ * 113 * , 190501 ( 2014 ) ; t. ferreira da silva , d. vitoreti , g. b. xavier , g. c. do amaral , g. p. temporao , j. p. von der weid .
a _ , 88 , 052303 ( 2013 ) .
t. schmitt - manderbach , h. weier , m. frst , r. ursin , f. tiefenbacher , t .
scheidl , j. perdigues , z. sodnik , c. kurtsiefer , j. g. rarity , _ et al .
_ , _ phy .
_ , * 98 * , 010504 ( 2007 ) .
c. z. peng , j. zhang , d. yang , w. b. gao , h. x. ma , h. y , et al . , _ ibid .
_ , * 98 * , 010505 ( 2007 ) .
s. h. sun , m. gao , m. s. jiang , c. y. li , and l. m. liang , _ phys .
a _ , * 85 * , 032304 ( 2012 ) ; y. l. tang , h. l. yin , x. f. ma , c. h. f. fung , y. liu , h. l. yong , _ et al .
_ _ ibid . _ * 88 * , 022308 ( 2013 ) .
the time shift appears because when the laser cavity is seeded with an external field , the relaxation oscillations are dampened and the first oscillation occurs earlier . and
the amount of the time shift is much larger for the two id300 lasers than those of the homemade lasers , because the linewidth of two id300 lasers is larger than that of the homemade lasers .
thus the probability that a photon is infused into the id300 lasers is larger than that of the homemade lasers .
the reason is that it takes time for the initial field to appear ( owing to the finite spontaneous emission rate and geometry of the laser chip ) .
while the field takes time to build up , there is no or little stimulated emission , yet the electrical pumping continues at the full rate .
the population inversion then rises above the steady - state value and far overshoots it .
then the field far overshoots the steady - state , because for a short time the laser has much higher gain , then the stronger field depletes it by stimulated emission below the steady - state value , and so on , there are a few oscillations before the emission settles to a stable value .
but in the presence of the seed field in the cavity , the population inversion does not initially overshoot as high , because the emission stimulated by the seed field begins earlier .
|
the security of source has become an increasingly important issue in quantum cryptography . based on the framework of measurement - device - independent quantum - key - distribution ( mdi - qkd ) ,
the source becomes the only region exploitable by a potential eavesdropper ( eve ) .
phase randomization is a cornerstone assumption in most discrete - variable ( dv- ) quantum communication protocols ( e.g. , qkd , quantum coin tossing , weak coherent state blind quantum computing , and so on ) , and the violation of such an assumption is thus fatal to the security of those protocols . in this paper , we show a simple quantum hacking strategy , with commercial and homemade pulsed lasers , by eve that allows her to actively tamper with the source and violate such an assumption , without leaving a trace afterwards . furthermore , our attack may also be valid for continuous - variable ( cv- ) qkd , which is another main class of qkd protocol , since , excepting the phase random assumption , other parameters ( e.g. , intensity ) could also be changed , which directly determine the security of cv - qkd .
|
in this appendix , we begin by discussing the transformation of the driven bhm from the lab frame to the rotating frame , before doing a bogoliubov stability analysis .
the rotating frame is defined by the unitary transformation @xmath88\cdot \sum_j { \bf r}_j n_j \right),\ \ \ \ \
b_j\to b_j e^{-i\mathcal{a}(t)\cdot{\bf r}_j},\end{aligned}\ ] ] where @xmath11
. this time - dependent change of basis is equivalent to the standard gauge transformation in electromagnetism @xmath89 .
physically , the transformation trades the fast time - dependence of the quasimomentum for a time oscillating dispersion relation .
the infinite - frequency limit is non - trivial when the amplitude of the gauge potential @xmath90 remains finite .
this rotation facilitates the analytic computation the time - averaged floquet hamiltonian .
in fact , going to the rotating frame is equivalent to re - summing an infinite inverse - frequency subseries in the lab - frame@xcite . as a result ,
the effective hopping matrix element is a non - perturbative function of @xmath29 .
the rotating frame hamiltonian reads @xmath91 where in the second equality we separated out the time average explicitly .
the time - average hamiltonian @xmath92 is defined in eq .
( 3 ) of the main text , and the effective hopping is @xmath93 . to avoid confusion in the notation
, we note in passing that this effective hamiltonian is _ not _ equivalent to the full floquet hamiltonian at any finite driving frequency , see eg
. refs .
@xcite .
before we dive into the details of the parametric stability analysis for the driven bose - hubbard model , let us demonstrate how to derive the stability criterion with the help of the rotating wave approximation ( rwa ) [ a very similar method was used to study the parametric instability in periodically - driven luttinger liquids @xcite ] . for this purpose
, we choose the parametric oscillator with hamiltonian @xmath94 .
writing the hamiltonian using ladder operators @xmath95 and @xmath96 , and dropping any ( time - dependent ) constants leads to @xmath97 if we parametrise @xmath98 , with @xmath99 and @xmath100 , we can write heisenberg s eom as @xmath101 \left ( \begin{array}{c } u ' \\ v ' \end{array } \right ) + \frac{\alpha}{2}\omega_0\cos\omega t \left ( \begin{array}{cc } 0 & 1 \\ -1 & 0 \end{array } \right ) \left ( \begin{array}{c } u ' \\ v ' \end{array } \right),\end{aligned}\ ] ] where the matrix @xmath102 has zero time - average and @xmath103 is the pauli matrix in bogoliubov space .
we now apply the transformation @xmath104 , @xmath105 which brings the eom into the form @xmath106 \left ( \begin{array}{c } \tilde u ' \\ \tilde v ' \end{array } \right).\end{aligned}\ ] ] so far the treatment of the parametric oscillator eom has been exact .
however , the present form of the equation allows to easily identify the terms responsible for parametric resonance . to this end
, we apply the rotating wave approximation ( rwa ) ( i ) keeping in mind that the time - average of @xmath107 vanishes identically , and ( ii ) dropping any counter - rotating terms .
thus , we find that the dominant contribution to the dynamics appears for @xmath108 , which sets the critical driving frequency on resonance .
the resulting effective rwa - eom assumes the simple form : @xmath109 diagonalising the matrix on the right - hand side , we find the two lyapunov exponents @xmath110 .
hence , the maximum instability growth rate is set by @xmath111 . the stability criterion and the instability growth rate _ on resonance _ derived above by means of the rwa
agree precisely with the standard results obtained using two - time perturbation theory or by other means @xcite . in the following ,
we apply the same method and show explicitly the stability analysis for the driven bhm leading to the resonance condition @xmath112 of the main text .
had it not been for the subindex @xmath113 , this result follows immediately from the above arguments for the parametric oscillator .
we work in bogoliubov theory which , as we show , captures the onset of instability .
we begin by writing the rotating frame hamiltonian of eq . in momentum space and apply to it the bogoliubov approximation .
parametrising the bosonic annihilation operator by @xmath114 , with @xmath115 and @xmath116 , the heisenberg equations of motion read @xmath117 \end{array } \right ) \left ( \begin{array}{c } u_{\bf k } \\
v_{\bf k } \end{array } \right)\nonumber\\ & = & \left ( \begin{array}{cc } \varepsilon_\text{ave}({\bf k } ) + zj_0 + u n_0 & u n_0 \\
-u n_0 & -(\varepsilon_\text{ave}({\bf k } ) + zj_0 + u n_0 ) \end{array } \right ) \left ( \begin{array}{c } u_{\bf k } \\
v_{\bf k } \end{array } \right ) + \left ( \begin{array}{cc } g_{\bf k}(t ) & 0
\\ 0 & -g_{\bf -k}(t ) \end{array } \right ) \left ( \begin{array}{c } u_{\bf k } \\
v_{\bf k } \end{array } \right ) , \end{aligned}\ ] ] where in the second line we separated the time average . here
@xmath70 is the condensate fraction [ of the time - averaged hamiltonian ] , and the periodic function @xmath118 .
it will prove convenient to first perform a static bogoliubov transformation @xmath119 , which diagonalises the time - averaged hamiltonian : @xmath120 the bogoliubov angle is defined via @xmath121 and @xmath122 with @xmath123 the corresponding bogoliubov dispersion ( see main text ) .
the pseudounitary operator @xmath119 has the property @xmath124 [ note that @xmath125 . the idea behind performing this bogoliubov transformation is to bring the time - averaged hamiltonian in diagonal form . at high - frequencies , the former represents the leading - order floquet hamiltonian [ when expanded in powers of the inverse frequency ] , and thus this bogoliubov transformation brings the state at time @xmath126 to a basis which is close but not equal to the exact floquet basis [ finite @xmath85-corrections are missing ] : @xmath127\sinh(2\theta_{\bf k } ) \left ( \begin{array}{cc } 0 & 1 \\ -1 & 0 \end{array } \right ) \left ( \begin{array}{c } u'_{\bf k } \\
v'_{\bf k } \end{array } \right).\end{aligned}\ ] ] introducing the short - hand notation @xmath128,\end{aligned}\ ] ] the eom readily assumes the form : @xmath129 \left ( \begin{array}{c } u'_{\bf k } \\
v'_{\bf k } \end{array } \right ) + \sinh(2\theta_{\bf k } ) h_{\bf k}(t ) \left ( \begin{array}{cc } 0 & 1 \\ -1 & 0 \end{array } \right ) \left ( \begin{array}{c } u'_{\bf k } \\
v'_{\bf k } \end{array } \right).\end{aligned}\ ] ] notice that the diagonal matrix @xmath130 has a zero time average , a property inherited from the function @xmath131 . in the non - interacting limit , where one can integrate the eom exactly
, we have @xmath132 and the time - dependent term @xmath130 results in a trivial dynamical phase whose origin can be traced back to the kick operator of floquet theory@xcite .
thus , we find that at small @xmath7 , the kick operator attains additional contributions due to the bogoliubov dressing in the diagonal matrix @xmath130 . although parametric instability is a phenomenon believed to originate in the micromotion itself [ the stroboscopic floquet hamiltonian , if it exists as a local operator , is a hermitian operator and therefore has real eigenvalues ] , in both the weakly - interacting and the non - interacting case this @xmath130-term has no significance for the onset of the parametric instability , as we show below .
we stress that @xmath130 is not the only contribution to the micromotion ; part of the latter is due to the function @xmath133 .
thus , special attention must be paid to the term proportional to @xmath134 . in the instantaneous diagonal basis , this results to the coupling of the form @xmath135 , and is responsible for the drive - assisted scattering of bosons out of the condensate mentioned in the main text . in order to make the nature of parametric resonance visible
, we perform yet another unitary time - dependent transformation @xmath136 , @xmath137 .
this transformation is needed to bring the relative dynamical phases of @xmath138 and @xmath139 with energy @xmath123 at the same footing . in some physical sense , the onset of the parametric instability for a quantum harmonic oscillator is due to an enhanced mismatch of the relative dynamical phases accumulated by the operators @xmath140 and @xmath141 in the presence of the periodic drive .
the eom now takes the form @xmath142 \left ( \begin{array}{c } \tilde u'_{\bf k } \\
\tilde v'_{\bf k } \end{array } \right).\end{aligned}\ ] ] so far the above analysis was rather involved but exact .
we can make further progress by flourier - expanding the periodic function @xmath143 [ note that @xmath131 has zero mean by definition ] which leads to @xmath144e^{il\omega t}$ ] .
the fourier coefficients @xmath145 are closely related to bessel functions .
now we apply the rotating wave approximation ( rwa ) to the above equation .
this brings about a pronounced dominant contribution from the drive for @xmath112 coming from the slowly rotating oscillatory off - diagonal terms .
this behaviour is in precise agreement with the parametric resonance condition from the main text which was verified numerically using the wcca .
recalling that the time - average of @xmath131 vanishes , the dynamics of the system on resonance is governed by the following set of effective equations @xmath146 \\
k}(\zeta)+ c^{1}_{\bf -k}(\zeta)\right ] & e_\text{ave}({\bf k } ) \end{array } \right).\end{aligned}\ ] ] diagonalising this matrix , we find the lyapunov exponents on resonance [ we avoid the terminology quasienergies since unlike quasienergies the lyapunov exponents can be complex numbers ] , @xmath147\left[c^{+1}_{\bf k}(\zeta)+ c^{+1}_{\bf -k}(\zeta)\right ] } , \ ] ] whose imaginary part is responsible for the exponential growth of the parametrically - unstable solution within bogoliubov theory . notice that @xmath148 defines the instability growth rate which is directly tied to the heating rate at short times in the parametrically unstable regime . at small driving amplitudes ,
the imaginary part of the lyapunov exponent scales linearly with the drive strength @xmath149 , as expected .
higher - order photon absorption resonances can be taken into account by using the higher - order fourier coefficients @xmath150 .
the behaviour of the system in the vicinity of the parametric resonance , on the other hand , can be analysed by introducing a small detuning @xmath151 .
finally , we remark that , the above analysis within bogoliubov theory is not expected to produce the correct dynamics in the unstable regimes due to the lack of particle number conservation .
instead , one needs to further develop the perturbation theory by extending it to the wcca .
figure [ fig : ph_diag_resolved ] shows a direct comparison between the stability criterion for the transition boundary @xmath152 , shown as a solid black line , and the numerical solution to the wcca equations .
numerically , the condensate is defined to be stable whenever , after @xmath153 periods of stroboscopic evolution , no decay is visible . in general , we find an excellent agreement between the two approaches ; the only difference comes for the points lying right on the transition boundary which appear to be stable for small @xmath29 and unstable for large @xmath29 .
we attribute this to the uncertainty in determining the numerical phase boundary : indeed , coming from the stable phase the lifetime of the condensate can be very long , thus exceeding the @xmath153 periods of evolution time .
in this appendix , we derive the equations of motion ( eom ) using the weak - coupling conserving approximation ( wcca ) .
we are interested in studying the periodically driven bose - hubbard model on a 2d lattice : @xmath154 in order to treat the spontaneous symmetry breaking of the condensate efficiently , we introduce the bogoliubov spinor for the bosonic fields @xmath155 , with @xmath156 , where @xmath157 and @xmath158 . adopting the notation @xmath159 , the time - dependent action can be cast into the compact form @xmath160 & = & s_0 + s_\text{int}\nonumber\\ s_0 [ b , b^ * ] & = & \frac{1}{2}\int_c\mathrm{d}x b^*_{a}(x)\left(g_\text{free}^{(-1)}\right)_{ab}(x , y ) b_{b}(y)\nonumber\\ s_\text{int}[b , b^ * ] & = & -\frac{u}{2}\int_c\mathrm{d}x\mathrm{d}y \delta_c(x - y ) b^*(x ) b^*(x ) b(x ) b(x).\end{aligned}\ ] ] where the integral over time is taken along the keldysh roundtrip contour @xmath161@xcite and we introduced the delta function @xmath162 . in bogoliubov space , the noninteracting green s function has thus the form @xmath163 we define the vacuum expectation value ( vev ) @xmath164 and the quasiparticle ( phonon ) propagator @xmath165 as @xmath166 the microscopically occupied fields are denoted with a tilde @xmath167 .
hence the green s function @xmath168 defined above does _ not _ include the condensate fraction .
the effective action is given by the double legendre transform of the original action w.r.t .
the vev @xmath164 and the correlator @xmath169@xcite : @xmath170 & = & s[\varphi,\varphi^ * ] + \frac{1}{2}\text{tr}[\log g^{-1 } ] + \frac{1}{2}\text{tr}[g_0^{-1}(\varphi)g ] - \gamma_2[\varphi , g],\nonumber\\ s[\varphi,\varphi^ * ] & = & \int\mathrm{d}x\mathrm{d}y \varphi^*(x)g^{-1}_\text{free}(x , y)\varphi(y ) - \frac{u}{2}\int\mathrm{d}x\left|\varphi(x)\right|^4 \equiv \int\mathrm{d}x\mathrm{d}y \varphi^*(x)g^{-1}_0(x , y;\varphi)\varphi(y),\end{aligned}\ ] ] where the sum over the bogoliubov - nambu index @xmath171 is implicit .
the bogoliubov propagator @xmath172 generates the motion of the gross - pitaevskii equation . notice that it depends on the field @xmath173 itself since the gpe is nonlinear . from
that we obtain the inverse bogoliubov propagator @xmath174 via : @xmath175}{\delta \varphi_{a}^*(x)\delta \varphi_{b}(y ) } = \frac{1}{2}\left(g_\text{free}^{-1}\right)_{ab}(x , y ) - \frac{u}{2}\delta_c(x - y)\left(\begin{array}{cc } 2|\varphi(x)|^2 & \varphi(x)^2\\ ( \varphi(x)^*)^2 & 2|\varphi(x)|^2 \end{array}\right)_{ab}.\nonumber\\\end{aligned}\ ] ] so far the calculation is exact , however , we have not specified the luttinger - ward functional @xmath176 $ ] yet which is the sum of all two - particle irreducible diagrams and thus has to be treated approximately . here
, we consider a weak - coupling expansion which amounts to consider diagrams to first order in @xmath7 , see fig .
[ fig : gamma2_nonc ] .
the eom for the vev and the propagator are obtained by making the effective action @xmath177 stationary with respect to the fields , @xmath178}{\delta\varphi^*}=0 $ ] and @xmath178}{\delta g_{ab}}=0 $ ] , which lead to : @xmath179g_{bc}(t , t ' ) & = & \delta_{ac}\delta_c(t - t ' ) , \label{eq : eom_phi_nonc}\end{aligned}\ ] ] and the green s function multiplication in the second equation above is understood in the matrix - multiplication sense : @xmath180 .
we remark that these eom are equivalent to the bogoliubov - hartree - fock eom derived in ref .
when starting from the lab frame hamiltonian ( see main text ) , making the ansatz @xmath181 , @xmath182 , and then linearises any cubic terms in @xmath183 .
next , we open the closed time contour@xcite by decomposing the green s function into a spectral part @xmath184 and a statistical part @xmath185 according to @xmath186\rangle_c = i\left(\begin{array}{cc } \langle[\tilde b(x),\tilde b^\dagger(x')]\rangle & \langle[\tilde b(x),\tilde b(x')]\rangle \\ \langle[\tilde b^\dagger(x),\tilde b^\dagger(x')]\rangle & \langle[\tilde b^\dagger(x),\tilde b(x')]\rangle .
\end{array}\right)_{ab}\end{aligned}\ ] ] the following relations follow immediately from the above definitions : @xmath187 we now assume that the system is translationally invariant , with periodic boundary conditions .
we find the following system of coupled nonlinear eom in momentum space for the condensate @xmath188^*\left[\varphi(t)\right]^2 + 2\varphi(t ) \int_q f_{11}(t , t;q ) + \left[\varphi(t)\right]^*\int_q f_{12}(t , t;q ) \right],\nonumber\\ \label{eq : wcca_gpe}\end{aligned}\ ] ] and the statistical correlator @xmath189 @xmath190 ^ 2 + \int_q f_{12}(t , t;q)\right ) \left[f_{12}(t , t';k)\right]^ * \right],\nonumber\\ i\partial_tf_{12}(t , t';k ) & = & ( zj_0-\mu)f_{12}(t , t';k ) + \varepsilon_k(t)f_{12}(t , t';k)\nonumber\\ & & + \frac{u}{n_s}\left [ 2\left ( |\varphi(t)|^2 + \int_q f_{11}(t , t;q ) \right)f_{12}(t , t';k ) + \left ( \left[\varphi(t)\right]^2 + \int_q f_{12}(t , t;q)\right)\left[f_{11}(t , t';k)\right]^ * \right].\nonumber\\\end{aligned}\ ] ] for completeness , we also give the equations of motion for the spectral correlators @xmath191 which , on the other hand , obey @xmath192 ^ 2 + \int_q f_{12}(t ,
t;q)\right)\left[\rho_{12}(t , t';k)\right]^ * \right],\nonumber\\ i\partial_t\rho_{12}(t , t';k ) & = & ( zj_0-\mu)\rho_{12}(t , t';k ) + \varepsilon_k(t)\rho_{12}(t , t';k)\nonumber\\ & & + \frac{u}{n_s}\left [ 2\left ( |\varphi(t)|^2 + \int_q f_{11}(t , t;q ) \right)\rho_{12}(t , t';k ) + \left ( \left[\varphi(t)\right]^2 + \int_q f_{12}(t , t;q)\right)\left[\rho_{11}(t , t';k)\right]^ * \right].\nonumber\\\end{aligned}\ ] ] in the above equations , @xmath193 is the coordination number , @xmath194 is the time - periodic free dispersion in the rotating frame , and the integrals are all taken over the brillouin zone .
furthermore , if one is interested in the equal - time correlation of the statistical correlator @xmath189 , using the symmetry relations in eq .
one arrives at the somewhat simplified equations @xmath195 ^ 2 + \int_q f_{12}(t , t;q)\right)\left[f_{12}(t , t;k)\right]^ * \right\},\nonumber\\ i\partial_tf_{12}(t , t;k ) & = & 2\bigg\ { ( zj_0-\mu)f_{12}(t , t;k ) + \varepsilon_k(t)f_{12}(t , t;k ) \nonumber\\ & & + \frac{u}{n_s}\left [ 2\left ( |\varphi(t)|^2 + \int_q f_{11}(t , t;q ) \right)f_{12}(t , t;k ) + \left ( \left[\varphi(t)\right]^2 + \int_q f_{12}(t , t;q)\right)\left[f_{11}(t , t;k)\right]^ * \right]\bigg\}. \label{eq : wcca_f}\end{aligned}\ ] ] we now generalise the wcca eom to any bipartite lattice . consider a bipartite lattice with the two sublattices labelled by @xmath196 and @xmath197 and periodic boundary conditions .
each sublattice contains @xmath198 number of sites . with this definition , additionally to the bogoliubov index @xmath156 , all correlators carry an additional index @xmath199 , and so does the condensate fraction .
the extended system of equal - time equations of motion for the wcca of the bhm reads @xmath200^*\left[\varphi^a(t)\right]^2 + 2\varphi^a(t ) \int_q f^{aa}_{11}(t , t;q ) + \left[\varphi^a(t)\right]^*\int_q f_{12}^{aa}(t , t;q ) \right],\nonumber\\ i\partial_t\varphi^b(t ) & = & ( zj_0-\mu)\varphi^b(t ) + \varepsilon_{k=0}(t)\varphi^a(t ) \nonumber\\ & & + \frac{u}{n_a}\left [ \left[\varphi^b(t)\right]^*\left[\varphi^b(t)\right]^2 + 2\varphi^b(t)\int_q f^{bb}_{11}(t , t ) + \left[\varphi^b(t)\right]^*\int_q f_{12}^{bb}(t , t ) \right ] , \label{eq : wcca_gpe_ab}\end{aligned}\ ] ] @xmath201^ * + \frac{u}{n_a}\left ( \left[\varphi^a(t)\right]^2 + \int_q f^{aa}_{12}(t , t;q)\right)\left[f^{aa}_{12}(t , t;k)\right]^ * \right\},\nonumber\\ \partial_t f^{bb}_{11}(t , t;k ) & = & 2\text{im}\left\ { \varepsilon^*_k(t)f^{ab}_{11}(t , t;k ) + \frac{u}{n_a}\left ( \left[\varphi^b(t)\right]^2 + \int_q f^{bb}_{12}(t , t;q)\right)\left[f^{bb}_{12}(t , t;k)\right]^ * \right\},\nonumber\\ i\partial_t f^{ab}_{11}(t , t;k ) & = & \varepsilon_k(t)\left ( f^{bb}_{11}(t , t;k ) - f^{aa}_{11}(t , t;k ) \right ) + \frac{u}{n_a}\bigg [ \nonumber\\ & & + 2\left ( |\varphi^a(t)|^2 - |\varphi^b(t)|^2 + \int_q f^{aa}_{11}(t , t;q ) - f^{bb}_{11}(t , t;q ) \right)f^{ab}_{11}(t , t;k)\nonumber\\ & & + \left ( \left[\varphi^a(t)\right]^2 + \int_q f^{aa}_{12}(t , t;q ) \right)\left[f^{ab}_{12}(t , t;k)\right]^ * - \left ( \left[\varphi^b(t)\right]^2 + \int_q f^{bb}_{12}(t , t;q ) \right)^*f^{ab}_{12}(t , t;k ) \bigg],\nonumber\\ i\partial_tf^{aa}_{12}(t , t;k ) & = & 2\bigg\ { ( zj_0-\mu)f^{aa}_{12}(t , t;k ) + \varepsilon_k(t)f^{ba}_{12}(t , t;k ) \nonumber\\ & & + \frac{u}{n_a}\left [ 2\left ( |\varphi^a(t)|^2 + \int_q f^{aa}_{11}(t ,
t;q ) \right)f^{aa}_{12}(t , t;k ) + \left ( \left[\varphi^a(t)\right]^2 + \int_q f^{aa}_{12}(t , t;q)\right)\left[f^{aa}_{11}(t , t;k)\right]^ * \right]\bigg\ } , \nonumber\\ i\partial_tf^{bb}_{12}(t , t;k ) & = & 2\bigg\{\varepsilon_k^*(t)f^{ab}_{12}(t , t;k ) + ( zj_0-\mu)f^{bb}_{12}(t , t;k ) \nonumber\\ & & + \frac{u}{n_a}\left [ 2\left ( |\varphi^b(t)|^2 + \int_q f^{bb}_{11}(t ,
t;q)\right ) f^{bb}_{12}(t , t;k ) + \left ( \left[\varphi^b(t)\right]^2 + \int_q f^{bb}_{12}(t , t;q ) \right)\left[f^{bb}_{11}(t , t;k)\right]^ * \right]\bigg\ } , \nonumber\\ i\partial_tf^{ab}_{12}(t , t;k ) & = & 2(zj_0-\mu)f^{ab}_{12}(t , t;k ) + \left(\varepsilon_k(t)^*f^{aa}_{12}(t , t;k ) + \varepsilon_k(t)f^{bb}_{12}(t , t;k)\right ) + \frac{u}{n_a}\bigg [ \nonumber\\ & & + 2\left ( |\varphi^a(t)|^2 + |\varphi^b(t)|^2 + \int_q f^{aa}_{11}(t , t;q ) + f^{bb}_{11}(t , t;q ) \right)f^{ab}_{12}(t , t;k ) \nonumber\\ & & + \left ( \left[\varphi^a(t)\right]^2 + \int_q f^{aa}_{12}(t , t;q ) \right)\left[f^{ab}_{11}(t , t;k)\right]^ * - \left ( \left[\varphi^b(t)\right]^2 + \int_q f^{bb}_{12}(t , t;q ) \right)^*f^{ab}_{11}(t , t;k ) \bigg ] .
\label{eq : wcca_fab } \end{aligned}\ ] ] all integrals in eqs . and
are taken over the reduced brillouin zone ( w.r.t . the @xmath202-sublattice symmetry ) .
equations and constitute a coupled set of non - linear equations , the solution of which produces the dynamics discussed in the main text . note that these eom can be applied to systems with arbitrary time - dependence ( not necessarily a periodic one ) and on an arbitrary bipartite lattice , such as the honeycomb lattice . for the analysis in the main text , the initial condition for the condensate fractions is chosen to be @xmath203 for @xmath204 and @xmath205 for @xmath206 , where @xmath70 is the total condensate fraction for the non - driven model in bogoliubov theory
in this appendix we estimate the timescales on which the wcca gives a reliable description of the physics , and discuss the dominant processes that ( in the weak coupling regime ) destabilize the various pre - thermal steady states discussed in the main text . we discuss each of the three regimes separately .
_ parametrically unstable region_. in this regime , the prethermalized phase is the one in which the momentum distribution is sharply peaked along momentum - space arcs as in fig . 3 ( a ) ( see main text ) . as in the main text , we treat the time - averaged dispersion as the unperturbed hamiltonian and look at the parametric instability growth rate growth rate due to a perturbation of the form @xmath207 .
the matrix element for pair creation is then @xmath208 ( the fourier coefficient @xmath209 from the parametric instability analysis above is essentially given by the bessel function ) , and for reasonably large drives this is linear in @xmath7 .
parametric instability predicts that these features will grow at the rate @xmath210 , where @xmath70 is the condensate amplitude .
the decay rate ( i.e. , inverse lifetime ) of the quasiparticles along these arcs , once they are formed , is limited by collisions , and fermi s golden rule implies that this decay rate is of order @xmath211 ; this is the rate at which these features spread out in momentum space .
thus there is a parametric separation in @xmath7 between the formation and decay rate of these peaks .
the leading collisional process comes from cubic terms of the form @xmath212 ( plus appropriate conjugates ) in the hamiltonian .
the golden rule rate for this particular process is @xmath213 where @xmath214 is the accessible two - particle density of states . here , @xmath215 is the energy of an excitation with quasimomentum @xmath216 . on dimensional grounds
this two - particle density of states must be inversely proportional to @xmath217 ; thus , the overall golden rule lifetime of a particular quasiparticle state will go as @xmath218 up to a multiplicative constant .
the ratio between the decay rate and the creation rate scales as @xmath219 .
thus the decay rate of a mode is slower than the creation rate whenever the condensate amplitude is large compared with the population of the mode ( essentially because the matrix element is not bose - enhanced to the same degree ) .
however , the decay rate is also suppressed with decreasing the interactions ( expected ) or increasing the drive amplitude . at short times ,
when the condensate is not appreciably depleted , the wcca is therefore reliable ; however , when the depletion becomes large the wcca also fails .
thus the regimes of validity of the wcca and bogoliubov theory in the parametrically unstable regime are essentially the same , although the wcca has the advantage of respecting particle number conservation _ exactly _ at all times .
_ stable region .
_ in the stable region there are two types of physical processes beyond the wcca .
( 1 ) the excitations created by the original quench into the phase have finite collisional lifetimes , as discussed above .
the momentum - space patterns in the stable region will dephase on this golden - rule timescale @xmath220 ; however , the condensate fraction will remain large and stable even after dephasing .
( 2 ) eventually , the system will absorb energy from the drive .
if the drive frequency is @xmath37 and the bandwidth of single - particle excitations is @xmath217 , then resonant absorption must involve at least @xmath221 quasiparticles .
it is straightforward to check that the associated golden rule rate , at weak coupling , is of the form @xmath222 .
when @xmath7 is sufficiently small , this heating timescale is much longer than the timescale on which the momentum - space patterns dephase ; thus the system should remain _ stable _ for extremely long times at high frequencies .
_ dynamically unstable region .
_ in this regime , the growth rates of unstable modes are of order @xmath223 , whereas the collision rates are of order @xmath224 at best , so at weak coupling we have a parametric window in @xmath7 where the wcca remains valid .
in this appendix , we discuss the dynamics governed by the wcca close to the first zero of the bessel function , @xmath226 , where the dispersion of the @xmath227 hamiltonian becomes flat ( central grey region in fig . 1 , main text ) .
for @xmath204 the dispersion of the free theory @xmath228 supports a stable minimum for @xmath62 , while for @xmath206 the stable minimum appears at @xmath63 . since the the two stable regions support different momentum modes , a phase transition occurs in between them .
therefore , it is required that one allows for a macroscopic population of both the modes in the immediate vicinity of @xmath229 .
this can be achieved by reducing the translational symmetry of the problem .
intuitively , a condensate at @xmath230 with amplitude @xmath231 flips a sign on every other site .
hence , one can choose to work in the original ( momentum - resolved ) basis @xmath232 , or in the site - resolved basis @xmath233 .
the two are related by a rotation . in order allow for a dynamical population of the @xmath231 condensate , eqs . and
require that the initial condition for @xmath234 be nonzero . in the @xmath202-basis ,
this is equivalent to saying that there is a slight difference in the condensate occupation on the two sublattices .
physically , this imbalance is caused by spontaneous symmetry breaking .
however , in the wcca one has to put in this imbalance by hand . in the following we refer to the small value @xmath235 as _
seed_. when the effective dispersion becomes flat @xmath236 [ fig [ fig : osc_condensate ] ] , the condensate undergoes oscillations between the @xmath237 and @xmath21 modes , with a period @xmath238 for small @xmath7 .
this behaviour is reminiscent of the collapse - and - revival effects seen for a bec that is suddenly quenched into the mott insulating phase@xcite , although the dynamics governed by wcca is classical .
the period of the transfer oscillations is also seed - dependent and increases with @xmath239 . even though our approximation does not capture a true mott insulating phase , the nonlinearities included in the wcca are sufficient to give rise to these oscillations . physically speaking ,
a quasiparticle - mediated channel is opened , through which particles flow from the condensate at @xmath14 to @xmath63 .
although it is present at any @xmath29 , this channel is only effective when the dispersion is sufficiently flat since the amplitude for the phonon- mediated transition @xmath240 scales as @xmath241 .
last , we briefly address the issue of heating . fig .
[ fig : energy ] shows the excess total ( i.e. , condensate plus quasiparticle ) energy density in the system , relative to the non - driven state . due to the abrupt turn - on of the periodic circularly polarised modulation , the energy changes discontinuously at @xmath126 .
as expected , the energy density increases due to heating in the parametrically unstable region , saturates in the stable region , and exhibits a small growth for @xmath242 .
notice the different behaviour in the parametrically unstable region compared to the dynamically unstable one : while in the former the energy grows due to the population of modes lying on the high - energy surface , in the latter the dynamically unstable modes appear close enough to the origin [ cf .
fig . 3 , panel ( c ) in main text ] so that the growth in energy density past the quench value is not substantial .
note that the system does not heat up even at fairly long times whenever the parameters are chosen to be in the stable region of the stability diagram . although ergodic periodically - driven systems are expected to eventually heat up to infinite temperature@xcite , in the weak - coupling limit this heating timescale ( which is due to collisions between quasiparticles ) is parametrically slower in the `` stable '' regimes of our phase diagram than in the `` unstable '' regimes . thus , for a range of present - day experiments , we expect that in the stable high - frequency regime there is no significant heating on experimentally relevant timescales .
for the time - evolution of the momentum distribution function , we refer to the three videos in the supplementary material .
the lower left panel shows the quasiparticle momentum distribution over the first brillouin zone , while the upper left panel is a top view of the same .
the upper right panel displays the time - evolution of the condensate fraction , while the lower right panel shows the energy density .
the parameters for each simulation can be found in the title .
the three movies correspond to the points in the stability diagram marked by ( a ) , ( b ) , ( c ) in fig . 1 ( main text ) .
|
we explore prethermal floquet steady states and instabilities of the weakly interacting two - dimensional bose - hubbard model subject to periodic driving .
we develop a description of the nonequilibrium dynamics , at arbitrary drive strength and frequency , using a weak - coupling conserving approximation .
we establish the regimes in which conventional ( zero - momentum ) and unconventional [ @xmath0-momentum ] condensates are stable on intermediate time scales .
we find that condensate stability is _ enhanced _ by increasing the drive strength , because this decreases the bandwidth of quasiparticle excitations and thus impedes resonant absorption and heating .
our results are directly relevant to a number of current experiments with ultracold bosons .
periodically driven systems@xcite often exhibit exotic phenomena that are absent in their non - driven counterparts@xcite .
classic examples include the kapitza pendulum and the periodically kicked rotor .
recently , periodically modulating optical lattices has attracted interest as a way of controlling hopping processes@xcite in order to engineer gauge fields@xcite , topological band structures@xcite , and associated exotic states of matter .
such exotic states are known to exist in noninteracting systems and in certain mean - field models ; the extent to which they survive in the presence of interactions is a central open question .
it is believed , from the eigenstate thermalization hypothesis@xcite , that driven interacting systems will generically heat up to infinite temperature at sufficiently late times@xcite . nevertheless , in some parameter regimes these heating times will be parametrically slower than the system s characteristic time scales . in that case , the system will rapidly approach a `` prethermalized '' floquet steady state@xcite , which governs the dynamics until the much later heating time scales . in the present work , we study these prethermal states in the weakly interacting , two - dimensional , periodically - driven bose - hubbard model ( bhm ) . the regime
we explore is directly relevant to experiments@xcite , in which weak interactions are present .
we employ a self - consistent weak - coupling conserving approximation ( wcca ) which treats the coupled nonlinear dynamics of the condensate and the quasiparticle spectrum while neglecting collisions between quasiparticles .
this approximation is justified at weak coupling since nonlinearities are important at much shorter times than the collisional time scales . within the wcca
, we find a phase diagram ( fig .
[ fig : stability_diag ] ) featuring at low drive frequency a regime in which the superfluid state is _ already _ unstable within bogoliubov theory , owing to the resonant creation of quasiparticle pairs , and a regime ( at high drive frequency ) where the superfluid is stable . in the wcca
, there is a sharp phase transition between these ; when effects beyond weak coupling are included , there will be a qualitative difference in heating rates .
thus , in the `` stable '' regions of fig .
[ fig : stability_diag ] , the system initially reaches a prethermalized superfluid state featuring a nonequilibrium quasiparticle distribution and then eventually heats up . for strong
driving , the prethermalized superfluid state is exotic , involving condensation at momentum @xmath1 .
the existence of this exotic phase in the high - frequency limit has previously been established@xcite ; we find that it persists for intermediate frequencies as well . remarkably , we find that the stable phase is _ enhanced _ for intermediate drive strengths , since the drive both creates quasiparticle pairs when this is a resonant process , and decreases the effective hopping rate and thus the effective bandwidth of quasiparticle excitations .
a key conclusion of our work is that , for weak interactions but general drive amplitude and frequency , the condensate becomes unstable when the drive frequency is parametrically resonant with the _ drive - renormalized _ time - averaged bandwidth .
therefore , parametric resonance occurs at _ lower _ frequencies when the drive strength is ramped up . _
_ model.__we consider the bose - hubbard model on a square lattice in the presence of a circularly - polarised time - periodic force @xmath2 : @xmath3 .
\label{eq : hlab(t)}\ ] ] the operator @xmath4 creates a boson on lattice site @xmath5 . the tunnelling and interaction strength are denoted by @xmath6 and @xmath7 , respectively . to achieve non - trivial dynamics in the high - frequency regime ,
we scale the driving amplitude linearly with the driving frequency @xmath8.@xcite ; we define @xmath9 .
we transform this hamiltonian into a rotating frame ( cf .
supplementary material@xcite ) , giving : @xmath10 thus , in the rotating frame , the system experiences an effective time - dependent gauge potential @xmath11 .
the time evolution of @xmath12-invariant quantities ( and thus the stability ) remains the same in both frames@xcite .
_ method . _ to study the driven system at arbitrary frequencies , we employ a self - consistent , weak - coupling conserving approximation ( wcca ) .
the wcca involves deriving equations of motion from a two - particle irreducible effective action@xcite within the nonequilibrium schwinger - keldysh formalism@xcite , keeping only diagrams to first order in @xmath7 ( see @xcite ) .
unlike simple perturbation theory or bogoliubov theory , the wcca respects unitarity and conservation laws@xcite , and thus gives physically sensible results for all times ; in particular , it allows the exponential growth of unstable modes to be cut off by the resulting depletion of the condensate . while the wcca is _ not
_ guaranteed to yield a gapless excitation spectrum@xcite , the low - frequency behavior of the spectrum is irrelevant for the phenomena discussed here .
our approach is equivalent to a fully self - consistent , time - dependent hartree - fock - bogoliubov ( hfb ) approximation@xcite ; our formulation , however , can more readily be extended to higher orders in @xmath7 .
( color online ) . ( a ) time evolution of the condensate fraction for @xmath13 driving cycles , starting from a bogoliubov initial state localised at @xmath14 for @xmath15 .
( b ) decay rate to @xmath16 of the condensate curves for @xmath17 ( boldface points in fig .
[ fig : stability_diag ] ) .
error bars are set by the difference of the inverse times , determined by the first and last time the curve passes through @xmath18 taking into account the oscillatory behaviour . ]
the wcca equations of motion@xcite were solved numerically .
for the results presented here , we prepared the system on a @xmath19 lattice in the ground state of bogoliubov theory .
we allow for a macroscopic population of the @xmath20 mode to allow for a condensate at momentum @xmath21 . to study the nonequilibrium dynamics , we abruptly turn on the periodic drive and propagate the initial state for @xmath13 driving cycles using eqs .
( 15 ) and ( 16 ) of@xcite .
we checked that the results are insensitive to system size . _
stability diagram_.the stability phase diagram is shown in fig .
[ fig : stability_diag ] .
previous work has investigated the driven bose - hubbard model@xcite and related models@xcite using various approximation schemes ; we go beyond these works by treating both the condensate and quasiparticle sectors , including the feedback between them .
thus , we are able to explore instabilities originating in either sector on equal footing .
we first discuss two analytically tractable limits , corresponding to high - frequency driving ( i.e. , going along the @xmath22 axis of fig . [
fig : stability_diag ] ) and to low - amplitude driving ( i.e. , going along the @xmath23 axis ) .
in the first case , the dynamics is approximately governed by an effective _ time - average _ hamiltonian@xcite : @xmath24 the periodic modulation renormalizes the hopping to @xmath25 , where @xmath26 is the zeroth - order bessel function of the first kind , which is a damped oscillatory function with the first zero at @xmath27 , the second at @xmath28 , etc .
thus , as @xmath29 is increased , the time - averaged hopping decreases , until the dispersion flattens at @xmath30 .
for @xmath31 the dispersion flips sign , and acquires a stable minimum at @xmath1 .
thus , in the high - frequency limit the condensate at @xmath32 is stable when @xmath33 , whereas the condensate at @xmath21 is stable when @xmath34 .
moreover , for commensurate filling , the superfluid phase should transition into a mott insulating state around @xmath35 determined by the phase boundary @xmath36.@xcite this transition regime , marked by the thin vertical strip in fig .
[ fig : stability_diag ] , is beyond the validity of the wcca ; our wcca simulations in this regime give oscillatory behavior , see @xcite .
a second analytically tractable limit is that of weak driving , at arbitrary @xmath37 .
the dominant effects can be inferred from linear stability analysis around the non - driven state . in terms of bogoliubov quasiparticle operators
@xmath38 , the system - drive coupling includes terms of the form @xmath39 , involving the emission of pairs of quasiparticles from the condensate .
the emission rate is related to the density of states of two - quasiparticle excitations at @xmath37 . specifically ,
if the non - driven system has quasiparticle excitations at energy @xmath40 such that @xmath41 , absorption will occur and the system will be unstable . on the other hand , if @xmath42 , where @xmath43 is the approximate bandwidth of bogoliubov excitations , then absorption does not occur and the system is stable . combining the insights from these two limits allows us to understand the entire stability phase diagram .
the drive creates pairs of _ renormalized _ bogoliubov quasiparticles , which have an effective bandwidth @xmath44 .
we define @xmath45 - \text{min}_k[e_\text{ave}({\bf k})]$ ] as the time - averaged floquet - bogoliubov bandwidth ; in terms of this , the stability condition reads @xmath46 equation is consistent with our numerical results ( fig .
[ fig : stability_diag ] ) .
this result is unexpected since the time - averaged hamiltonian is valid at infinite frequency whereas parametric resonance is a low - frequency phenomenon but can be understood as follows .
the hopping matrix element in the driven system can be expanded as @xmath47 .
we absorb the time - independent @xmath48 component in the unperturbed hamiltonian , and treat the @xmath49 term , which oscillates at @xmath37 , perturbatively .
the perturbation is small for @xmath50 , because the matrix element for creating two quasiparticles is proportional to both @xmath51 [ which need not be small ] and @xmath7 [ which is assumed to be small ] .
we then use parametric instability analysis@xcite with the renormalized dispersion , and conclude that an instability occurs when @xmath52 .
when @xmath53 , the critical driving frequency is given by @xmath54 note that in the present case , resonant absorption occurs for drive strengths up to _ twice _ the single - particle bandwidth ; by contrast , in noninteracting systems , no absorption occurs for @xmath55 .
the presence of absorption at frequencies exceeding the single - particle bandwidth is generic in interacting systems .
_ condensate evolution_.figure [ fig : condensate ] [ panel ( a ) ] shows the evolution of the condensate fraction in various regimes : in the parametrically unstable regime ( solid blue line ) , the condensate slowly decays ; in the stable regime ( dashed red line ) , it saturates to a prethermalized value , which is generally lower than the bogoliubov value ( since @xmath56 ) .
the system enters a steady - state with constant in time evolution when measured stroboscopically .
when the initial condensate is at the band maximum ( dash - dotted black line ) , the condensate decays rapidly .
panel ( b ) shows the decay rate as a function of drive amplitude in the parametrically unstable regime : note that the decay rate depends not only on drive strength @xmath29 , but also on @xmath7 and @xmath37 . very close to the region @xmath57 ( grey strip in fig .
[ fig : stability_diag ] ) , the wcca gives strong oscillations of the particle density between the condensates at @xmath58 and @xmath21 ( see @xcite ) ; however , as previously noted , the wcca is not reliable here .
a natural further observable is the total energy of the system , which grows in the unstable phases and saturates in the stable phases ( see @xcite ) . _
( quasi-)momentum distribution._fig .
[ fig : n_k ] plots snapshots of the quasimomentum ( i.e. , lattice momentum ) distribution ; the time evolution of this quantity is shown in@xcite .
specifically , the quantity plotted is @xmath59 , i.e. , the condensate peak is subtracted .
the quasimomentum distribution can be directly accessed through band mapping followed by time - of - flight imaging . moreover , as we are concerned with a single - band model , one can extract this distribution directly from time - of - flight imaging , by focusing on momenta within the first brillouin zone .
figure [ fig : n_k ] ( a ) shows the parametrically unstable case , where quasiparticles are strongly excited around the quasimomentum surface @xmath60 matching the resonance condition . within bogoliubov theory , the ( time - averaged ) excitation intensity should be uniform along this surface .
however , as the points along this surface are not symmetry - related , the nonlinearities included in the wcca favor some points on the excitation surface , as seen in the intensity pattern in fig .
[ fig : n_k ] ( a ) .
figure [ fig : n_k ] ( b ) shows the _ stable _ case . here ,
by contrast with panel ( a ) , the quasiparticle population remains low throughout the brillouin zone .
as expected from bogoliubov theory , bosonic modes satisfying @xmath61 should have appreciable occupation in the steady state ; this region expands as the dispersion flattens .
the intricate patterns in momentum space are due to the abrupt turn - on of the drive which initializes the floquet - bogoliubov quasiparticle states out of equilibrium and are absent when the drive is instead gradually ramped up .
these patterns evolve nontrivially with time ( see @xcite ) . finally , fig .
[ fig : n_k ] ( c ) illustrates the case in which the initial state is a condensate at @xmath62 , but the dispersion is inverted ( @xmath31 ) so that the only _ stable _ condensate is supported at @xmath63 .
thus the initial state is unstable regardless of @xmath37 .
let us consider the infinite - frequency limit ; which amounts to a sudden quench of the single - particle dispersion . computing the bogoliubov spectrum around a condensate at @xmath62 in an inverted dispersion , we find that modes with momenta near @xmath62 acquire imaginary frequencies ( and thus grow exponentially ) , whereas modes with large momenta are _ stable _ limit , _ all _ modes are stable as there are no decay processes . ] .
the unstable modes are determined by the condition @xmath64 , where @xmath65 is the single - particle floquet dispersion .
these modes are dynamically stabilized due to the nonlinear feedback of the self - consistent treatment@xcite .
our numerical results with the wcca confirm this picture : the unstable modes at small quasimomenta acquire large populations , whereas the large - quasimomentum modes do not .
this behavior is _ specific _ to the wcca ; in a real system it will correspond to intermediate - time dynamics @xmath66 . on longer times
, collisions between quasiparticles should cause large occupation numbers across the brillouin zone , see @xcite . _
discussion_.we briefly outline the validity of the wcca in the three regimes of interest ( for details see @xcite ) . in the parametrically unstable regime
, the stability analysis suggests that unstable modes grow at the rate @xmath67 , while the momentum arcs in fig .
[ fig : condensate ] ( a ) decay at a golden rule rate @xmath68 . hence , as long as @xmath69 , the formation rate is greater than the decay rate and the wcca is reliable . in the stable region ,
the condensate fraction @xmath70 remains large , and the wcca remains valid , until very late times , when resonant absorption involving @xmath71 quasiparticles becomes dominant . for large @xmath37
, this is a very high - order and therefore very slow process . finally , in the dynamically unstable phase , the wcca physics is valid up to times @xmath72 ( the collisional time scale ) .
thus , at weak coupling , there is a parametrically large window between @xmath73 and @xmath72 where the wcca description is correct .
the main experimental prediction of this work a parametric change in heating rates as a function of drive amplitude and frequency can be measured in present - day experiments , which are naturally in the weak - coupling regime . for the experiment in ref .
the parameters were chosen as @xmath74 , @xmath75 , and @xmath76 , which is within the regime we considered .
for realistic experiments in optical lattices , the presence of higher bands can lead to instability even at high drive frequencies @xmath37 . in this case
there are three regimes : ( i ) if @xmath37 is less than twice the renormalized bandwidth @xmath77 of the lower band , the system is parametrically unstable as discussed above ; ( ii ) if @xmath37 is larger than @xmath78 , smaller than the band gap to the upper band , and furthermore chosen such that any @xmath79-photon resonances to higher bands @xcite are suppressed , then the system is stable within wcca .
( iii ) if @xmath37 exceeds the band gap , the drive can mediate interband transitions , leading to instability again .
for a square optical lattice with typical lattice potential @xmath80 , @xmath81 , the bandwidth of the lowest band is @xmath82 [ the time - averaged bandwidth @xmath77 is reduced by a factor of @xmath26 ] , and the gap to the second bloch band is @xmath84 .
although we focused on a square lattice , the arguments generalize to other lattices , such as the honeycomb lattice , in which topologically non - trivial states exist .
note that topological gaps in _ mechanically shaken _ optical lattices scale as @xmath85@xcite .
hence , in order to engineer topological insulators with large gaps ( and a large region of non - zero berry curvature around them ) , it is desirable to go to lower frequencies .
our results impose a fundamental limit for _ weakly - interacting bosonic _ systems on how small the frequency can be , since for @xmath86
the system becomes unstable .
more generally , our results suggest that conserving approximations , whether controlled by weak coupling or some other parameter as in large-@xmath87 models@xcite , are ways of exploring dynamical phase transitions in models that are both _ interacting _
( unlike free - particle models ) and _ finite - dimensional _
( unlike the kapitza pendulum ) . the critical properties of such transitions are a fruitful theme for future work .
although in practice such phase transitions will be smeared out by higher - order effects , the associated crossovers should still be experimentally observable . _
we thank d. abanin , m. babadi , i. bloch , l. dalessio , e. dalla torre , n. goldman , m. kolodrubetz , a. polkovnikov , and u. schneider for interesting and fruitful discussions , and especially acknowledge the help of m. lohse in determining the experimentally relevant parameters .
the authors acknowledge support from the nsf grant dmr-1308435 , harvard - mit cua , afosr new quantum phases of matter muri , the aro - muri on atomtronics , aro muri quism program , afosr fa9550 - 13 - 1 - 0039 and bsf 2010318 , and technical university munich - institute for advanced study , funded by the german excellence initiative and the european union fp7 under grant agreement 291763 . 99 j. h. shirley , phys . rev . * 138 * , b979 ( 1965 ) .
h. sambe , phys .
rev . a * 7 * , 2203 ( 1973 ) . h. p. breuer , k. dietz , and m. holthaus , physica ( amsterdam ) * 46d * , 317 ( 1990 ) .
h. p. breuer and m. holthaus , ann .
phys .
( n.y . ) * 211 * , 249 ( 1991 ) .
n. goldman and j. dalibard , phys .
rev .
x * 4 * , 031027 ( 2014 ) .
m. bukov , l. dalessio , and a. polkovnikov , adv .
phys . * 64 * , 139226 ( 2015 ) .
a. eckardt , and e. anisimovas , new
. j. phys .
* 17 * , 093039 ( 2015 ) .
d. h. dunlap and v. m. kenkre , phys .
rev .
b * 34 * , 3625 ( 1986 ) ; phys . rev .
b * 37 * , 6622 ( 1988 ) .
a. eckardt , c. weiss , and m. holthaus , phys .
rev .
lett .
* 95 * , 260404 ( 2005 ) .
h. lignier , c. sias , d. ciampini , y. singh , a. zenesini , o. morsch , and e. arimondo , phys .
rev .
lett .
* 99 * , 220403 ( 2007 ) .
a. zenesini , h. lignier , d. ciampini , o. morsch , and e. arimondo , phys .
rev .
lett . * 102 * , 100403 ( 2009 ) . c. e. creffield , f. sols , d. ciampini , o. morsch , and e. arimondo , phys .
rev .
a * 82 * , 035601 ( 2010 ) .
g. jotzu , m. messer , f. grg , d. greif , r. desbuquois and t. esslinger , phys .
rev .
lett . * 115 * , 073002 ( 2015 ) .
j. struck , c. lschlger , r. le targatn , p. soltan- panahi , a. eckardt , m. lewenstein , p. windpassinger , and k. sengstock , science * 333 * , 996 ( 2011 ) .
d. jaksch and p. zoller , ann .
phys .
( amsterdam ) * 315 * , 52 ( 2005 ) .
e. j. mueller , phys .
rev .
a * 70 * , 041603 ( 2004 ) .
j. struck , m. weinberg , c. lschlger , p. windpassinger , j. simonet , k. sengstock , r. hppner , p. hauke , a. eckardt , m. lewenstein , and l. mathey , nature physics * 9 * , 738 - 743 ( 2013 ) .
m. aidelsburger , m. atala , m. lohse , j. t. barreiro , b. paredes , and i. bloch , phys .
rev .
lett . * 111 * , 185301 ( 2013 ) .
h. miyake , g. a. siviloglou , c. j. kennedy , w. c. burton , and w. ketterle , phys .
rev .
lett . * 111 * , 185302 ( 2013 ) .
m. atala , m. aidelsburger , m. lohse , j. t. barreiro , b. paredes , and i. bloch , nature physics * 10 * , 588593 ( 2014 ) . c. j. kennedy , w. c. burton , w. ch .
chung , and w. ketterle , nature physics * 11 * , 859864 ( 2015 ) . n. flschner , b. s. rem , m. tarnowski , d. vogel , d .- s .
lhmann , k. sengstock , and c. weitenberg , arxiv:1509.05763 , ( 2015 ) .
t. oka and h. aoki , phys . rev .
b * 79 * 081406(r ) , ( 2009 ) .
t. kitagawa , t. oka , a. brataas , l. fu , and e. demler , phys .
rev .
b * 84 * 235108 ( 2011 ) .
g. jotzu , m. messer , r. desbuquois , m. lebrat , t. uehlinger , d. greif , and t. esslinger , nature ( london ) * 515 * , 237 ( 2014 ) .
m. aidelsburger , m. lohse , c. schweizer , m. atala , j. t. barreiro , s. nascimbne , n. r. cooper , i. bloch , and n. goldman , nature physics * 11 * 162166 ( 2015 ) .
n. h. lindner , g. refael and v. galitski , nature physics * 7 * , 490 ( 2011 ) .
m. s. rudner , n. h. lindner , e. berg , and m. levin , phys .
rev .
x * 3 * , 031005 ( 2013 ) .
m. bukov , m. kolodrubetz and a. polkovnikov , arxiv:1510.02744 ( 2015 ) .
j. m. deutsch , phys .
rev .
a * 43 * , 2046 ( 1991 ) .
m. srednicki , phys .
rev .
e * 50 * , 888 ( 1994 ) .
m. rigol , v. dunjko , and m. olshanii , nature ( london ) * 452 * , 854 ( 2008 ) .
l. dalessio and a.polkovnikov , ann . phys . * 333 * , 19 ( 2013 )
. l. dalessio and m. rigol , phys .
rev .
x * 4 * , 041048 ( 2014 ) .
a. lazarides , a. das , and r. moessner , phys .
rev .
lett . * 112 * , 150401 ( 2014 ) .
a. lazarides , a. das and r. moessner , phys .
rev .
e * 90 * , 012110 ( 2014 ) .
p. ponte , z. papi , f. huveneers and d. a. abanin , phys .
rev .
lett . * 114 * , 140401 ( 2015 ) .
d. abanin , w. de roeck , and f. huveneers , arxiv:1507.01474 ( 2015 ) . t. kuwahara , t. mori , and k. saito , arxiv:1508.05797 , ( 2015 ) . t. mori , t. kuwahara , and k. saito , arxiv:1509.03968 , ( 2015 ) .
d. a. abanin , w. de roeck , and w. w. ho , arxiv:1510.03405 ( 2015 ) .
j. berges , s. borsanyi , and c. wetterich , phys .
rev .
lett . * 93 * , 142002 ( 2004 ) .
e. canovi , m. kollar , and m. eckstein , arxiv:1507.00991 ( 2015 ) .
m. bukov , and a. polkovnikov , phys .
rev .
a * 90 * , 043613 ( 2014 ) .
j. m. cornwall , r. jackiw and e. tomboulis , phys .
rev .
d * 10 * , 2428 ( 1974 ) .
j. bauer , m. babadi , and e. demler , phys .
rev .
b * 92 * , 024305 ( 2015 ) .
m. babadi , e. demler , and m. knap , phys .
rev .
x * 5 * , 041005 ( 2015 ) .
p. c. hohenberg and p. c. martin , ann .
phys .
( n.y . ) * 34 * , 291 ( 1965 ) .
a. griffin , phys .
rev .
b * 53 * , 9341 ( 1996 ) .
a. m. rey , b. l. hu , e. calzetta , a. roura , and c. w. clark , phys . rev .
a * 69 * , 033610 ( 2004 ) .
a. buchleitner and a. r. kolovsky , phys .
rev .
lett . * 91 * , 253002 ( 2003 ) .
a. r. kolovsky and a. buchleitner , phys .
rev .
e * 68 * , 056213 ( 2003 ) .
a. tomadin , r. mannella , and s. wimberger , phys .
rev .
lett . * 98 * , 130402 ( 2007 ) .
a. r. kolovsky , h. j. korsch , and e -m .
graefe , phys .
rev .
a * 80 * , 023617 ( 2009 ) . c. e. creffield , phys .
rev .
a * 79 * , 063612 ( 2009 ) .
a. kolovsky , europhys . lett . * 93 * , 20003 ( 2011 ) ; c. e. creffield and f. sols , europhys .
lett .
* 101 * , 40001 ( 2013 ) . c. a. parra - murillo , j. madroero , and s. wimberger , phys .
rev .
a * 88 * , 032119 ( 2013 ) . c. a. parra - murillo , j. madroero , and s. wimberger , comput .
. comm . * 186 * , 1930 ( 2015 ) .
t. stferle , h. moritz , c. schori , m. khl , and t. esslinger , phys .
rev .
lett . * 92 * , 130403 ( 2004 ) . c. schori , t. stferle , h. moritz , m. khl , and t. esslinger phys .
rev .
lett . * 93 * , 240402 ( 2004 ) .
m. krmer , c. tozzo , and f. dalfovo , phys .
rev .
a * 71 * , 061602(r ) ( 2005 ) . c. tozzo , m. krmer , and f. dalfovo , phys .
rev .
a * 72 * , 023613 ( 2005 ) .
i. carusotto , r. balbinot , a. fabbri , and a. recati , eur .
phys .
j. d * 56 * 391 - 404 ( 2009 ) . c. zhang , j. liu , m. g. raizen , and q. niu , phys .
rev .
lett . * 92 * , 054101 ( 2004 ) .
s. choudhury and e. j. mueller , phys .
rev . a * 91 * , 023624 ( 2015 ) .
t. bilitewski and n. r. cooper , phys .
rev . a * 91 * , 063611 ( 2015 ) .
s. choudhury and e. j. mueller , arxiv:1508.07572 , ( 2015 ) .
b. capogrosso - sansone , s. g. soyler , n. prokofev , and b. svistunov , phys .
rev .
a * 77 * , 015602 ( 2008 ) .
m. knap , e. arrigoni , and w. von der linden , phys .
rev .
b * 81 * , 024301 ( 2010 ) .
m. weinberg , c. lschlger , c. strter , s. prelle , a. eckardt , k. sengstock , and j. simonet , arxiv:1505.02657 ( 2015 ) .
m. moshe and j. zinn - justin , phys . rep . *
385 * , 69 ( 2003 ) .
b. sciolla and g. biroli , phys .
rev .
b * 88 * , 201110(r ) ( 2013 ) . p. smacchia , m. knap , e. demler , and a. silva , phys .
rev .
b * 91 * , 205136 ( 2015 ) .
a. chandran and s. l. sondhi , arxiv:1506.08836,(2015 ) .
see supplemental material , which includes refs .
@xcite .
m. bukov and m. heyl , phys .
rev .
b * 86 * , 054304 ( 2012 ) . l. d. landau and e. m. lifshitz , _ mechanics _ ( elsevier , amsterdam , 2008 ) .
a. kamenev , _ field theory of non - equilibrium systems _
( cambridge university press , cambridge , england 2011 ) .
j. berges , aip conf.proc . *
739 * , 3 - 62 ( 2005 ) .
m. greiner , o. mandel , t. w. hnsch , and i. bloch , nature ( london ) * 419 * , 51 ( 2002 ) .
|
the standard model and most of its supersymmetric extensions suffer from being a theory of massless particles .
therefore a mechanism that would provide masses is required . among several possibilities the higgs mechanism plays a major role .
it assumes the existence of an additional scalar field , the higgs field , which has non - zero vacuum expectation value ( vev ) .
the correct implementation of this mechanism leads in the standard model not only to massive gauge bosons ( with the photon correctly remaining massless ) , but also to massive fermions , and a proper electroweak symmetry breaking from the weak gauge groups @xmath2 to the electromagnetism @xmath3 .
these features make this mechanism an extremely convenient and elegant solution
. the experimental smoking gun confirming this theory would be the discovery of the higgs boson .
earlier this year the tevatron collaborations cdf and d0 reported an excess of events in the higgs to two photons channel ( @xmath4 ) observed in the mass region 120140 gev @xcite .
the significance of these data were reported on the level of @xmath5 . only recently the newly announced lhc7 results @xcite from the atlas and cms collaborations confirmed an excess of events in the same channel within 115130 gev range , with a maximum at 125 gev , at the statistical significance of @xmath6 .
even though @xmath5 can not be named a discovery , an effect independently obtained within very similar mass ranges by four project working on two biggest accelerators in the world may give hope that some new particle has been observed .
even though the experimental results are at first interpreted within the standard model , its supersymmetric and other exotic extensions of various kinds can also be tested against the newly reported findings @xcite . following this line of research , in this communication we discuss an @xmath1-parity violating minimal supergravity ( rpv msugra ) model and constrain it by the liberal condition for the lightest higgs boson mass @xmath7 . also
the specific case of @xmath8 is considered .
the @xmath1-parity is a multiplicative quantum number defined as @xmath9 and implies stability of the lightest supersymmetric particle . in this paper , following ref .
@xcite , we adopt the so - called generalized baryon parity in the form of a discrete @xmath10 symmetry @xmath11 which ensures the stability of the proton and lack of unwanted dimension-5 operators . in short
, @xmath1 is equal to + 1 for ordinary particles , and @xmath12 for supersymmetric partners , and this @xmath1 is usually assumed to be conserved in interactions .
this assumption , however , is based mainly on our will to exclude lepton and baryon number violating processes , which has not been observed in the low - energy regime .
notice , that the generation lepton numbers @xmath13 , @xmath14 , and @xmath15 , also conserved in the standard model , are broken in the neutrino oscillations .
there is in fact no underlying principle which forbids breaking of @xmath16 or @xmath17 .
the baryon number violation is highly constrained by the proton decay , but the lepton number violation may occur at high energies . in general supersymmetric models
one often discusses the possibility of having the @xmath1-parity violating terms , properly suppressed , in the theory .
these are the so - called @xmath1parity violating models .
we perform the calculations within the framework described in detail in ref . @xcite .
this model takes into account full dependence of the mass matrices and renormalization group equations on the @xmath1parity violating couplings .
we define it below by quoting the expressions for the superpotential and soft supersymmetry breaking lagrangian .
next , we discuss the free parameters of the model , and the higgs sector .
the interactions are defined by the superpotential , which consists of the @xmath1parity conserving ( rpc ) and violating part @xmath18 where @xmath19 , \label{wrpc}\\ w_{\rm rpv}&= & \epsilon_{ab}\left [ \frac{1}{2}({\bf \lambda}_{e^k})_{ij } l_i^a l_j^b{\bar e}_k + ( { \bf \lambda}_{d^k})_{ij } l_i^a q_j^{xb } { \bar d}_{kx } \right ] . \nonumber \\
\label{wrpv}\end{aligned}\ ] ] here * y * s are the 3@xmath203 trilinear yukawa - like couplings , @xmath21 the bilinear higgs coupling , and ( @xmath22 ) and ( @xmath23 ) are the @xmath1-parity violating trilinear and bilinear terms . @xmath24 and @xmath25 denote the @xmath26 left - handed doublets , while @xmath27 , @xmath28 and @xmath29 are the right - handed lepton , up - quark and down - quark @xmath26 singlets , respectively . @xmath30 and
@xmath31 mean two higgs doublets .
we have introduced color indices @xmath32 , generation indices @xmath33 and the @xmath26 gauge indices @xmath34 .
the supersymmetry is not observed in the regime of energies accessible to our experiments , therefore it must be broken at some point . a convenient method to take this fact into account
is to introduce explicit terms , which break supersymmetry in a soft way , ie .
, they do not suffer from ultraviolet divergencies .
we add them in the form of a scalar lagrangian @xcite , @xmath35 \nonumber \\ & + & [ ( { \bf a}_{e^k})_{ij } l_i l_j e_k + ( { \bf a}_{d^k})_{ij } l_i q_j d_k + ( { \bf a}_{u^i})_{jk } u_i d_j d_k \nonumber \\ & & - d_i l_i h_u + \mathrm { h.c . } ] , \nonumber\end{aligned}\ ] ] where the lower case letter denotes the scalar part of the respective superfield .
@xmath36 are the gaugino masses , and @xmath37 ( @xmath38 , @xmath39 ) are the soft supersymmetry breaking equivalents of the trilinear ( bilinear ) couplings from the superpotential .
one of the weaknesses of the supersymmetric models is their enormous number of free parameters , which easily may exceed 100 .
this lowers significantly their predictive power .
therefore it is a custom approach to impose certain boundary conditions , which allow in turn to _ derive _ other unknown parameters .
one may either start with certain known values at low energies and evaluate the values of other parameters at higher energies , or assume , along the grand unified theories ( gut ) line of thinking , common values at the unification scale and evaluate them down . in both cases the renormalization group equations ( rge )
we follow the top - down approach by assuming the following : * we introduce a common mass @xmath40 for all the scalars at the gut scale @xmath41 , * we introduce a similar common mass @xmath42 for all the fermions at the gut scale , * we set all the trilinear soft supersymmetry breaking couplings @xmath37 to be proportional to the respective yukawa couplings with a common factor @xmath43 , @xmath44 at @xmath45 .
the only couplings which we allow to evolve freely , are the rpv @xmath46 s , which are set to a common value @xmath47 at @xmath48 and are being modified by the rge running only .
the reason for this is , that we want to keep them non - zero , but on the other hand their influence at the low scale must be small . in the first part of this presentation we will fix @xmath49 , which assures only small admixture of the rpv interactions .
later , we discuss the impact of @xmath47 on the results for certain set of input parameters .
( for a discussion of the @xmath50 s impact on the rge running see , eg . , ref .
@xcite ) .
there are two remaining free parameters in the model , one describing the ratio of the vacuum expectation values of the two higgs doublets @xmath51 , the other being the sign of the higgs self - coupling constant , @xmath52 , which gives altogether only six free parameters . in this calculations
we have kept @xmath21 positive , and fixed @xmath47 , so the resultant parameter space to analyze is four dimensional .
it is also a common practice to restrict the calculations to the third family of quarks and leptons only , which is supposed to give the dominant contributions . since
this issue is difficult to keep under control , we keep dependence on all three families . also , unlike the authors of ref .
@xcite , we keep the full dependence on all the rpv couplings .
the procedure of finding the minimum for the scalar potential is quite involving .
the equations one has to solve in this model are given in ref .
@xcite , however , the numerical procedure used by us differs slightly from the one presented in the cited paper .
the goal is to find the values of @xmath21 , @xmath53 , @xmath38 , and @xmath39 , as well as the five vacuum expectation values @xmath54 of the two higgs bosons and three sneutrinos .
the initial values for the vev s are @xmath55 , @xmath56 , @xmath57 , where @xmath58 .
we start by setting @xmath59 and evaluating @xmath60 , @xmath61 , and @xmath62 to the @xmath45 scale .
there we impose the gut unification conditions and run everything down back to the @xmath48 scale . in this first iteration the best minimization scale for the scalar potential @xmath63^{1/2 } [ ( { \bf m}_q^2)_{33}]^{1/2}}\ ] ] is calculated . at this scale initial values of @xmath21 and @xmath38 are found , according to the relations @xmath64 next , a rge run is performed from @xmath65 to @xmath48 , but this time the non - zero values of @xmath21 and @xmath38 generate non - zero values for the @xmath53 and @xmath39 , providing starting point for the next iteration .
the second iteration repeats the same steps as the first one , with the exception that now all @xmath21 , @xmath53 , @xmath38 , and @xmath39 contribute to the rge running .
getting back to the ( new ) @xmath65 scale , we solve for @xmath21 , @xmath38 , and @xmath66 using the loop - corrected equations @xmath67 \nonumber \\ & - & \left [ m_{h_u}^2 + |\kappa_i|^2 -\frac{1}{2}(g^2+g_2 ^ 2)v_i^2 -d_i~\frac{v_i}{v_u } \right ] \tan^2\beta \biggr \ } \nonumber \\ & - & \frac{m_z^2}{2 } , \label{eq : mub1 } \\ b & = & \frac{\sin 2\beta}{2 } \big [ ( m_{h_d}^2 - m_{h_u}^2 + 2|\mu|^2 + |\kappa_i|^2 ) \nonumber \\ & & \hskip 0.85truecm + ( { \bf m}_{l_ih_d}^2 + \kappa_i^*\mu)\frac{v_i}{v_d } -d_i~\frac{v_i}{v_u } \big ] , \label{eq : mub2}\end{aligned}\ ] ] and the so - called tadpole equations for the sneutrino vev s , which explicitely read @xmath68 + v_2 [ ( { \bf m}_l^2)_{21 } + \kappa_1 \kappa_2^ * ] + v_3 [ ( { \bf m}_l^2)_{31 } + \kappa_1 \kappa_3^ * ] = - [ { \bf m}_{h_d l_1}^2 + \mu^*\kappa_1]v_d + d_1 v_u , \nonumber \\ & & v_1 [ ( { \bf m}_l^2)_{12 } + \kappa_2 \kappa_1^ * ] + v_2 [ ( { \bf m}_l^2)_{22 } + |\kappa_2|^2 + d ' ] + v_3 [ ( { \bf m}_l^2)_{32 } + \kappa_2 \kappa_3^ * ] = - [ { \bf m}_{h_d
l_2}^2 + \mu^*\kappa_2]v_d + d_2 v_u , \label{eq : tadpoles } \\ & & v_1 [ ( { \bf m}_l^2)_{13 } + \kappa_3 \kappa_1^ * ] + v_2 [ ( { \bf m}_l^2)_{23 } + \kappa_3 \kappa_2^ * ] + v_3 [ ( { \bf m}_l^2)_{33 } + |\kappa_3|^2 + d ' ] = - [ { \bf m}_{h_d
l_3}^2 + \mu^*\kappa_3]v_d + d_3 v_u , \nonumber\end{aligned}\ ] ] where @xmath69 .
this set of three equations can be easily solved and we get @xmath70 where @xmath71 and @xmath72 can be obtained from @xmath73 by replacing the @xmath74-th column with @xmath75v_d + d_1 v_u \\ - [ { \bf m}_{h_d l_2}^2 + \mu^*\kappa_2]v_d + d_2 v_u \\ - [ { \bf m}_{h_d l_3}^2 + \mu^*\kappa_3]v_d + d_3 v_u \end{array } \right ) .\ ] ] the equations ( [ eq : mub1])([eq : tadpoles ] ) are solved subsequently until convergence and self - consistency of the results is obtained .
after this procedure we add also the dominant radiative corrections @xcite , and get back to the @xmath48 scale to obtain the mass spectrum of the model .
for the details of the mass matrices which need to be diagonalized see ref .
not all initial parameters result in an acceptable mass spectrum
. one may impose several different constraints to test the model .
the problem , however , is in the fact that the available experimental data are in most cases not confirmed in other experiments , not to mention that they are very model dependent .
therefore caution is needed before such constraints will be imposed .
first , we are going to check whether the recently suggested @xmath76 may be obtained within the described model . the only additional constraints that we have used are the mass limits for different particles , as published by the particle data group in 2010 .
they read @xcite : @xmath77 , @xmath78 , @xmath79 , @xmath80 , @xmath81 , @xmath82 , @xmath83 , @xmath84 , @xmath85 , @xmath86 , @xmath87 .
we have performed a scan over the whole parameter space given by the following ranges : @xmath88 with step of 20 gev , @xmath89 with step 5 , @xmath90 with step 100 . during this scan
we have kept @xmath91 and a fixed @xmath92 . for each point
the mass spectrum was calculated and confronted with the imposed constraints .
solutions for @xmath93 .
for each point there is an @xmath94 such that the mass of lightest higgs boson is in the region 120140 gev . ] like fig .
[ mgozdz : fig1 ] but for @xmath95 . ] like fig .
[ mgozdz : fig1 ] but for @xmath96 . ] like fig .
[ mgozdz : fig1 ] but for @xmath97 . ] like fig .
[ mgozdz : fig1 ] but for @xmath98 . ] like fig .
[ mgozdz : fig1 ] but for @xmath99 . ] like fig .
[ mgozdz : fig1 ] but for @xmath100 . ] like fig .
[ mgozdz : fig1 ] but for @xmath101 . for such high value
the model breaks down and the results are not reliable . ]
the results are presented in figs .
[ mgozdz : fig1][mgozdz : fig8 ] .
there is a separate diagram in the @xmath102 plane for each @xmath103 .
every point on these diagrams says , that for certain values of @xmath40 , @xmath42 , and @xmath104 there is an @xmath43 between 200 gev and 1000 gev for which @xmath0 is between 120 gev and 140 gev
. therefore figs .
[ mgozdz : fig1][mgozdz : fig8 ] represent the ranges of free parameters , for which the model is compatible with the newest experimental suggestions within error margins .
one sees that for small values of @xmath105 the parameter range is quite constrained . for higher values of @xmath104
the common fermion mass @xmath42 is preferred to be below roughly 600 gev .
a characteristic feature of the model is presented in fig .
[ mgozdz : fig8 ] , ie .
, that it breaks down for @xmath106 .
this is because for such high values the bottom and tau yukawa couplings tend to obtain unacceptably high values during the rge running .
similarly , too small values of @xmath107 make the top yukawa coupling explode .
.[tab : higgs1 ] list of free parameters of the model for which @xmath108 ( cf .
[ fig : param - space ] ) . here , @xmath91 and @xmath109 .
[ cols="<,<,<,<,<,<,<,<,<,<",options="header " , ] if we assume , according to the newest data , the lightest higgs boson mass to be centered around 125 gev , the allowed parameter space shrinks drastically .
first , let us allow for a 5 gev spread in the lightest higgs boson mass . on fig .
[ fig : param - space ] the points corresponding to @xmath110 are presented . as a reference
, we give the whole list in tab .
[ tab : higgs1 ] . if we narrow our field of interest to the , say , @xmath111 region only , we end up with the parameter space listed in tab . [ tab : higgs2 ] .
it is apparent , that higher values of @xmath43 and @xmath104 are favoured .
also , quite often if one of the @xmath112 s takes smaller value , it is compensated by a high value of the other .
it is worth to comment at this point at the recent constraint on supersymmetry formulated by the lhc collaborations cms and atlas @xcite .
it excludes the existence of supersymmetric particles with masses below roughly 1 tev .
however , this conclusion has been drawn for the simplest supersymmetric models in which , among others , the @xmath1-parity is conserved , and as such do not directly apply to the model discussed here .
all the calculations have been presented so far for a fixed @xmath92 parameter .
this is just the value for which the rpv effects start to appear , however , their contribution is very small . for @xmath109 and below
, the model essentially becomes @xmath1parity conserving .
on the other hand , values of @xmath47 of the order of @xmath113 have a very big impact on the results , often throwing the mass spectrum out of the allowed ranges .
there is therefore a rather modest region of the @xmath47 parameters in which the model is still physically acceptable , and at the same time the rpv contributions are not marginal .
let us now check what is the @xmath47 dependence of the @xmath0 mass for the 19 candidate points listed in tab .
[ tab : higgs2 ] .
we do not expect all of them to behave in the same way under the change of the @xmath47 parameter , especially that some of them are found in the @xmath101 region , which is in parts numerically unstable .
the results are presented in fig .
[ fig : mhl ] , where we have grouped the points according to their functional dependence on @xmath47 . on the lower left hand side pannel two point
are presented which seem to be found by accident only , and they yield the wanted lightest higgs boson mass just for the @xmath92 value . on the lower right hand side and the upper left hand side pannels we present the category of points , which converge to the correct @xmath0 value for decresing @xmath47 .
these points would be the sought solutions in the @xmath1parity conserving model , and also in the rpv case presented here for fine - tuned values of @xmath47 .
we see that deviations from the @xmath114 may be substantial for @xmath47 greater than @xmath115 , with general tendency to increase ( lower right hand side pannel ) or decrease / oscillate ( upper left hand side pannel ) .
the last , upper right hand side pannel shows three points which very weakly depend on the changing of @xmath47 .
the solutions obtained form them are stable , regardles in the rpc and rpv regime
. these points , surprisingly , also have @xmath101 .
it was interesting to check , that for the typical minimal supergravity model with broken @xmath1-parity there is a quite wide parameter space , for which the mass of the lightest higgs boson is compatible with the recent tevatron observations .
however , if one refines the constraints using the lhc7 results , the parameter space shrinks drastically to a set of points roughly given in tab .
[ tab : higgs2 ] . in this communication
we have used a very modest set of constraints on the low - energy spectrum , keeping only the most obvious ones . a more detailed analysis containing a discussion on the higgs and higssino contributions to the neutrino magnetic moment and
1-loop neutrino masses will be given elsewhere .
we may conclude that , in the way presented above , we have found 17 good candidate points in the rpv msugra model which result in a physically acceptable mass spectrum and are at the same time compatible with the newest higgs boson searches .
there are many more points , like the one numbered ( 1 ) and ( 17 ) , which would also give the desired mass spectrum , but for which a specific fine tuning of the parametes is necessary .
this project has been financed by the polish national science centre under the decission number dec-2011/01/b / st2/05932 .
also the support of the european community research infrastructure action under the fp7 `` capacities '' specific programme is greatly acknowledged .
h. baer , v. barger , p. huang , a. mustafayev , phys .
d * 84 * , 091701(r ) ( 2011 ) ; u. ellwanger , arxiv : 1112.3548 [ hep - ph ] ; o. buchmueller _ et al .
_ , arxiv : 1112.3564 [ hep - ph ] ; s. akula _
et al . _ ,
arxiv : 1112.3645 [ hep - ph ] ; m. kadastik , k. kannike , a. racioppi , m. raidal , arxiv : 1112.3647 [ hep - ph ] .
|
the first results on the searches of the higgs boson appeared this summer from the lhc and tevatron groups , and has been recently backed up by the atlas and cms experiments taking data at cern s lhc . even though the excitement that this particle has been detected is still premature , the new data constrain the mass of the lightest higgs boson @xmath0 to a very narrow 120140 gev region with a possible peak at approximately 125 gev . in this communication we shortly present the higgs sector in a minimal supergravity model with broken @xmath1-parity . imposing the constraint on @xmath0
we show that there is a relatively large set of free parameters of the model , for which that constraint is fulfilled .
we indetify also points which result in the lightest higgs boson mass being approximately 125 gev .
also the dependence on the magnitude of the @xmath1parity admixture to the model is discussed .
|
the way people are thinking about neutrino oscillations and how a neutrino factory fits into a program of measurements has changed dramatically between previous nufact workshops and this one .
although we still consider a neutrino factory the ultimate facility to do precision oscillation measurements , we are becoming more aware of what one can learn with conventional neutrino beams . understanding how much of the possible parameter space one can explore with conventional beams , and how these experiments might complement neutrino factory measurements
has kept many phenomenologists busy this past year .
the oscillation working group also has included for the first time discussions on different proposals for these super conventional neutrino beams , as well as a new idea for making a pure @xmath0 or @xmath1 beam , which can be done by accelerating radioactive ions which undergo beta decay . with these new techniques for measurements , new issues
become relevant : cross sections , beam systematics , and backgrounds , the three things that were so straightforward to tackle in the case of neutrino factory experiments . while the experimental picture is changing ,
so too is the theoretical motivation for studying neutrino oscillations . with the concept of extra dimensions becoming
more fleshed out in the past year , neutrinos have surfaced as an example of particles which have significant activity outside the brane we live on .
furthermore , given that the sno experiment has confirmed solar oscillations , and in particular , confirmed that the mixing angle describing solar neutrino oscillations is large , the prospect of seeing leptonic cp violation in the laboratory is still alive .
the see - saw mechanism remains an attractive candidate for accounting for the smallness of neutrino mass and can easily accommodate the observed large solar angle .
furthermore the see - saw mechanism allows the possibility that baryon asymmetry is generated in the early universe , but only if cp is violated in the lepton sector . in the past year
there has been much theoretical work trying to make the link between the cp violation required by baryon asymmetry and that potentially observable in low energy neutrino experiments . in this summary
we try to give an overview of what the issues are in both understanding the origin of neutrino oscillations , how to go from experimental measurements to oscillation parameters , and finally how to make the experimental measurements themselves .
perhaps the zeroth order question is whether we are actually seeing atmospheric and solar neutrino oscillations , or some other non - standard effect ?
valle @xcite showed that while the case for atmospheric oscillations looks convincing , there are alternatives to solar oscillations which fit the data well , such as spin flavor precession or flavor changing neutrino interactions . however the magnetic moments and fields , or flavor changing couplings do not appear very natural , and an oscillation signal from kamland would rule out such interpretations .
assuming we are seeing oscillations , and hence neutrino mass , then , ignoring lsnd for the moment ( discussed later ) , the atmospheric and solar data imply two possible patterns of neutrino mass , distinguished by the sign of @xmath2 , together with two large mixing angles and one small mixing angle ( also discussed later ) . from a theoretical point of view
the challenge is to account for such a neutrino spectrum .
the first point to make is that neutrino masses are zero in the standard model , so neutrino oscillations are the first solid evidence for new physics beyond the standard model .
three possible ways to extend the standard model in order to account for the neutrino mass spectrum above were discussed in this working group , namely the see - saw mechanism , extra dimensions , and r - parity violating supersymmetry . in his plenary talk ross @xcite reviewed the see - saw mechanism in which super - heavy right - handed majorana neutrinos are introduced into the standard model , which induce light effective left - handed majorana neutrino masses according to the formula @xmath3 where @xmath4 represents a typical quark or charged lepton mass and @xmath5 is the heavy right - handed neutrino mass scale .
he showed that if one of the right - handed neutrinos plays the dominant role in the see - saw mechanism then a mass hierarchy and large mixing angles may naturally emerge .
blazek @xcite applied the see - saw mechanism to gut models such as @xmath6 where three families of right - handed neutrinos with majorana masses not far below the gut scale are predicted .
gut models require supersymmetry ( susy ) in order to allow the gauge couplings to meet , and also to stabilize the hierarchy .
one of the consequences of having a see - saw mechanism with susy is lepton flavor violation ( lfv ) which was discussed extensively in this working group .
blazek @xcite showed that sizable branching ratios for @xmath7 are expected .
shimizu @xcite confirmed this and presented results for @xmath8 which also showed large branching ratios .
shimoyama @xcite showed that in a simplified model where right - handed neutrinos have universal masses the standard susy model implies branching ratios for @xmath8 which exceed the experimental limit .
finally koike @xcite studied the lfv process @xmath9 and concluded that @xmath10 represents the optimal range for experimental searches .
he also showed how it may be possible to distinguish models of new physics by measurements on several different kinds of nuclei .
the see - saw mechanism also opens up the possibility of generating the baryon asymmetry of the universe via leptogenesis .
the idea due to fukugita and yanagida @xcite is that in the early universe the super - heavy lepton number violating majorana right - handed neutrinos would have been produced , and , if cp is violated , could then decay out of equilibrium giving a lepton asymmetry which is subsequently converted into baryon asymmetry via electroweak sphalerons . the possible link between the cp violating phases necessary for leptogenesis and the cp violating phases measurable by low energy neutrino physics was also explored in this working group . working in a simple framework of two right - handed neutrinos
, morozumi @xcite discussed the possible correlation between leptogenesis and low energy phases .
ibarra @xcite discussed the case of three right - handed neutrinos , and showed that although in general there is no direct link , the low energy observable phases may contribute to leptogenesis over much of the parameter space .
neutrino masses in the framework of extra dimensions were considered by shafi @xcite , who specialized to the case of a 5d standard model with warped geometry .
the basic idea is that the standard model fermions reside in the 5d bulk , and the quark and lepton mass hierarchies can be interpreted in a geometrical way , with the higgs field localized at the tev brane , and the fermion masses being in direct proportion to the overlap of their wavefunctions with the higgs field . in such a set - up the see - saw mechanism can not be implemented because of the warped geometry , however the desired majorana masses may arise from non - renormalizable operators , and the masses again depend on the position of the left - handed leptons in the bulk .
neutrino masses in r - parity violating ( rpv ) susy were discussed by valle @xcite . in the simplest bilinear rpv scheme one
adds an explicit mass term coupling lepton doublets to a higgsino doublet .
the left - handed neutrinos are then mixed in the neutralino mass matrix , resulting in a single neutrino majorana mass term .
the other neutrino masses and mixings are generated by loop corrections , depending on the rpv susy parameters .
this results in a testable relation between neutrino masses and rpv susy physics at colliders .
the observation of atmospheric neutrinos ( see , e.g. , ref.@xcite ) and solar neutrinos ( see , e.g. , ref.@xcite ) implies that not only are neutrinos massive , but that the mass eigenstates ( @xmath11 ) with mass ( @xmath12 ) are not identical to the flavor eigenstates ( @xmath13 ) . the unitary matrix which translates from the mass to the flavor basis is usually parameterized by three mixing angles ( @xmath14 ) , and a cp violating phase ( @xmath15 ) , and is often referred to as the mnsp @xcite matrix , after maki , nakagawa , sakata , and pontecorvo .
the atmospheric data largely constrain ( @xmath16 , @xmath17 ) and the solar data constrain ( @xmath18 , @xmath19 ) in this three flavor framework of neutrino oscillations@xcite , where @xmath20 .
the constraint from the chooz result @xcite implies @xmath21 .
so in vacuum the mnsp matrix looks roughly like @xmath22 where we have used @xmath23 and @xmath24 , and @xmath25 , and @xmath26 .
choubey @xcite performed model independent and model dependent analyses of solar neutrino data including the neutral current event rate from sno @xcite .
she concluded that the large mixing angle solution gives the best fit ( @xmath27 ) and the low solution is allowed only at the @xmath28 level .
valle @xcite also gave results on his solar neutrino analyses and his results agree with those in @xcite .
while whether the lma solution is the correct one or not will be soon confirmed by the kamland experiment , it should be noticed that the goodness of fit ( gof ) for each of the other solar solutions ( low , quasi vacuum oscillation , small mixing angle solutions ) is still acceptable .
namely , according to @xcite , we have @xmath29(lma)=40.57 , @xmath29(low)=50.62 , @xmath29(qvo)=56.11 , and @xmath29(sma)=70.97 for 45 degrees of freedom .
this implies that lma is 65.99% ( 0.44@xmath30cl ) likely , low is 26.14% ( 1.12@xmath30cl ) likely , qvo is 12.39% ( 1.54@xmath30cl ) likely , and sma is 0.81% ( 2.65@xmath30cl ) likely . as far as gof is concerned , none of these solutions is excluded from a conservative point of view . or 5@xmath30 is necessary to exclude a hypothesis . ]
they are disfavored when they are compared with the lma solution , i.e. , @xmath31 = @xmath32(lma ) = 15.54 ( 3.53@xmath30cl for d.o.f . = 2 ) , @xmath31 = @xmath33(lma ) = 30.40 ( 5.16@xmath30cl for d.o.f .
= 2 ) . on the other hand
, @xcite argues that gof is a necessary but not sufficient criterion for a model to survive .
@xcite claims that the sma solution is rejected because it creates strong tension between not only total rate versus spectrum data , but also between sno data versus all the others , while the low and qvo solutions are still acceptable because the tension is much smaller for them than that for sma . with the mass hierarchy @xmath34
there are two possible mass patterns , depending on whether @xmath2 is positive ( normal hierarchy : @xmath35 ) or negative ( inverse hierarchy : @xmath36 ) .
the quantities one has to determine are @xmath37 , the sign of @xmath2 and the cp phase @xmath15 . to discuss cp violation in long baseline experiments , it is helpful to have the exact analytical expressions for the oscillation probabilities in matter .
kimura @xcite gave the expressions in a nice way .
the point of his work is that the quantity @xmath38 ( @xmath39 are the flavor indices , @xmath40=1,2,3 is the index of the mass eigen states , and no sum on @xmath40 is implied ) in matter with constant density can be expressed linearly in the quantity @xmath41 in vacuum with some coefficients which are functions of the eigenvalues @xmath42 of the @xmath43 matrix @xmath44diag(@xmath45)@xmath46+diag(@xmath47,0,0 ) , where @xmath48 .
his results are more useful than previous works ( e.g. , @xcite ) since the expression does not use a particular parametrization for @xmath49 and it is relatively easy to see the condition in which enhancement of the oscillation probability occurs .
once we know the form for @xmath50 , we can write down the the oscillation probability as @xmath51 where @xmath52 .
naively one might assume that with two general unknowns , ( @xmath53 ) , and one discrete unknown , ( the sign of @xmath2 ) , and two measurements of the oscillation probability at a neutrino factory ( i.e. @xmath54 and @xmath55 ) it would be straightforward to extract the mixing parameters .
however , since the work @xcite , it has been realized that there is in general more than one solution for the set of oscillation parameters , even if we know both @xmath56 and @xmath57 .
reference @xcite shows that there are intrinsically two sets of solutions ( @xmath37 , @xmath15 ) and ( @xmath58 , @xmath59 ) .
reference @xcite describes the degeneracy under @xmath60 , and reference @xcite found the ambiguity under @xmath61 .
thus there is an eight - fold degeneracy @xcite , and the resolution of this degeneracy was one of the main subjects at nufact02 .
the degeneracy in @xmath61 is difficult to resolve as long as we look at only channels @xmath62 , @xmath63 and @xmath64 , since the first two create the degeneracy while the third gives the information only on @xmath65 . as was pointed out in @xcite ,
the best way is probably to look at the channel @xmath66 whose oscillation probability in vacuum is given by @xmath67 , which is compared with @xmath68 . to determine @xmath37 , minakata @xcite proposed to tune the neutrino energy at the oscillation maximum ( @xmath69 ) so that the ellipse in the @xmath56 @xmath70 plane that results from letting @xmath15 run from 0 to @xmath71 collapses to a line .
then the ambiguity in @xmath37 is reduced .
whisnant @xcite discussed systematically whether or not a set of measurements in neutrino oscillation appearance experiments can completely resolve the ambiguities .
he showed that experiments with neutrinos and antineutrinos at two different energies , which give the four independent measurements , can in principle resolve all of the parameter degeneracies if @xmath72 .
mena @xcite showed that by combining a superbeam and a neutrino factory , it is possible to reduce the intrinsic , sign(@xmath73 ) and @xmath17 degeneracies .
she pointed out that the short neutrino factory baseline ( e.g. @xmath74 km ) becomes an interesting distance to combine with a superbeam experiment for values of @xmath37 near its present limit . for smaller @xmath37 , an intermediate neutrino factory baseline of @xmath75(3000 km ) is still required .
meloni @xcite showed how the @xmath76 ( `` golden '' ) and @xmath77 ( `` silver '' ) transitions observed at an opera - like 2 kton lead - emulsion detector at @xmath78 km , in combination with the @xmath54 transitions observed at a 40 kton magnetized iron detector with a baseline of @xmath79 km , strongly reduce the intrinsic @xmath80 ambiguity .
he also showed how a moderate increase in the opera - like detector mass ( 4 kton instead of 2 kton ) completely eliminates the degenerate regions even for small values of @xmath37 .
huber @xcite compared the physics potential of planned superbeams with the one of neutrino factories .
he showed that jhf
superk ( 0.75mw , 110kt@xmath81yr ) , jhf hyperk ( 4mw , 8mt@xmath81yr ) and nufact
i ( 0.75mw , 50kt@xmath81yr ) , nufact ii ( 4mw , 400kt@xmath81yr ) can explore @xmath82 down to @xmath83 , @xmath84 , @xmath85 , respectively , while only jhf hyperk and nufact ii can probe the cp phase for @xmath86several@xmath87 , @xmath88 , respectively . a previous study showed that two simultaneous baselines for a neutrino factory could be used to resolve degeneracies @xcite where something like a combination of 3000 km plus either 700 km or 7000 km baselines might be optimal .
for one neutrino factory to access two baselines , a triangular or bow - tie ring would be necessary . depending on what the first generation of neutrino superbeams measures ,
the argument for making two simultaneous baselines may be less compelling .
in very long baseline experiments , it is important to estimate the magnitude of the matter effect . unlike in shorter or medium long baseline experiments ,
in which neutrinos go through only the crust or the upper mantle of the earth , a constant density approximation is not good in very long baseline experiments . by using the method of a fourier series expansion of the matter profile ,
ota @xcite showed that neglecting the matter profile effect leads to an extra uncertainty ( @xmath89 5% ) on the average density for 5,000 km @xmath90 @xmath91 @xmath90 10,000 km .
he proposed to treat the first coefficient in the fourier series as a fitting parameter .
winter @xcite compared three different methods ( single perturbation , random fluctuations , and the measured mean ) to evaluate uncertainties in the earth s matter density profile .
he gave advantages and disadvantages for each method and he concluded that matter density uncertainties probably would not be the bottleneck of the statistical analysis of a planned experiment , while he claimed that somewhat more effort should be spent on improving the results from geophysics .
although the lsnd result @xcite is not yet confirmed , we can still try to explain the lsnd result in at least three different ways .
the first way is to assume a fourth mass eigenstate , which implies one sterile neutrino ( with a ( 2 + 2 ) or ( 3 + 1 ) mass scheme , depending on the mass pattern ) .
the lsnd signature is then simply the result of three independent mass squared differences .
valle @xcite gave the results of his recent analyses on the solar and atmospheric neutrino data for the ( 2 + 2)scheme , and it is strongly disfavored .
the ( 3 + 1)scheme is allowed only at 99%cl .
a second possibility is to give up cpt invariance @xcite and to take full advantage of the difference between neutrinos and antineutrinos .
this possibility would be excluded if kamland observes the deficit .
the third idea @xcite is to assume a lepton number violating interaction @xmath92 ( @xmath93 ) .
interestingly , this scenario predicts that the miniboone experiment will see no signal .
there are other exotic scenarios which can be tested at neutrino factories .
wong @xcite considered the situation where the equivalence principle is violated in neutrino oscillations with three flavors .
the @xmath43 matrix which describes the propagation of neutrinos is @xmath94 where @xmath95 is the relative gravitational coupling constants , violation of the equivalence principle ( vep ) implies @xmath96 , @xmath97 stands for the mixing matrix between the gravitational and flavor eigenstates , and @xmath98 is the gravitational potential . assuming @xmath99 for simplicity , she gets the following ratio of t violation @xmath100 with vep to that @xmath101 without vep @xmath102 where @xmath103 is the mixing angle between the first and the third eigenstates in @xmath97 and other mixings in @xmath97 are assumed to be zero for simplicity .
she showed that t violation @xmath100 with vep is deviated from @xmath101 without vep if the violation @xmath104 is large and therefore may be measured in long baseline experiments .
sato @xcite discussed the measurement of new physics which is described by an interaction such as @xmath105 ( @xmath106 ) in long baseline neutrino oscillation experiments .
through the neutrino oscillation , the probability to detect the new physics effects such as flavor violation is enhanced by the interference with the weak interaction . assuming a neutrino factory and an upgraded conventional beam
, he examined the possibility to observe new physics numerically and pointed out that it is possible to search for new interactions using some channels , for example @xmath107 , in these experiments .
he also showed that there is a chance to observe a flavor - changing effect in @xmath108 appearance experiments such as icarus and opera .
campanelli @xcite discussed a new flavor changing interaction which gives the contribution @xmath109 ( @xmath110 ) to the @xmath111 matrix in the flavor basis .
he showed that the new physics effect becomes important in the channels @xmath112 ( @xmath113 ) at sufficiently high neutrino energies , since the contribution from the conventional physics is suppressed by @xmath37 and @xmath114 .
therefore it may be possible to observe the new physics effect particularly by looking at the energy spectrum which would be different in the presence of new physics .
it is clear from the discussion above that to measure the oscillation parameters and to probe exotic scenarios , we will need more than one measurement of oscillations to understand phenomena visible at the atmospheric @xmath115 .
furthermore , there is much than can be learned at high intensity , conventional ( that is , meson decay ) beams .
if a signal is seen in a conventional beam , that information may still be an important constraint even in the advent of a neutrino factorybecause of the different energies of the beams .
the challenge with doing any of the measurements described above in conventional neutrino beams is that the @xmath116 transition probability is limited by the chooz non - observation of @xmath117 disappearance to be small , less than @xmath118 in the absence of enhancement from cp violation or matter effects .
intrinsic @xmath0 backgrounds in conventional beams at the per cent level , due primarily to the decay chain @xmath119 , and significant background processes present if neutrino energies are above a few hundred mev from neutral current events mimicking electron signals will make these measurements difficult . for a given detector mass ,
one gets the best sensitivity for seeing a non - zero value of the muon to electron neutrino transition probability by placing an experiment near or slightly below the first oscillation maximum ( @xmath120 ) . in order to see evidence for matter effects , and
thereby deduce the mass hierarchy , however , one needs to go to baselines on the order of 1000 km , and therefore use neutrino energies well above 1gev . finally , to look for cp violation ,
one might still run at the first oscillation peak , as described by minakata@xcite , but it has also been argued that one should consider running an experiment at the second or third oscillation maximum @xcite , since the cp - violating contribution to the transition grows with @xmath121 . to get around the problem of backgrounds ,
several different strategies have been proposed .
for example , the jhf to superk , the numi off - axis , and the cngt proposals all take advantage of the two - body decay kinematics of the pion , and put the far detector slightly off the main axis of the neutrino beamline @xcite .
this trick , first suggested in bnl proposal e889 @xcite , allows one to maximize the @xmath117 flux in a very narrow energy region , reduces the flux at higher neutrino energies , and does not alter the flux from three - body decays , which are the source of intrinsic @xmath0 s in the beam .
this results in an improved signal to background ratio at the peak beam energy since the background from high @xmath122 neutral current interactions that fake electron charged current events at this peak energy is reduced .
however , since the @xmath117 spectrum is significantly narrower than the shape of the appearance probability , one suffers from statistics .
this must be mitigated by the use of massive detectors and very high proton beam power .
another way to reduce the effect of backgrounds is to consider a distance of thousands of kilometers where the matter effects will significantly amplify either the neutrino or the antineutrino probability , so that even for a small value of @xmath123 a large probability would be expected .
this is the strategy of the brookhaven to nusl proposal , as described by s.kahn @xcite .
however , to go such large distances , even with a wide band beam , requires a `` megaton class '' detector . by combining a large distance and a broad band neutrino energy spectrum ,
this proposal can see not only the first but the second oscillation maximum , and as such can have more constraints on the oscillation parameters .
a third way to reduce the backgrounds inherent in a conventional beam is to go to neutrino energies of a few hundred mev and to construct only a short decay tunnel so that few muons decay to produce @xmath0 s , as described by m.mezzetto .
the intrinsic @xmath0 contamination of the proposed cern spl beam is at the parts per mil level , and at this neutrino energy the few neutral pions that do get produced are easy to distinguish from electrons since there is a large angle between the two decay photons .
however in the few hundred mev energy range , the cross - section varies quadratically with energy , suppressing the rate , so again a megaton - class detector must be used .
finally , a novel technique for creating a beam with extremely low backgrounds is to accelerate a beam of radioactive ions in a long storage ring , and let them decay to electron neutrinos .
the intrinsic backgrounds here are just as low as that of a neutrino factory , and in fact the beam is really only @xmath0 or @xmath1 depending on the ion in the ring , with contaminations from any other flavor at less than the @xmath88 level
. the resulting neutrino beam would be at a very low neutrino energy , and a water qerenkov detector located in the frejus tunnel could be used for both this `` beta beam '' concept and the cern spl neutrino beam .
this concept was presented in the plenary session by p. zucchelli and in a parallel session by m. mezzetto .
table [ tab : exps ] gives a summary of the different neutrino experiments that are being considered worldwide , and the different strategies being adopted .
the experiments listed below are at very different stages of completion , as well : the jhf to superkamiokande uses an existing @xmath124 kton detector , but will require a new beamline to be built at the japanese hadron facility .
the beamline is being designed not only to handle both a first stage experiment with a 1mw proton source , but also an upgraded stage with a 4mw proton source , and a megaton - class water qerenkov detector @xcite .
the numi off - axis experiment would use the ( almost ) existing numi beamline , but a new @xmath125 kton detector would need to be built somewhere in northern minnesota or perhaps even in canada @xcite .
the cngt project would require instrumenting a large body of water , perhaps in the gulf of taranto for a detector , and the cngs beamline would have to be tuned at a lower energy which would preclude measuring @xmath108 appearance @xcite at gran sasso . finally , the other programs ( cern spl and the bnl proposal ) require both new beamlines and new detectors @xcite .
[ tab : exps ] .vital characteristics of future possible neutrino experiments . @xmath126
proton power of cngt assumed to be approximately that of cngs .
speakers in this workshop were : @xmath127a .
ichikawa , @xmath128s .
kopp , @xmath129s .
kahn , @xmath130 m .
mezzetto,@xmath131b .
autin , @xmath132s . geer . [ cols="<,^,^,^,^,^,^,^ " , ] for most of the proposals listed in table [ tab : exps ] , the detector of choice is a water qerenkov device either the superkamiokande detector , or a new larger version , which would also presumably be instrumented well enough to search for proton decay and supernovae neutrinos . at energies at or below @xmath133 gev ,
the superkamiokande detector is able to keep detector misidentification at a level comparable to or below the intrinsic electron neutrino backgrounds in the jhf off - axis beamline . for a neutrino factory ,
the energies are @xmath134s of gev , and because charge identification is required , there is an appropriate and proven candidate detector technology , a steel - scintillator detector like the one being built for the minos experiment @xcite . for the @xmath135 measurements that were described by sato and meloni , an opera - like detector , even one that was twice the mass of the current design , could provide very interesting measurements .
a detector technology which in principle could work for any of the conventional beamlines , is a liquid argon tpc , such as that being built by the icarus collaboration .
its very precise imaging capabilities would mean very good background rejection , while maintaining high efficiency , as described by d. cline . because the backgrounds from neutral current events can be eliminated , and the signal efficiency remains high
, the reach per kton of a liquid argon detector can be as much as 3 times that of other technologies .
therefore , one would require in principle a much smaller detector mass than those listed in the table above .
although using the 600 ton module size of icarus would be prohibitively expensive even for a 6kton fiducial mass , k. mcdonald showed that for large enough single vessels , and large enough wire plane separation , this may become a very economical technology .
certainly the technology to make extremely large ( order 100kton ) cryogenic vessels is well - understood thanks to the natural gas companies , and the biggest challenge will likely be elsewhere . for the numi off axis beamline ,
another detector technology being considered is fine - grained calorimetry , as described in a talk by a. para .
fine grained calorimeters have a long history in the field of neutrino physics , and are well - understood and well - simulated . by making a low - z calorimeter
which was segmented every third of a radiation length , one could build a detector with reasonably high efficiencies , which would be able to reduce the neutral current background in the numi off - axis beamline to below that of the intrinsic @xmath0 contamination , which is itself about @xmath136 under the peak energy .
the challenge here will be to design a device that is economical to construct and realizes these large background reductions .
as can be seen by the large number of proposals being considered , it is clear that one or more `` neutrino superbeams '' are likely to provide measurements before the advent of a neutrino factory .
what s more , these neutrino superbeams are likely to be between a few hundred mev region , such as the cern spl proposal , to the few ( up to 6 ? ) gev region as in the bnl proposal . the experimental challenges associated with measurements with these beamlines are extremely different from those associated with those from a neutrino factory . in the absence of oscillations , the number of events at a detector is simply the product of the neutrino flux times the cross section times the detector efficiency .
uncertainties in any of these can degrade the ability to use precise measurements of multiple transition probabilities to extract oscillation parameters , and in fact all of them may have complicated energy dependence , in particular in the sub to few gev region . to address these issues , we convened a joint session with the neutrino scattering working group . at very low energies ,
the cross - section is dominated by elastic or quasi - elastic processes which are few body reactions and relatively easy to model , at least in a free nucleon target . at very high energies ,
the cross - section is dominated by deep inelastic scattering , where complex multi - body final states can be effectively modeled with small corrections to a dominant picture of neutrino - quark scattering with an outgoing `` jet '' created by recoil hadronic system .
however , the neutrino energies are on the order of the nucleon mass the second portion of the cross - section becomes dominated by intermediate resonant hadronic states with discrete masses , which complicate the picture significantly .
in addition , the few body nature of the final state requires modeling exclusive reactions in order to correctly predict signal and background efficiencies .
quasi - elastic cross - sections can be predicted very accurately , although the effects of final - state interactions in nuclei give @xmath134@xmath137% uncertainties in observed rates .
deep inelastic cross - sections are known to a few percent in the high energy limit . however , in the `` resonance '' region the rates for exclusive final states typically have @xmath124@xmath138% uncertainties , and often differential distributions are completely unmeasured as discussed by casper at this workshop , and as shown in reference @xcite .
bodek has been looking at a way to extend deep inelastic scattering predictions for cross sections to the resonance regime using global fits of many different data sets , and has shown that for charged lepton scattering , this treatment results in an accurate prediction of the mean cross section in the resonance region . by then using the data from charged leptons in the resonance region ,
it is hoped that this treatment can result in correct modeling of resonances in neutrino scattering at low energies @xcite .
to first order , flux predictions will be made in these upcoming superbeam experiments by simply measuring the neutrino event distributions as a function of energy in a near detector and then by using a beamline simulation to extrapolate from near to far .
however , this is complicated by the fact that the neutrino source is usually so close to the near detector that the source does not appear point - like . also , if the neutrinos are from 3-body or non - pion decays , ( i.e. the @xmath0 s , which are from three - body decays of the muon and kaon ) the near - far extrapolation is different .
several different strategies for making flux predictions were discussed in this workshop .
one very straightforward strategy in principle is to simply measure directly ( or indirectly ) the distributions of the secondaries that are creating the tertiary neutrinos in the first place .
kobayashi discussed the data from the pion monitor used by k2k , which gives the angular distribution of pions at a given radius .
the pion angular distribution , combined with the near detector measurements , gives a systematic uncertainty on the flux extrapolation to the far detector of 5% @xcite .
coney discussed how miniboone would be using muon counters located at the end of a small pipe set at a large angle with respect to the decay pipe at miniboone .
most of the muons which were accepted at the end of that pipe are products of high angle kaon decays , and by measuring the energies of those muons , one could get a good constraint ( estimated uncertainty goal of 10% ) on the ratio of pion to kaon decays in the beamline .
this constraint is important for limiting the uncertainty in the kaon - induced @xmath0 background in miniboone .
another strategy for extrapolating from the near to the far detector was outlined by szleper , using as an example the both the on and off - axis fluxes expected at the numi beamline .
he considered , among other effects , the uncertainties in near to far extrapolation due to hadron production uncertainties . because the bulk of the @xmath117 s expected in both the near and far detectors are from pion decays , and the low energy @xmath0 s from muon decays , where the muons were
also from pion decays , szleper showed how there is a straightforward correlation between what is measured in a near on - axis detector and far detectors , both on and off axis .
the standard technique is simply to take the near detector energy spectrum and use the beamline simulation to create a `` far over near ratio '' for each energy bin , and then multiply by that ratio . the new technique developed by szleper and para ,
however , involves a `` matrix approach '' where for every neutrino event at the near on - axis detector there is a spectrum of events at the far detector , and those events are not be all reconstructed with the same energy as the event in the near detector @xcite . for the @xmath117 s coming from the focused pions ,
the two techniques result in similar hadron production uncertainties of 1 - 2% , but for neutrinos for higher energy pions , those just underfocused , the differences are reduced from 15% ( using the traditional approach ) to 5% .
para offered a wager of a bottle of good sake during the conference that if k2k were to use this technique their far detector flux would be predicted to 1% , rather than to the 5% uncertainty currently assigned @xcite .
the error has yet to be reduced and no sake has yet changed hands . of course at a muon storage ring ,
the issues of flux prediction are quite different because the beam and decay kinematics as well known .
as evidence of this point , broncano showed that the @xmath139 corrections to the neutrino distributions were below the @xmath140 level @xcite .
therefore , it is likely that a neutrino factory will be in the enviable position of having the primary challenge to knowing the flux be knowledge of the phase space distribution and number of muons in the beam . near detectors play a very different role depending on whether or not one is discussing disappearance or appearance measurements . in the case of disappearance measurements , where it is assumed
the oscillation probability will be large and the backgrounds to the signal small , the near detector is used to make an accurate prediction of the @xmath117 flux . in particular for a wide band neutrino beam , the energy dependence of both the flux and the detector response must be well understood . for appearance measurements ,
however , one wants to use a near detector to measure the backgrounds , and in particular how the backgrounds enter as a function of the neutrino energy , and the measured ( visible ) energy . for both kinds of measurements
there are conflicting requirements for the near detector : one would like a detector that resembles the far detector as much as possible , but at the same time one would like a near detector that could make detailed studies of neutrino interactions in order to best model the far detector response .
one solution to this near detector problem is to have a suite of near detectors , as has been done very successfully by the k2k experiment , reported by yoshida .
while they have a 1kton water qerenkov device for a near to far detector ratio where the cross sections cancel , they can also compare events in the near water qerenkov detector with those seen in their fine - grained scintillating fiber detector .
the fine - grained detector can see the recoil proton in the neutrino interaction , and k2k can use this to deduce the energy resolution and acceptance of their far detector . furthermore , there is also a large muon chamber which can see the profile of the neutrino beam and verify that indeed the beam is pointing towards the far detector .
the success of this near detector suite is evidenced by the fact that very similar plans are in place for an off - axis near detector suite for the jhf to superkamiokande experiment , as described by kajita .
there will be near detectors at 280 m to study cross sections and verify the position of the neutrino beam , and a `` less near '' detector at 2 km consisting of a water qerenkov device to make the most accurate far / near prediction @xcite .
because of the difference in energies and detector technologies , the minos experiment can use one near detector which will be used to not only measure the flux and event rates ( using both quasielastic and non - quasielastic events , respectively ) , but which can also verify the direction of the neutrino beam ( during special high energy runs ) . as described by morfin
, minos is also considering adding a specially - modified cross section detector in front of the near detector , which can study neutrino interactions more precisely , and also can study the z dependence of neutrino cross sections by having planes of various materials : carbon , steel , and tungsten @xcite . for a numi
off - axis experiment , however , one would like to study the detector - related backgrounds with an off - axis detector .
this is because the on - axis beam spectrum is very different from the off - axis spectrum , and also because the visible energy in the background events is often far from the initial neutrino energy .
one definite requirement for an off - axis experiment would be an off - axis near detector which mimics a far detector , but there have also been thoughts about also including a more fine - grained near detector , ` a la k2k .
bodek described a possible new detector for cross section studies that were relevant for a numi off - axis experiment , as well as a possible site for such a detector in the numi beamline @xcite .
from this workshop , it is clear that conventional neutrino beams will occupy the foreground of oscillation measurements until perhaps 2010 .
then , a crucial decision will have to be made in choosing between hyper conventional experiments on one hand combining multi megawatt proton beams combined with megaton detectors , and neutrino factories on the other hand whose design should be mature by that time .
99 t. kajita and y. totsuka , rev . mod .
* 73 * ( 2001 ) 85 .
j. n. bahcall , phys .
* 333 * ( 2000 ) 47 .
z. maki , m. nakagawa and s. sakata , prog .
* 28 * ( 1962 ) 870 .
b. pontecorvo , sov .
jetp * 26 * ( 1968 ) 984 [ zh . eksp .
* 53 * ( 1968 ) 1717 ] .
k. hagiwara _ et al .
_ [ particle data group collaboration ] , phys . rev .
d * 66 * ( 2002 ) 010001 .
m. apollonio _ et al . _
[ chooz collaboration ] , phys .
b * 466 * ( 1999 ) 415 [ arxiv : hep - ex/9907037 ] .
s. choubey , in these proceedings .
q. r. ahmad _ et al .
_ [ sno collaboration ] , phys . rev . lett .
* 89 * ( 2002 ) 011302 [ arxiv : nucl - ex/0204009 ] .
r. m. barnett _ et al . _
[ particle data group collaboration ] , phys .
d * 54 * ( 1996 ) 1 .
g. l. fogli , e. lisi , a. marrone , d. montanino and a. palazzo , phys .
d * 66 * ( 2002 ) 053010 [ arxiv : hep - ph/0206162 ] , and private communication .
k. kimura , in these proceedings .
h. w. zaglauer and k. h. schwarzer , z. phys .
c * 40 * ( 1988 ) 273 .
j. burguet - castell , m. b. gavela , j. j. gomez - cadenas , p. hernandez and o. mena , nucl .
b * 608 * ( 2001 ) 301 [ arxiv : hep - ph/0103258 ] .
h. minakata and h. nunokawa , jhep * 0110 * ( 2001 ) 001 [ arxiv : hep - ph/0108085 ] . g. l. fogli and e. lisi , phys . rev .
d * 54 * ( 1996 ) 3667 [ arxiv : hep - ph/9604415 ] .
v. barger , d. marfatia and k. whisnant , phys .
d * 65 * ( 2002 ) 073023 [ arxiv : hep - ph/0112119 ] .
v. d. barger , d. marfatia and k. whisnant , arxiv : hep - ph/0108090 .
k. whisnant , in these proceedings .
b. jacobsson , t. ohlsson , h. snellman and w. winter , in these proceedings [ arxiv : hep - ph/0209147 ] . c. athanassopoulos
_ et al . _
[ lsnd collaboration ] , phys .
* 77 * ( 1996 ) 3082 [ arxiv : nucl - ex/9605003 ] . h. murayama and t. yanagida , phys .
b * 520 * ( 2001 ) 263 [ arxiv : hep - ph/0010178 ] .
babu , in these proceedings .
w. j. marciano , arxiv : hep - ph/0108181 . y. itow _ et al .
_ , arxiv : hep - ex/0106019 . f. dydak , expression of interest in an experiment at the cngs to measure @xmath37 , ` http://home.cern.ch/dydak/oscexp.ps ` ( april 2002 ) .
a. cervera , a. donini , m. b. gavela , j. j. gomez cadenas , p. hernandez , o. mena and s. rigolin , nucl .
b * 579 * ( 2000 ) 17 [ erratum - ibid .
b * 593 * ( 2001 ) 731 ] [ arxiv : hep - ph/0002108 ] .
h. j. grabosch _ et al . _
[ skat collaboration ] , z. phys .
c * 41 * ( 1989 ) 527 .
a. bodek , in these proceedings .
|
issues on the physics of , beamlines for , and detectors of neutrino oscillation discussed in working group 2 at nufact02 are summarized .
|
topological properties of fermion band theory have dominated research in condensed matter physics and other areas over the past decade or so @xcite . recently , it has been shown that the magnon bulk bands of heisenberg ( anti)ferromagnet on the honeycomb lattice exhibit dirac points at the corners of the brillouin zone ( bz ) @xcite .
the low - energy hamiltonian near these points realizes a massless 2d dirac - like hamiltonian with dirac nodes at nonzero energy
. this system preserves pseudo spin time - reversal ( @xmath0 ) symmetry .
it was also shown that the dirac points are robust against magnon - magnon interactions and any perturbation that preserves the pseudo spin @xmath0-symmetry of the bogoluibov hamiltonian . in this paper , we provide evidence of non - trivial topology ( magnon edge states ) in the magnon bulk bands of heisenberg ( anti)ferromagnet and xy model on the honeycomb lattice , when a gap opens at the dirac points . we show that the simplest practical way to open a gap at the dirac points is by breaking the inversion symmetry of the lattice , which subsequently breaks the pseudo spin @xmath0-symmetry of the bogoliubov hamiltonian . we show that this can be achieved by introducing a next - nearest neighbour dzyaloshinskii - moriya ( dm ) interaction .
the opening of a gap at the dirac points leads to magnon edge states reminiscent of haldane model in electronic systems @xcite . in the case of xy model
, we observe the same topological effects with magnon edge states propagating in the vicinity of the magnon bulk gap .
remarkably , the resulting hamiltonian for the xy model maps to interacting hardcore bosons .
therefore , these magnon edge states can be simulated numerically .
as magnons are uncharged particles , noninteracting topological magnons can propagate for a long time without dissipation , thus they are considered as a good candidate for magnon spintronics @xcite .
the topological properties of these dirac magnons are not just analogues of fermion band theory .
they are called topological magnon insulators @xcite and has been recently observed on the kagome magnet cu(1 - 3 , bdc)@xcite . for the honeycomb lattice , they can actually be searched for in many accessible quantum magnets such as na@xmath1cu@xmath2sbo@xmath3 @xcite and @xmath4-cu@xmath2v@xmath2o@xmath5 @xcite which are spin-@xmath6 heisenberg antiferromagnetic materials with a honeycomb structure .
they can also be investigated in ultra - cold atoms trapped in honeycomb optical lattices as the bosonic tight binding model is analogous to that of haldane model , which has been realized experimentally in optical fermionic lattice @xcite . in 3d quantum magnets ,
dirac and weyl points are possible in the magnon excitations .
recently , weyl points have been investigated in quantum magnets using 3d kitaev fermionic model @xcite .
weyl points were recently shown to occur in the magnon excitations of breathing pyrochlore lattice antiferromagnet @xcite . in this case , the criteria that give rise to them seem to be unknown unlike in fermionic systems . here , we show that the resulting bogoliubov hamiltonian has the form of weyl hamiltonian in electronic systems and that the weyl points should break the pseudo spin @xmath0-symmetry by expanding the bogoliubov hamiltonian near the non - degenerate dispersive band - touching points and projecting onto the bands . hence , the criterion for weyl nodes to exist in electronic systems also applies to magnons .
in 2d quantum spin magnetic materials , non - degenerate band - touching points or dirac points require at least two energy branches of the magnon excitations . therefore , ordered quantum magnets that can be treated with one sublattice are devoid of dirac points .
the simplest two - band model that exhibits dirac nodes is the heisenberg ferromagnet or antiferromagnet on the honeycomb lattice @xcite .
the hamiltonian is governed by @xmath7 where @xmath8 depends on the bonds along the nearest neighbours . as mentioned above , eq .
[ hhh1 ] describes several realistic compounds @xcite . for simplicity
we take @xmath9 . the ground state of eq .
[ hhh1 ] is a ferromagnet with two - sublattice structure on the honeycomb lattice ; see fig .
[ honey_comb ] .
this is equivalent to heisenberg antiferromagnet by flipping the spins on one sublattice . in many cases of physical interest ,
magnon excitations are studied by linear spin wave theory via the standard linearize holstein primakoff ( hp ) transformation .
this is an approximation valid at low - temperature when few magnons are excited . and @xmath10.,width=384 ]
the momentum space hamiltonian is given by @xmath11 , where @xmath12 and the mean field energy has been dropped . in this model , the bogoliubov quasiparticle operators are the same as the bosonic operators .
the bogoliubov hamiltonian is given by @xmath13 where @xmath14 is an identity @xmath15 matrix , and @xmath16 are pauli matrices acting on the sublattices ; @xmath17 is the coordination number of the lattice and @xmath18 .
the structure factor @xmath19 is complex given by @xmath20 where @xmath21 are the three nearest neighbour vectors on the honeycomb lattice , @xmath22 , @xmath23 and @xmath24 .
the eigenvalues of eq .
[ honn ] are given by @xmath25 the energy bands have dirac nodes at the corners of the bz reminiscent of graphene model .
in contrast to graphene , the dirac nodes occur with a nonzero energy @xmath26 as shown in fig .
in addition to the dirac nodes , there is a zero energy mode in the lower band , which corresponds to a goldstone mode due to the spontaneous symmetry breaking of su(2 ) rotational symmetry of the quantum spin hamiltonian .
as many physical systems are anisotropic , it is important to note that with spatial anisotropy @xmath27 , several dirac points can be obtained by tuning the anisotropy in each bond .
the density of states per unit cell as a function of energy is shown in fig .
[ dos ] for the two energy bands in eq .
the interesting properties of this system are manifested near the dirac points .
there are only two inequivalent dirac points located at @xmath28 as shown in fig .
[ honey_comb ] . in the case of heisenberg antiferromagnet ,
only a single dirac point occurs at @xmath29 @xcite .
expanding eq .
[ honn ] near @xmath30 we obtain a linearized model @xmath31 where @xmath32 , @xmath33 , and @xmath34 describes states at @xmath30 .
thus , the low - energy excitation spectrum near the dirac points is similar to the bloch hamiltonian of graphene model .
let us now compute the specific heat at a constant volume given by c_v=_v , where @xmath35 is the internal energy density of the system given by u=_= _ ( q)n_b[_(q ) ] , @xmath36 , @xmath37=[e^{\beta \xi_\lambda(\bold q)}-1]^{-1}$ ] is the bose function , @xmath38 is the inverse temperature , and @xmath39 is the chemical potential .
the specific heat can be integrated exactly by turning the chemical potential at the dirac nodes @xmath40 . in this case
we find @xmath41 where @xmath42 .
this is similar to the @xmath43-law found in graphene .
a very crucial point is the role of time - reversal symmetry .
since the excitations of quantum magnets are usually described in terms of the hp bosons , an ordered state must be assumed .
hence , the system must contain an even number of half integral spins with @xmath44 , where @xmath45 is even .
thus , magnons behave like bosons .
however , in the pseudo spin space @xmath46-operator can be defined for the bogoliubov hamiltonian , @xmath47 where @xmath48 denotes complex conjugation and @xmath49 .
this pseudo spin symmetry is preserved provided dirac points exist in the bz .
topological magnon insulators are the analogues of topological insulators in electronic systems .
they are characterized by the existence of edge state modes when a gap opens at the dirac points . for the honeycomb ferromagnets ,
a next - nearest neighbour interaction of the form @xmath50 ( @xmath51 ) only shifts the positions of the dirac points as it contributes a term of the form @xmath52 , where @xmath53 , and @xmath54 .
the next - nearest neighbour vectors are @xmath55 , @xmath56 , @xmath57 .
the simplest realistic way to open a gap at the dirac points is by breaking the inversion symmetry of the lattice , which in turn breaks the @xmath46-symmetry of the bogoliubov hamiltonian .
this can be achieved by introducing a next - nearest neighbour dm interaction @xmath58 where @xmath59 is the dm interaction between sites @xmath60 and @xmath61 .
the total hamiltonian of a honeycomb ferromagnetic insulator can be written as @xmath62 in the hp bosonic mapping , we obtain @xmath63 where @xmath64 , @xmath65 , @xmath66 , and @xmath67 is the coordination number of the nnn sites . we have assumed a dm interaction along the @xmath68-axis .
the phase factor @xmath69 , where @xmath70 is a magnetic flux generated by the dm interaction on the nnn sites , similar to the haldane model with @xmath71 as in electronic systems . the total flux enclosed in a unit cell vanishes as depicted in fig .
[ unit_cell_1 ] .
in contrast to electronic systems , the phase factor @xmath72 depends on the parameters of the hamiltonian .
label sites on the triangular plaquettes and give rise to a nonzero spin chirality @xmath73.,width=192 ] the bogoliubov hamiltonian is given by @xmath74 where @xmath75 , @xmath76 , @xmath77 , and @xmath78 , where @xmath79 .
expanding near the dirac points we generate a gap at @xmath80 and the full hamiltonian for @xmath81 ( @xmath82 in this case ) becomes @xmath83 where @xmath84 .
this model can be regarded as the bosonic analogue of haldane model in electronic systems @xcite . in fermionic systems
, there is a topological invariant quantity which is quantized when the fermi energy lies between the gap , such that the lower band is occupied . in the bosonic model
, there is no fermi energy and not all states are occupied .
the bosons can condense at the goldstone mode in the lower band .
however , the topological invariant quantity is , in principle , independent of the statistical property of the particles .
it merely predicts edge states in the vicinity of the bulk gap . in the magnon excitations
the chern number @xmath85 simply predicts a pair of counter - propagating magnon edge states in the vicinity of the bulk gap as shown in figs .
[ edge ] and [ edge1 ] for @xmath82 and @xmath86 respectively .
hence , the heisenberg ( anti)ferromagnet on the honeycomb lattice realizes topological magnon insulator with magnon edge states propagating at the edge of the sample as depicted in fig .
[ edge2 ] . as mentioned above , the propagation of magnon edge states differs from those in electronic systems .
therefore , they a useful in many technological devices and magnon spintronics . besides
, they can be accessible in many accessible quantum magnetic systems .
dirac points occur in a variety of quantum honeycomb ferromagnetic insulators .
let us consider the xy model on the honeycomb lattice @xmath87 the ground state of this model is an ordered ferromagnet or nel state in the @xmath88 plane .
choosing @xmath89 quantization axis , the momentum space hamiltonian in linear spin wave theory is generally written as @xmath90 where @xmath91 label the sublattices .
equation [ main ] can be written as h=_0 + _ ^_()_+const . , [ hp]where @xmath92 , @xmath93 , and @xmath45 is the number of sublattice , @xmath94 , and ( ) = _ 0 +
_ + + _ - , where @xmath95 and @xmath96 are @xmath97 matrices .
this hamiltonian is diagonalized by a matrix @xmath98 via the transformation @xmath99 , which satisfies the relation ^ ( ) = ( ) ; ^= , with @xmath100 .
@xmath101 contains the bogoliubov operators @xmath102 and @xmath103 and @xmath104 is the eigenvalues .
this is equivalent to saying that we need to diagonalize a non - hermitian bogoliubov hamiltonian @xmath105 , where @xmath106 and @xmath107/2 $ ] .
the eigenvalues of @xmath108 are given by @xmath109 $ ] , where _ ( ) = [ posi],@xmath110 and @xmath111 are the eigenvalues of @xmath95 and @xmath96 respectively .
for the xy model @xmath108 is given by @xmath112 , \label{ham3}\end{aligned}\ ] ] with @xmath113 and @xmath114 .
the positive eigenvalues [ eq . [ posi ] ] are given by @xmath115 the magnon excitations exhibit dirac nodes at @xmath116 with an energy of @xmath117 .
to generate a gap , we follow the same approach above . for simplicity , we ignore an external magnetic field and a nnn isotropic interaction @xmath118 .
hence , the dm interaction that would open a gap has to be parallel to the @xmath119-quantization axis , @xmath120 in the hp bosonic mapping , this corresponds to a magnetic flux of @xmath82 . in the @xmath89 quantization axis , @xmath121 and @xmath122 are off - diagonals .
the momentum space hamiltonian is given by @xmath123 the positive eigenvalues of the full hamiltonian are given by @xmath124^{1/2 } , \label{ham6}\end{aligned}\ ] ] at @xmath116 , a gap of magnitude @xmath125 is generated as shown in fig .
[ xyband ] .
similar to the heisenberg model , there exist magnon edge states in the vicinity of the bulk gap as depicted in fig .
[ hcb ] .
surprisingly , eqs .
[ ham0 ] and [ ham4 ] actually map to interacting hardcore bosons on the honeycomb lattice via the matsubara - matsuda transformation @xcite @xmath126 , where @xmath127 .
the hardcore boson hamiltonian is given by @xmath128 , \label{ham6}\end{aligned}\ ] ] where @xmath129 , @xmath130 , and @xmath82 .
this model [ eq .
[ ham6 ] ] offers a physical realization of these magnon edge states using ultracold atoms trapped in honeycomb optical lattices .
unfortunately , this model can not be simulated by quantum monte carlo ( qmc ) methods due to a sign problem .
however , it is amenable to other numerical simulations such as exact diagonalization .
introducing a magnetic field introduces additional phases into the system .
for instance , a magnetic field along the @xmath68-axis introduces superfluid and mott insulating phases , whereas a staggered magnetic field introduces a charge - density wave insulator . in this case
, the corresponding hard - core boson model can be written as @xmath131 where @xmath39 is the chemical potential and @xmath132 is a staggered onsite potential on sublattice @xmath133 and @xmath134 .
they correspond to a ( staggered ) magnetic field along the @xmath68-axis in the spin language . for @xmath135 , eq . [ hamm6 ] is amenable to qmc simulation as recently shown @xcite .
also recently , we have complemented the qmc results using the method presented in this paper @xcite .
hence , the hp spin wave method offers a simple approach to capture the topological properties of bosonic models that can not be simulated by qmc .
finally , we address the weyl magnons in 3d lattices . recently
, weyl points were observed on breathing pyrochlore lattice governed by @xcite @xmath136 where @xmath137 and @xmath138 are the exchange couplings between the nearest - neighbour spins on the up - pointing and down - pointing tetrahedra respectively ( see ref .
@xcite ) , and @xmath139 is a single - ion anisotropy .
it can be easy - axis ( @xmath140 ) or easy - plane ( @xmath141 ) . in the former case
, the spins would prefer the @xmath68-axis ; whereas in the latter case the @xmath88 plane is auspicious .
although a comprehensive analysis of this model has been studied in ref .
@xcite , the criteria for the existence of weyl magnons were not mentioned and understood properly . here , we argue that breaking of pseudo spin @xmath0-symmetry is a condition for weyl points to exist in quantum magnetic systems .
the linear spin wave theory hamiltonian derived in ref .
@xcite has the general form given in eq .
now , the bogoliubov hamiltonian can be cast into the form of eq .
[ bogoliubovb ] . from this equation
, we see that the momentum space hamiltonian resembles that of electronic systems .
in addition to weyl nodes obtained along the bz paths for @xmath141 and @xmath142 @xcite , there is additional non - degenerate band - touching points at the corners of the bz .
the system should realize dirac hamiltonian at the corners of the bz as shown above and also a weyl hamiltonian near the weyl points . to check whether pseudo spin @xmath46-symmetry is preserved or broken at the dirac or weyl points respectively , one should follow the approach outlined above . basically , one has to expand @xmath95 and @xmath143 near the band - touching points and project the resulting hamiltonian onto the bands . in principle , the bogoliubov hamiltonian [ eq . [ bogoliubovb ] ] near the weyl points should break @xmath46-symmetry .
therefore , one recovers the usual criteria for weyl semimetals @xcite .
in contrast to 2d systems , a gap is not needed to observe edge states in 3d weyl magnons .
edge states exist in 3d weyl magnons provided the momentum lies between the weyl nodes @xcite .
in summary , we have shown that physical realistic models of honeycomb quantum spin magnets exhibit nontrivial topology in the magnon excitations . in 2d
ordered honeycomb quantum magnets , we showed that the non - degenerate band - touching points ( at the corners of the brillouin zone ) in the magnon excitation spectrum realize a massless dirac hamiltonian .
opening of a gap at the dirac points requires the breaking of inversion symmetry of the lattice .
this leads to nontrivial topological magnon insulator with magnon edge states propagating on the edges of the material , similar to topological insulators in electronic systems .
these magnon edge states also manifest in hardcore bosons on honeycomb lattice . the hardcore boson model proposed in eqs .
[ ham6 ] and [ hamm6 ] should be studied by numerical approach to further substantiate the existence of magnon edge modes in this system .
since there are many physical 2d honeycomb quantum magnetic materials in nature , these results suggest new experiments in ordered quantum magnets and ultracold atoms in honeycomb optical lattices , to search for magnon dirac materials and topological magnon insulators on the honeycomb lattice . for 3d ordered quantum magnets ,
weyl points are possible in the magnon excitations @xcite .
we argued that the bogoliubov hamiltonian near the weyl points should yield a low - energy hamiltonian that breaks time - reversal symmetry of the pseudo spins .
at nonzero temperature and external magnetic field , there is a possibility of topological magnon hall effect @xcite and spin nernst effect @xcite , similar to the kagome , lieb , and pyrochlore lattices @xcite .
the analysis of magnon hall effect for the honeycomb lattice will be reported elsewhere .
the author would like to thank african institute for mathematical sciences .
research at perimeter institute is supported by the government of canada through industry canada and by the province of ontario through the ministry of research and innovation .
hasan , c.l .
kane , rev .
mod . phys . * 82 * , 3045 ( 2010 ) .
xiao - liang qi , shou - cheng zhang , rev .
phys . * 83 * , 1057 ( 2011 ) a. h. castro neto _ et al_. , rev . mod . phys.*81 * , 109 ( 2009 ) .
et al_. , phys .
b*85 * , 195320 ( 2012 ) ; x. wan _ et al . _ ,
b * 83 * , 205101 ( 2011 ) ; a. a. burkov and l. balents , phys . rev
. lett . * 107 * , 127205 ( 2011 ) ; s. a. owerre , j. phys . : condens .
matter 28 ( 2016 ) 235501 .
_ et al_. , science , * 349 * , 613 ( 2015 ) ; b. q. lv _ et al_. , phys .
x * 5 * , 031013 ( 2015 ) .
j. fransson , a. m. black - schaffer , a. v. balatsky , arxiv:1512.04902 f. d. m. haldane , phys .
lett . * 61 * , 2015 ( 1988 ) .
b. lenk , h. ulrichs , f. garbs , m. mnzenberg , physics reports * 507 * , 107 ( 2011 ) .
a. v. chumak , v. i. vasyuchka , a. a. serga and b. hillebrands , nature physics * 11 * , 453 ( 2015 ) . r. shindou et.al . ,
b 87 , 174427 ( 2013 ) ; phys . rev .
b 87 , 174402 ( 2013 ) .
r. matsumoto , r. shindou , and s. murakami , phys .
b 89 , 054420 ( 2014 ) .
l. zhang , j. ren , j. s. wang , and b. li , phys .
b * 87 * , 144101 ( 2013 ) .
a. mook , j. henk , and i. mertig , phys .
b * 90 * , 024412 ( 2014 ) .
r. chisnell _
et al_. , phys .
115 * , 147201 ( 2015 ) .
et al_. , j. phys .
jpn . * 75 * , 084707 ( 2006 ) .
a. a. tsirlin , o. janson , and h. rosner , phys .
b * 82 * , 144416 ( 2010 ) ; see also the citations .
et al . _ ,
nature * 515 * , 237 ( 2014 ) .
maria hermanns , kevin obrien , simon trebst , phys .
lett . * 114 * , 157202 ( 2015 ) .
et al_. , arxiv:1602.04288 . upon finishing this work we became aware of ref .
@xcite , which focuses on weyl points on the breathing pyrochlore lattice .
t. matsubara and h. matsuda , prog .
* 16 * , 569 ( 1956 ) ; * 17 * , 19 ( 1957 ) .
h. guo _ et al_. , phys
b * 93 * , 121401(r ) ( 2016 ) .
s. a. owerre , arxiv:1603.07989 .
h. katsura , n. nagaosa and p. a. lee , phys .
104 * , 066403 ( 2010 ) ; y. onose _
et al_. , science * 329 * , 297 ( 2010 ) .
x. cao , k. chen and d. he , j. phys .
: condens .
matter * 27 * , 166003 ( 2015 ) .
|
it has been recently shown that in the heisenberg ( anti)ferromagnet on the honeycomb lattice , the magnons ( spin wave quasipacticles ) realize a massless two - dimensional ( 2d ) dirac - like hamiltonian .
it was shown that the dirac magnon hamiltonian preserves time - reversal symmetry defined with the sublattice pseudo spins and the dirac points are robust against magnon - magnon interactions
. the dirac points also occur at nonzero energy . in this paper
, we propose a simple realization of nontrivial topology ( magnon edge states ) in this system .
we show that the dirac points are gapped when the inversion symmetry of the lattice is broken by introducing a next - nearest neighbour dzyaloshinskii - moriya ( dm ) interaction .
thus , the system realizes magnon edge states similar to haldane model for quantum anomalous hall effect in electronic systems .
however , in contrast to electronic spin current where dissipation can be very large due to ohmic heating , noninteracting topological magnons can propagate for long time without dissipation as magnons are uncharged particles .
we observe the same magnon edge states for the xy model on the honeycomb lattice .
remarkably , in this case the model maps to interacting hardcore bosons on the honeycomb lattice .
quantum magnetic systems with nontrivial magnon edge states are called topological magnon insulators .
they have been studied theoretically on the kagome lattice and recently observed experimentally on the kagome magnet cu(1 - 3 , bdc ) with three magnon bulk bands .
our results for the honeycomb lattice suggests an experimental procedure to search for honeycomb topological magnon insulators within a class of 2d quantum magnets and ultracold atoms trapped in honeycomb optical lattices . in 3d lattices , dirac and weyl points were recently studied theoretically , however , the criteria that give rise to them were not well - understood .
we argue that the low - energy hamiltonian near the weyl points should break time - reversal symmetry of the pseudo spins .
thus , recovers the same criteria in electronic systems . _
_ k__eywords : topological magnon insulators , magnon edge states , magnon spintronics
|
the discovery of high - temperature superconductivity in rare earth ( r ) oxypnictides , rofeas @xcite , has initiated extensive research activities in this class of compounds @xcite .
the main structural building block of the feas - based superconductors is the fe@xmath2as@xmath2 layer with tetrahedrally coordinated fe covalently bonded to 4 as atoms .
the fe@xmath2as@xmath2 blocks are the active layers for the observed superconductivity and they are separated by charge reservoir layers that can modify or change the average charge of the active layer through appropriate chemical substitutions . in rofeas the appropriate doping can be achieved by substituting f for o @xcite or by removing oxygen @xcite . however , the amount of charge carriers introduced is limited and attempts to achieve higher carrier densities in the fe@xmath2as@xmath2 layers of the rofeas system frequently results in chemical disproportionation and decomposition and the formation of impurity phases @xcite .
the fe@xmath2as@xmath2 layers are very rigid and the charge reservoir blocks can be replaced by different oxygen - free layers , for example , by single layers of alkali metal or alkaline earth ( ae = ca , sr , ba ) ions .
stable structures of kfe@xmath2as@xmath2 , csfe@xmath2as@xmath2 @xcite , and lifeas @xcite have been synthesized and found superconducting at different temperatures .
it is interesting that the alkali metal - based compounds apparently do not need any additional doping to induce superconductivity , and they belong to a small family of intrinsic ( self - doped ) feas superconductors .
the critical temperatures of kfe@xmath2as@xmath2 and csfe@xmath2as@xmath2 are relatively low , below 4 k , but can be significantly increased up to 37 k by replacing the alkali metals with alkaline earth metals ( corresponding to a change of the carrier density in the fe@xmath2as@xmath2 layer ) @xcite .
the aefe@xmath2as@xmath2 compounds are nonsuperconducting but they show a transition into a spin density wave ( sdw ) state at higher temperatures .
the continuous substitution of k for sr or ba in k@xmath0(sr , ba)@xmath1fe@xmath2as@xmath2 leads to the onset of superconductivity at a critical doping of x@xmath5=0.17 , an increase of t@xmath5 to a maximum t@xmath737 k near the optimal doping ( [email protected] ) and , with further increasing x , a decrease of t@xmath5 to 3.8 k at x=1 ( kfe@xmath2as@xmath2 ) .
thus a complete superconducting phase diagram is obtained at all doping levels between x=0 and x=1 @xcite .
the doping dependence of t@xmath5 in this system seems to be qualitatively similar to the high - t@xmath5 copper oxide superconductors suggesting that the underlying physics in both systems is determined by the control of the charge density in the active fe@xmath2as@xmath2 layers .
this conjecture recently received support from a high - pressure investigation of k@xmath0sr@xmath1fe@xmath2as@xmath2 @xcite .
one major difference of the feas- and cuo - based high - temperature superconductors is the existence or coexistence of magnetic order in the carrier density - dependent phase diagram . in the copper oxides antiferromagnetism exists in the undoped , insulating state
and it is completely suppressed before superconductivity arises with doping . in k@xmath0sr@xmath1fe@xmath2as@xmath2 magnetic order in form of a spin density wave has been reported for the non - superconducting srfe@xmath2as@xmath2 @xcite and bafe@xmath2as@xmath2 @xcite .
the sdw transition is suppressed with doping of potassium , however , in a narrow range of the phase diagram , for 0.17@[email protected] , the spin density wave and superconducting transitions appear as successive transitions upon cooling @xcite . this raises the question about the coexistence of magnetic and superconducting orders in different regions of the phase diagram and it demands a detailed investigation of the physical properties in the whole phase diagram of k@xmath0(sr , ba)@xmath1fe@xmath2as@xmath2 with special attention to the narrow region between 0.17 and 0.3 .
we have therefore synthesized k@xmath0sr@xmath1fe@xmath2as@xmath2 , for 0@xmath3x@xmath31 , and measured the normal state and superconducting transport properties .
we show that the superconducting phases exhibit very high upper critical fields that scale with the transition temperatures .
the seebeck coefficients , s(t ) , of the series of samples are positive over the whole temperature range ( except for x@xmath60 ) , and s exhibits a significant change at [email protected] where the spin density wave state forms at higher temperatures .
polycrystalline samples of k@xmath0sr@xmath1fe@xmath2as@xmath2 were prepared from stoichiometric amounts of the ternary end members , kfe@xmath2as@xmath2 and srfe@xmath2as@xmath2 , thoroughly mixed , compressed , and annealed in welded nb containers ( jacketed in quartz ) at 700 @xmath10c for 100 to 120 hours .
the high phase purity of the ternary compounds , prepared from high purity k , sr , and feas as described previously @xcite , guarantees the exceptional quality of the solid solutions and the precise control of the chemical compositions .
electrical conductivities were measured by a four probe technique using the low - frequency ( 19 hz ) resistance bridge lr700 ( linear research ) and a physical property measurement system ( quantum design ) for temperature ( t@xmath11 1.8 k ) and magnetic field ( h@xmath9 7 tesla ) controls .
the seebeck coefficients were measured between 300 k and 5 k employing a high precision ac technique with a dual heater system @xcite .
the error in the measurements is about 0.05 @xmath12v / k .
the room temperature x - ray diffraction data of all samples were indexed to the thcr@xmath2si@xmath2-type tetragonal i4/mmm structure . the systematic change of the a- and c - axis lattice constants ( fig .
1 ) indicate the formation of a uniform solution between the two compounds kfe@xmath2as@xmath2 and srfe@xmath2as@xmath2 .
the increase of the c - axis by nearly 12 % with increasing x is consistent with the larger k atom replacing the smaller sr atom .
the change of the a - axis ( 2 % decrease from srfe@xmath2as@xmath2 to kfe@xmath2as@xmath2 ) is much smaller .
the systematic changes in the unit cell parameters result in an increase of the volume as well as the c / a ratio with x. sr@xmath1fe@xmath2as@xmath2 vs. composition x.,width=3 ] the resistivity of srfe@xmath2as@xmath2 shows the distinct anomaly ( inflection point ) at the spin density wave ( sdw ) transition temperature , t@xmath13=204 k , as reported earlier @xcite .
this anomaly is very sharp and it was shown recently that the sdw transition in srfe@xmath2as@xmath2 coincides with a structural deformation accompanied by a change of symmetry from tetragonal ( t@xmath11t@xmath13 ) to orthorhombic ( t@xmath9t@xmath13 ) @xcite . with increasing k content , the sdw anomaly shifts to lower temperatures as shown in fig .
2 . for x=0.12 ,
the sdw transition temperature decreased to 175 k from 204 k ( x=0 ) . with further increasing x ,
the critical temperature t@xmath13 drops quickly and the sharp anomaly of dr / dt disappears near [email protected] , limiting the sdw state to the very strontium - rich region of the phase diagram .
it is interesting to note that , before the sdw state is completely suppressed , superconductivity arises at a k concentration of about 0.17 , with a t@xmath5 of 2.9 k ( fig .
therefore , in a narrow range of chemical composition , between x=0.17 and x=0.3 , k@xmath0sr@xmath1fe@xmath2as@xmath2 passes from a nonmagnetic into a sdw state at t@xmath13 and then into the superconducting state at t@xmath5 .
it is not clear , however , if the sdw order survives the transition into the superconducting state , and if both orders coexist at low temperatures .
sr@xmath1fe@xmath2as@xmath2 for [email protected] .
t@xmath13 and t@xmath5 are defined by the sharp peaks of dr / dt in ( b ) .
the curves in ( a ) are vertically offset for better clarity and the attached numbers are the values of x. note that the vertical scale in ( b ) is changed at 50 k.,width=4 ] upon further increasing x ( hole doping ) the superconducting t@xmath5 rises to a maximum of 37 k and decreases again between x@xmath14=0.45 and x=0 . the resistivity data for [email protected] are shown in fig .
the temperature derivative , dr / dt , clearly defines the critical temperature and shows the sharp resistive transition into the superconducting state .
the phase diagram , derived from the resistivity data and shown in fig .
4 , is slightly asymmetric with respect to x. the nearly parabolic t@xmath5(x ) curve develops a tail - like extension to large values of x , and the ternary kfe@xmath2as@xmath2 remains superconducting with a t@xmath5 of 3.8 k. the major part of the phase diagram is similar to that known for the high - t@xmath5 cuprate superconductors having a dome - shaped parabolic t@xmath5 dependence on the carrier density . the main qualitative difference is the tail of the phase boundary in the k@xmath0sr@xmath1fe@xmath2as@xmath2 system .
sr@xmath15fe@xmath2as@xmath2 for [email protected] .
different curves are vertically offset and separated by 0.15 units to enhance clarity of presentation .
inset : derivative dr / dt with the peak position defining t@xmath5.,width=2 ] sr@xmath1fe@xmath2as@xmath2 .
the dashed lines serve as a guide to the eye.,width=3 ] the extrapolation of the sdw phase boundary to lower temperatures suggests that the magnetically ordered state becomes unstable for values of [email protected] .
the coexistence of both , the sdw and the superconducting transitions between x=0.17 and x=0.3 is another distinct feature that is different from the cuprate superconductors . in the latter system the antiferromagnetic order of the copper spins
is completely suppressed before superconductivity sets in at a higher doping level , and both transitions do not coexist at any point of the t@xmath5(n ) ( n carrier density ) phase diagram .
the extrapolation of t@xmath13 to zero temperature , as schematically shown in fig .
4 ( dotted line ) , may suggest a `` hidden '' quantum critical point ( qcp ) at t=0 and superconductivity induced by magnetic fluctuations . the exact value of x defining the qcp is difficult to determine since the sdw phase boundary is cut off by the superconducting transition and , as shown in fig .
2 ( b ) , the resistive anomaly that is characteristic for the sdw transition broadens significantly with increasing x. if magnetic fluctuations are involved in the superconducting pairing mechanism one should expect a maximum of t@xmath5 close to the critical doping x of the qcp , where the magnetically ordered state comes closest to the superconducting state . in the phase diagram of fig
4 , the highest t@xmath5 is observed at a k content of about 0.45 .
more detailed experiments are therefore needed near the maximum composition of the superconducting dome .
alternatively , the sdw and superconducting states may compete for the ground state in which case magnetic order would be counterproductive in stabilizing the superconducting state .
recent results of high - pressure experiments have shown that , in the narrow coexistence range of both transitions , the application of external pressure did drastically reduce the sdw temperature whereas the superconducting t@xmath5 was enhanced @xcite , suggesting a possible competition between both states .
the electrical resistivity of polycrystalline samples depends , to a large extent , on the grain boundary resistance and contributions from grain - grain contacts .
it is therefore difficult to extract detailed information about the electronic states .
the seebeck coefficient is less sensitive to grain boundary effects ( since it is a zero - current property ) and it provides information about both thermodynamic and transport properties of the carriers . its sensitivity to the electronic band structure makes it a perfect probe in detecting changes in the density of states and the fermi surface .
the seebeck coefficient for a single band metal can be expressed as @xcite @xmath16\ ] ] d(e ) , @xmath17(e ) , and @xmath18(e ) are the density of states , an average charge velocity , and the carrier scattering relaxation time , respectively .
the first term in the brackets shows the sensitivity of s with respect to changes of the band structure , shifts of the fermi energy , or instabilities of the fermi surface .
sr@xmath1fe@xmath2as@xmath2 for selected values of x ( labels next to the curves).,width=3 ] the seebeck coefficients , s(t ) , for some selected compositions of k@xmath0sr@xmath1fe@xmath2as@xmath2 are shown in fig .
s is positive over the whole temperature range ( with the exception of srfe@xmath2as@xmath2 where it is negative near room temperature and at low t ) , and drops to zero at the superconducting transition .
the positive sign is consistent with the positive hall coefficient @xcite indicating that the majority carriers are holes in the system k@xmath0sr@xmath1fe@xmath2as@xmath2 .
however , band structure calculations @xcite and arpes measurements @xcite have shown that the fermi surface consists of states from four different bands , some with electron - like and others with hole - like states .
the hole fermi surfaces are located around the zone center whereas the electron surfaces are at the zone corner ( m point ) of the primitive tetragonal brillouin zone .
contributions from all bands to the seebeck coefficient are therefore expected and the doping and temperature dependence of s(t ) will be more complex than expressed by the simple formula ( 1 ) .
the room temperature values of s initially increase with x as a consequence of hole doping .
this could be explained within a rigid band model by a lowering of the fermi energy and an increase of the density of states .
however , results of a recent density functional band structure calculation within a virtual crystal approximation @xcite have questioned the validity of the rigid band model and have shown ( for the related system k@xmath0ba@xmath1fe@xmath2as@xmath2 ) that the density of states depends only weakly on the doping level .
the major effect of hole doping is a redistribution of states between the electron and hole pockets of the fermi surface without significant change of the total n(e@xmath19 ) . with increasing x the hole states tend to dominate as is also obvious from the positive signs of the seebeck coefficient and the hall number .
this redistribution of states with x explains the increase of the room temperature value of s. since different contributions from electron and hole carriers to s(t ) partially cancel out , the weight shift towards the hole pockets at the fermi surface results in an increase of s(t ) even if n(e@xmath19 ) is not changed .
the effect is even larger at lower temperatures and it is also reflected in the t - dependence of s. for small values of x , s(t ) shows a pronounced maximum near the sdw transition .
the maximum value , s@xmath14 , rapidly increases with hole doping until x reaches a critical value [email protected] . at the critical x
, the value of s@xmath14 peaks and decreases with further doping .
the x - dependence of s@xmath14 is shown in fig .
the sharp and distinct peak at the critical composition indicates a dramatic change of the fermi surface and the density of states with the onset of the sdw order .
this might be an indication of an important change in the electronic system and the possible existence of a magnetic quantum phase transition at t=0 that could only be revealed if the superconductivity was completely suppressed .
theoretical calculations show that the sdw state is close in energy to other magnetically ordered states such as ferromagnetism and checkerboard antiferromagnetism . whenever different magnetic states are competing for
the ground state spin fluctuations in the paramagnetic phase may be large and significant .
the doping - induced instability of the magnetic order in the presence of large spin fluctuations , possibly giving rise to the pairing interaction in the superconducting state , can then be understood as a delicate balance between the sdw and the superconducting states .
experiments at much lower temperature are suggested to reveal more details about the critical behavior at the sdw phase boundary and the qcp . however , as shown in the phase diagram of fig .
4 , the onset of superconductivity at about 30 k ( for [email protected] ) does not allow the study of the sdw phase boundary closer to zero temperature .
the suppression of superconductivity , for example in high magnetic fields , is therefore suggested to investigate the qcp .
however , as discussed in the following section , the superconducting critical field in this composition range is very high and static fields of this magnitude may not be available .
it is also not clear how the high field would affect the sdw state .
( x ) appears at the onset x@xmath13 of the sdw phase.,width=3 ] the dependence of superconducting properties on external magnetic fields is an essential property which determines the performance of superconducting devices in magnets , for energy transport , etc . for type - ii superconductors ,
the upper critical field ( h@xmath4 ) limits the stability range of the superconducting state and its zero temperature value , h@xmath4(0 ) , is a characteristic quantity that can be determined or extrapolated in high - field measurements . in the k@xmath0sr@xmath1fe@xmath2as@xmath2 system h@xmath4(0 ) can be very high , exceeding 100 tesla , as shown for the optimally doped compounds @xcite .
sr@xmath1fe@xmath2as@xmath2 in magnetic fields at the superconducting transition for different compositions x. different curves are measured in magnetic fields from 0 to 7 tesla in steps of 1 tesla ( right to left , black to pink ) .
the insets show the critical fields estimated from the midpoint of the resistivity drop.,width=5 ] ( 0 ) as function of x ( a ) and t@xmath5 ( b).,width=3 ] to extrapolate the zero temperature h@xmath4 for the complete series of k@xmath0sr@xmath1fe@xmath2as@xmath2 compounds we measured the resistivity in magnetic fields up to 7 tesla .
some selected data for x=0.2 , 0.42 , 0.67 , and 0.9 are shown in figs . 7 ( a ) to ( d ) .
the insets to the figures show the temperature dependence of h@xmath4 .
t@xmath5 at different fields was determined from the midpoint of the resistivity drop .
this t@xmath5 criteria may appear somewhat arbitrary and the exact value of the extrapolated h@xmath4(0 ) may depend on the way t@xmath5 was defined ( alternative criteria , for example , are the 10 % or 90 % resistivity drop ) , however , since we are more interested in comparing the estimated h@xmath4 values for different x throughout the phase diagram the choice of the 50 % resistivity drop is reasonable .
the zero temperature h@xmath4(0 ) was extrapolated using the ginzburg - landau formula , @xmath20 , with @xmath21 , from the data similar to those shown in the insets to fig .
the estimated values are within 5 % of the h@xmath4 determined from the alternative relation @xmath22 using the slope of h@xmath4(t ) near t@xmath5 . the extrapolated values for h@xmath4(0 ) as a function of the k content , x ,
are shown in fig .
h@xmath4(0 ) is very large , up to 200 tesla , in the optimally doped region of the phase diagram with t@xmath5 s up to 37 k. however , it is drastically reduced , on both sides , where t@xmath5 drops to zero ( x@xmath9x@xmath14 ) or to small values for x@xmath231 .
this suggests that h@xmath4 scales with the critical temperature . plotting h@xmath4(0 ) versus t@xmath5 in fig .
8 ( b ) indeed shows a parabolic dependence , @xmath24 ( dashed line in fig . 8 ( b ) , a=0.152 tesla / k@xmath25 ) .
the large values of h@xmath4 are typical for a short ginzburg - landau coherence length , @xmath26 , that can be extracted through @xmath27 $ ] , where @xmath28 is the flux quantum .
for the highest t@xmath5 s in the k@xmath0sr@xmath1fe@xmath2as@xmath2 system coherence lengths of about 13 are thus obtained .
the short coherence length in the superconducting state is another indication of the analogy of the feas - based superconductors to the high - t@xmath5 copper oxides .
while the discussion about the pairing symmetry and the structure of the superconducting gap in feas superconductors is not yet decided there is increasing evidence for the existence of nodes in the superconducting gap function and two - gap superconductivity with carriers from two bands forming cooper pairs @xcite .
a more detailed investigation of the superconducting properties of the feas superconductors has to take into account the specifics of the pairing symmetry and the different bands at the fermi surface .
we have investigated the phase diagram and the normal state transport properties of the superconducting system k@xmath0sr@xmath1fe@xmath2as@xmath2 .
the dome - shaped phase diagram of the superconducting state resembles that of the high - t@xmath5 cuprates with a maximum of t@xmath5 near an optimal doping range .
magnetic order exists on the sr - rich side and the spin density wave transition is followed by a superconducting transition within a narrow doping range .
the seebeck coefficient shows a distinct anomaly near x=0.3 where the sdw phase arises above the superconducting dome .
this anomaly indicates a significant change of the fermi surface and the possible existence of a quantum critical point is proposed .
the most interesting property of the superconducting state is a very high upper critical field estimated from resistivity measurements in magnetic fields .
the large values of h@xmath4(0 ) imply a short ginzburg - landau coherence length , similar to the high - t@xmath5 cuprate superconductors .
this work is supported in part by the t.l.l .
temple foundation , the j.j . and
r. moores endowment , the state of texas through tcsuh , the usaf office of scientific research , and at lbnl through usdoe .
a.m.g . and
b.lv acknowledge the support from the nsf ( che-0616805 ) and the r.a .
welch foundation .
10 h. luetkens , h .- h .
klauss , m. kraken , f. j. litterst , t. dellmann , r. klingeler , c. hess , r. khasanov , a. amato , v. baines , j. hamann - borrero , n. leps , a. kondrat , g. behr , j. werner and b. bchner .
arxiv:0806.3533 ( unpublished ) .
|
the normal state and superconducting properties are investigated in the phase diagram of k@xmath0sr@xmath1fe@xmath2as@xmath2 for 0@xmath3x@xmath31 .
the ground state upper critical field , h@xmath4(0 ) , is extrapolated from magnetic field dependent resistivity measurements .
h@xmath4(0 ) scales with the critical temperature , t@xmath5 , of the superconducting transition . in the normal state
the seebeck coefficient is shown to experience a dramatic change near a critical substitution of [email protected] .
this is associated with the formation of a spin density wave state above the superconducting transition temperature .
the results provide strong evidence for the reconstruction of the fermi surface with the onset of magnetic order .
|
magnetic helicity is a measure of twists , kinks , and inter - linkages of magnetic field lines ( berger & field 1984 ) and has been an important parameter in solar dynamo theories ( parker 1955 ) .
while the source of the magnetic helicity lies below the surface of the sun , it was recently recognized as a useful parameter in describing solar features observed above surface such as spiral patterns of sunspot fibrils , helical patterns in filaments and coronal mass ejections ( cmes ; for a review , see brown et al .
naturally magnetic helicity studies have been directed to the energy buildup and instability leading to eruptions and cmes ( e.g. , rust 2001 ; kusano et al .
2004 ; phillips et al . 2005 ) .
more recently , several studies were carried out to relate the change of magnetic helicity to the problem of impending or triggering solar flares .
moon et al .
( 2002a ) studied the magnetic helicity change around major flares to find its rapid helicity change before flares , and concluded that a sudden helicity injection may trigger flares .
moon et al .
( 2002b ) applied the same approach to seven homologous flares in the active region , noaa 8100 , over a period of 6.5 hr to find a good correlation between the amount of incremental helicity and the soft x - ray flux during each homologous flare .
the results from both studies thus point to the idea that the helicity change occurring over short timescales ( around a half hour ) can be a significant factor in triggering flares .
kusano et al .
( 2003 ) proposed annihilation of magnetic helicity as a triggering mechanism for solar flares .
numerical simulations were carried , which show that , if the helicity is sharply reversed within a magnetic arcade , reconnection quickly grows in the helicity inversion layer , driving explosive dynamics .
yokoyama et al .
( 2003 ) studied flare activities in active region noaa 8100 to find that most of the flare events occurred about half a day after the helicity injection rate changed its sign , and the positions of h@xmath10 emission in flares well correspond to the helicity inversion lines in space .
sakurai & hagino ( 2003 ) studied two active regions appeared in 2001 ( noaa 9415 and 9661 ) , both of which have produced x - class flares .
their finding was , on the contrary , that the magnetic helicity integrated over the regions evolved slowly and did not show abrupt changes at the time of the flares , although the distributions of magnetic helicity changed significantly over a few days in the regions . in this paper , we study long term ( a few days ) variations of the magnetic helicity around major x - class flares . while some of the above studies suggest short - term helicity change as an important topic for flare triggering , hartkorn & wang ( 2004 ) found that the rapid helicity change at the time of a flare can occur as an artifact under the influence of flare emission on the spectral line adopted in mdi measurements .
this means that a short term variation during strong flares can hardly be measured with enough accuracy .
however we can study a long - term variation of the magnetic helicity with active region magnetograms gathered over many days excluding the times of flares .
our research is therefore focused on a possible characteristic helicity evolution pattern that is associated with flare impending mechanisms .
by magnetic helicity , we refer to the _ relative _ magnetic helicity in the rest of this paper , i.e. , the helicity relative to that of the potential field state . with the time - dependent measurement of longitudinal magnetic fields in the photosphere , we can
only approximately determine the change rate of the relative magnetic helicity ( demoulin & berger 2003 ) .
we further use a simplified expression for the helicity change rate ( chae 2001 ) given by @xmath11 where @xmath12 is the normal component of the magnetic field ; @xmath13 is the vector potential of the potential field ; @xmath14 represents the apparent horizontal motion of field lines ; @xmath15 is the surface integral element and the integration is over the entire area of the target active region .
although this expression does not explicitly include the helicity change by the vertical motion of field lines ( see kusano et al .
2002 ) , dmoulin & berger ( 2003 ) pointed out that it actually accommodates both the vertical and horizontal motions of flux tubes as far as no flux tube newly emerges from or totally submerges into the surface . as we will show later in presenting the result of helicity calculation , this requirement is not always met .
we determine the quantities in the above equation using full disk mdi ( scherrer et al .
1995 ) magnetograms following the procedure described in chae & jeong ( 2005 ) .
first , we approximately determine @xmath12 from the line - of - sight magnetic field @xmath16 in the mdi magnetograms , simply considering the projection effect , i.e. , @xmath17 where @xmath18 is the heliocentric angle of the point of interest , assuming that the magnetic field on the solar photosphere is normal to the solar surface .
second , @xmath19 is calculated from @xmath12 by using the fast fourier transform method as usual .
the extent of the spatial domain of the fourier transform is taken about twice wider than the active region area in order to minimize the artifacts arising from the periodic boundary condition in the fast fourier transform ( alissandrakis 1981 ) .
third , @xmath20 is calculated using the local correlation tracking ( lct ) technique ( november & simon 1988 ) . for local correlation tracking ,
we align all magnetograms in each event to the first image of the data set after correcting the differential rotation .
we set the fwhm of the apodizing window function to 10 and the time interval between two frames to 60 minutes , and performed lct for all pixels with an absolute flux density greater than 5 g. only the pixels with cross correlation above 0.9 are considered . in selecting data we found that use of 1 minute cadence full - disk mdi ( scherrer et al .
1995 ) magnetograms is adequate for our purpose of investigating the long - term helicity evolution .
however , there are occasionally found data gaps in the 60 minute cadence data set in which case we supplement the data gaps with 96 minute mdi magnetograms .
the time interval of the supplemented data set is therefore not longer than 96 minute . to reduce the effect of the geometrical projection
, we selected the active regions lying within 60% of the solar radius from the apparent disk center .
note that we use only full disk mdi magnetograms that have 2@xmath52 pixel size .
therefore the lct velocities calculated here may have been systematically underestimated compared with the lct velocities calculated with the higher resolution ( [email protected] ) mdi data ( longcope et al .
2007 ) .
after the helicity change rate is determined as a function of time , we integrate it with respect to time to determine the amount of helicity accumulation : @xmath21 where @xmath22 and @xmath23 are the start and end time of the helicity accumulation , respectively .
if @xmath22 is a time when the magnetic field is in the potential state , @xmath24 is simply the helicity , @xmath25 , at time @xmath23 .
however , there is no guarantee that we can observe , by chance , an active region in the potential energy state .
we therefore set @xmath22 as the earliest time without significant helicity accumulation at which the average value of the helicity change rate over four hours is less than our nominal threshold in helicity change rate , 1@xmath510@xmath8 mx@xmath7 hr@xmath9 .
if that time can not be determined , we define @xmath22 as the time when the data set starts or when the previously accumulated helicity is released by a flare .
the exact time of @xmath22 here is unimportant because it is only a trial value . after determining @xmath25 ,
we redefine @xmath22 as the time when the resulting helicity starts to increase from a nearly constant value .
we present the helicity variation calculated for seven active regions in figures 1 and 2 . in both figures
we plot the magnetic helicity accumulation together with the goes soft x - ray light curve and magnetic flux as functions of time .
the soft x - ray light curve is shown to indicate the flare times and the magnetic flux is shown to check the above - mentioned requirement for the approximation made in equation ( 1 ) .
note that the fluxes shown in this paper are total unsigned magnetic flux , i.e. , sum of the absolute amounts of positive and negative fluxes , because net magnetic flux may show little change despite significant flux change in each polarity . for the events shown in figure 1 we can see that the helicity accumulates at a monotonic rate of change about 0.52 days before the flare onset , and then becomes almost constant before the flares . for convenience , we distinguish the magnetic helicity variation in two stages : a phase of monotonically increasing helicity ( phase i ) and the following phase of relatively constant helicity ( phase ii ) .
this pattern is obvious for the four flares ( 2001 october 25 , 2004 november 7 , and 2005 january 16 and 17 ) . for the 2005 january 15 event ,
the helicity increased up to 22:00 ut on january 14 and then decreased afterward . in this case
, we do not consider that the flare occurred in phase ii .
it is then noted that these flares took place after a significant amount , @xmath26(1.811)@xmath510@xmath6 mx@xmath7 , of helicity accumulation . in figure 2
, we show the result for the other four active regions . like the events in figure 1
, these flares also occurred after a significant helicity accumulation , @xmath26(1.916)@xmath510@xmath6 mx@xmath7 .
however , they occurred in the middle of the continuous helicity accumulation , unlike those events shown in figure 1 . in other word , the flares in figure 2 occurred in phase i , while those in figure 1 occurred in phase ii .
one common trend is , however , that all the events are apparently associated with a considerable amount of helicity build - up before the flares , whether they occurred in phase i or in phase ii . in case where flares occur in phase ii
, it may imply that solar active regions can wait for major flares after the helicity accumulated to some limiting amount .
this is seemingly contrary to the general belief that a flare occurs as soon as the system reaches some threshold .
an active region may evolve to a certain stage where the helicity no longer increases , and the system waits until it unleashes the stored energy by producing flares due to certain mechanism of triggering . since we claimed that these large flares are always preceded by significant accumulation of helicity , as a reference we check the corresponding helicity variation in non - flaring times .
we show such data in figure 3 . for all active regions under investigation ,
the amount of helicity change during nonflaring periods ( fig .
3 ) is much less than that around the major flare time .
this convinces us that the above monotonically increasing helicity before major flares is a process associated with the flares and is not occurring in nonflare times .
another point to note in figure 3 is that not only the helicity but also the total unsigned magnetic flux changes much less during the nonflaring time compared with the period before major flares .
this implies that the increase of the magnetic helicity before major flares is , in part , related to the simultaneous increase of total unsigned magnetic flux .
it is also worthwhile to mention how the characteristic pattern of the helicity variation found here will depend on the sign of helicity . in our result obtained for seven active regions , similar amounts of both of positive and negative helicity
were accumulated continuously and simultaneously during the whole time .
it is therefore unlikely that counting the helicity in one and the other polarity separately yields a significantly different conclusion . on the other hand
, some studies suggested that the sign - reversal of the helicity injection rate is important for flare activity and we need to compared them with the present result .
kusano et al .
( 2003 ) emphasized spatially sharp reversal of helicity sign triggers magnetic reconnection based on model simulation . while the model prediction is interesting and compelling , we excluded from the outset ( see 1 ) the rapid helicity change during the flare time due to observational limitation .
our conclusion is valid only for the long term variation of helicity .
yokoyama et al .
( 2003 ) have found that flares tend to occur after reversal of helicity injection rate changed its sign .
although this is occasionally seen in our samples , ( i.e. , in the case of ar 10030 and ar 10720 ) as well , it is not always the case and we are unsure whether this is a necessary condition for the flares . more often than not , the helicity either remains constant or increases in one sign when the flare occurs .
we compare the helicity change rate , @xmath27 and accumulation amount , @xmath24 with the time integrated soft x - ray flux taken as the proxy for the flare energy release .
in addition we check the helicity accumulation time , @xmath28 , defined as the time interval of helicity accumulation measured from @xmath22 and the first coming flare . prior to make such a comparison , the range of uncertainty of each quantity needs to be known . in general
it is hard to trace all the possible uncertainties involved with each quantities in equation ( 1 ) .
fortunately , our targeted quantity given by equation ( 2 ) involves integration in space and time and the uncertainty in each measured quantity is not propagating , but rather may cancel out in the process of spatial and time integration if it is random in nature .
we thus focus the uncertainty estimate only on the linear approximation of the helicity variation that we are after .
we first find out the best - fit linear function to the points @xmath29 lying in phase i ( i.e. , @xmath30 ) in the form of @xmath31 .
next we calculate the standard deviation , @xmath32 , of the scatter points with respect to this linear function , and plot two additional lines corresponding to the @xmath33 levels of the scatter points . finally , we read the @xmath34-axis offsets of these two lines to determine @xmath35 and the x - axis offsets to determine @xmath36 , respectively .
in addition , we calculate the uncertainty of the slope @xmath37 itself in the form of @xmath38 .
the center value of @xmath37 here is taken as the average helicity change rate in the rest of this paper .
therefore , by the average helicity change rate , we do not mean the average of the quantities given in equation ( 1 ) , but we refer to the best fit slope to the helicity variation ( eq . [ 2 ] ) in phase i. the uncertainties shown in figure 4 and table 1 are those associated with our linear function fit only . in figure 4 ,
we plot , as symbols , the helicity parameters against the goes soft x - ray fluxes integrated over the flaring time ( @xmath39 , hereafter ) .
each symbol is identified with the event i d number in the figure together with uncertainty range represented by the bar ( see also table i ) .
the solid lines show the least - squares linear fits to the data points .
the correlation coefficients ( ccs ) of the linear fits are also given in each panel .
figure 4@xmath37 shows that there is a fairly good correlation ( cc=0.86 ) between the helicity change rate and @xmath39 .
the amount of helicity accumulation also shows a modest correlation with @xmath39 ( cc=0.68 ) as shown in figure 4@xmath40 , although not as good as for the helicity change rate . on the other hand ,
the correlation between helicity accumulation time @xmath28 and the soft x - ray flux is very poor with a weak tendency that the longer accumulation time @xmath28 , the weaker soft x - ray flux @xmath39 ( fig .
4@xmath41 ) .
we initially expected , on a general basis , that the helicity change @xmath24 would strongly correlate with @xmath39 .
it is therefore puzzling why the helicity change rate @xmath27 shows even a better correlation with @xmath39 in figure 4 .
as a possibility , we considered that @xmath28 may be a factor in complicating the relationship between @xmath24 and @xmath39 .
a intriguing idea is that the magnetic energy decays much faster than the magnetic helicity in the presence of magnetic diffusion ( berger 1999 ) .
we thus compare @xmath24 with @xmath28 in figure 4d , which unfortunately shows no obvious correlation between them .
the small number of events used in this study is another restriction for finding a trend here . with the present result alone
, it is fair to presume that the weaker correlation between @xmath39 and @xmath24 may arise from our inaccurate determination of the helicity accumulation amount due to unknown initial time of helicity build - up .
we have investigated the variation of magnetic helicity over a time span of several days around the times of 11 x - class flares which occurred in seven active regions using mdi magnetograms to find the following results .
first , a substantial amount of helicity accumulation is found before the flare in all the events .
the helicity increases at a nearly constant rate , ( 4.548)@xmath510@xmath8 mx@xmath7 hr@xmath9 , over a period of 0.6 to a few days , resulting in total amount of helicity accumulation in the range of ( 1.816)@xmath510@xmath6 mx@xmath7 .
such a wide range of helicity accumulation indicates that each active region has its own limit of helicity storage to keep a stable magnetic structure in the corona .
the finding of a monotonically increasing phase is similar to the earlier one by sakurai & hagino ( 2003 ) that the magnetic helicity integrated over the regions evolved slowly and did not show abrupt changes at the time of the flares .
the helicity increase over days before the flares reconfirms the conventional idea that helicity accumulation by a certain amount is necessary for a large flare to occur ( kusano et al . 1995 ; choe & lee 1996 ) .
second , there is a strong positive correlation between the average helicity change rate of phase i and the corresponding goes x - ray flux integrated over the flaring time .
the amount of helicity accumulation during phase i also correlates with the soft x - ray flux , as expected , but the correlation is stronger with the helicity change rate .
this result probably implies that the helicity change rate is more accurately determined than the amount of helicity change itself as the initial time of helicity build - up is poorly determined .
if the above correlations hold for a large number of events , we may predict the flare strength ( the integrated x - ray flux ) based on the helicity change rate .
monitoring of helicity variation in target active regions may also aid the forecasting of flares .
a warning sign of flares can be given by the presence of a phase of monotonically increasing helicity , as we found that all the major flares occur after significant helicity accumulation .
as a reference we have checked helicity variation of the six active regions in non - flaring times to find much lower helicity change rates compared with those around the major flares .
we thus conclude that the relative magnetic helicity can be a powerful tool for predicting major flares .
the authors wish to thank the referee for valuable comments on the manuscript .
the work is supported by nsf grant atm-0548952 and nasa grant nng0 - 6gc81 g .
j.l . was supported by nsf grant ast 06 - 07544 and nasa grant nng0 - 6ge76 g .
g.s.c . was supported by doe contract de - ac02 - 76-ch03073 and nasa grant nnh04aa16i .
alissandrakis , c. e. 1981 , , 100 , 197 berger , m. a. , & field , g. b. 1984 , j. fluid mech . , 147 , 133 berger , m. a. 1999 , in m. r. brown , r. c. canfield , and a. a. pevtsov ( eds . ) magnetic helicity in space and laboratory plasmas , geophys .
, 111 , 11 brown , m. r. , canfield , r. c. , & pevtsov , a. a. ( eds . ) 1999 , magnetic helicity in space and laboratory plasmas , geophys .
, 111 chae , j. 2001 , , 560 , l95 chae , j. , & jeong , h. 2005 , j. korean astron .
soc . , 38 , 295 choe , g. s. , & lee , l. c. 1996 , , 472 , 372 dmoulin , p. , & berger , m. a. 2003 , , 215 , 203 hartkorn , k. , & wang , h. 2004 , , 225 , 311 kusano , k. , suzuki , y. , & nishikawa , k. 1995 , , 441 , 942 kusano , k. , maeshiro , t. , yokoyama , t. , & sakurai , t. 2002 , , 577 , 501 kusano , k. , yokoyama , t. , maeshiro , t. , & sakurai , t. 2003 , adv .
space res . , 32 , 1931 kusano , k. , maeshiro , t. , yokoyama , t. , & sakurai , t. 2004 , apj , 610 , 537 longcope , d. w. , ravindra , b. , & barnes , g. 2007 , , 668 , 571 moon , y .- j . , chae , j. , wang , h. , & park , y. d. 2002a , , 580 , 528 moon , y .-
j . , chae , j. , choe , g. s. , wang , h. , park , y. d. , yun , h. s. , yurchyshyn , v. , & goode , p. r. 2002b , , 574 ,
1066 november , l. j. , & simon , g. w. 1988 , , 333 , 427 parker , e. n. 1955 , , 122 , 293 phillips , a. d. , macneice , p. j. , & antiochos , s. k. 2005 , , 624 , l129 rust , d. m. 2001 , j. geophys .
res . , 106 , 25075 sakurai , t. , & hagino , m. 2003 , adv .
space res . , 32 , 1943 scherrer , p. h. , et al .
1995 , , 162 , 129 yokoyama , t. , kusano , k. , maeshiro , t. , & sakurai , t. 2003 , adv .
space res .
, 32 , 1949 llcccccc 1 & x 1.3 on oct 25 , 2001 & 9672 & 15:02 & 2.3 & [email protected] & [email protected] & 29@xmath422 + 2 & x 3.0 on jul 15 , 2002 & 0030 & 20:08 & 1.4 & [email protected] & [email protected] & 14@xmath421 + 3 & x 1.5 on mar 17 , 2003 & 0314 & 19:05 & 1.3 & [email protected] & [email protected] & 63@xmath428 + 4 & x 1.5 on mar 18 , 2003 & 0314 & 12:08 & 1.3 & [email protected] & [email protected] & 17@xmath421 + 5 & x 18 on oct 28 , 2003 & 0486 & 11:10 & 20.0 & [email protected] & [email protected] & 30@xmath422 + 6 & x 10 on oct 29 , 2003 & 0486 & 20:49 & 9.1 & [email protected] & [email protected] & 34@xmath421 + 7 & x 1.2 on feb 26 , 2004 & 0564 & 02:03 & 0.75 & [email protected] & [email protected] & 70@xmath423 + 8 & x 2.2 on nov 07 , 2004 & 0696 & 16:06 & 2.1 & [email protected] & [email protected] & 54@xmath421 + 9 & x 1.3 on jan 15 , 2005 & 0720 & 00:43 & 1.3 & [email protected] & [email protected] & 19@xmath422 + 10 & x 2.8 on jan 15 , 2005 & 0720 & 23:00 & 6.6 & [email protected] & [email protected] & 19@xmath421 + 11 & x 4.1 on jan 17 , 2005 & 0720 & 09:52 & 9.1 & [email protected] & [email protected] & 8@xmath421 x - ray flux for three active regions .
the helicity is shown as cross symbols and the magnetic flux is shown as diamonds .
the @xmath43 x - ray flux is shown as the dotted lines .
phase i , the interval over which the helicity accumulation is considered , and phase ii , the following phase of relatively constant helicity , are marked .
[ fig1 ] ] x - ray flux integrated over the flaring time .
correlations of the integrated soft x - ray flux with ( @xmath37 ) average helicity change rate of phase i , ( @xmath40 ) the amount of helicity accumulation during phase i , and ( @xmath41 ) helicity accumulation time .
correlation coefficient ( cc ) is specified in each panel . in ( @xmath44 ) , the amount of helicity accumulation is plotted as a function of the accumulation time .
the uncertainties of the average helicity change rate , the amount of helicity accumulation , and the helicity accumulation time are shown as error bars in each panel .
|
we have investigated the variation of magnetic helicity over a span of several days around the times of 11 x - class flares which occurred in seven active regions ( noaa 9672 , 10030 , 10314 , 10486 , 10564 , 10696 , and 10720 ) using the magnetograms taken by the michelson doppler imager ( mdi ) on board the @xmath0 @xmath1 @xmath2 @xmath3 ( @xmath4 ) . as a major result we found that each of these major flares was preceded by a significant helicity accumulation , ( 1.816)@xmath510@xmath6 mx@xmath7 over a long period ( 0.5 to a few days ) .
another finding is that the helicity accumulates at a nearly constant rate , ( 4.548)@xmath510@xmath8 mx@xmath7 hr@xmath9 , and then becomes nearly constant before the flares .
this led us to distinguish the helicity variation into two phases : a phase of monotonically increasing helicity and the following phase of relatively constant helicity .
as expected , the amount of helicity accumulated shows a modest correlation with time - integrated soft x - ray flux during flares .
however , the average helicity change rate in the first phase shows even stronger correlation with the time - integrated soft x - ray flux .
we discuss the physical implications of this result and the possibility that this characteristic helicity variation pattern can be used as an early warning sign for solar eruptions .
|
recent years have seen an increasing interest in the spin properties of nanostructures @xcite .
manipulation and readout of spins in solids could open the way to the development of a generation of electronic devices such as spin transistors , spin filters , and spin memory devices .
in addition , the spin of an electron confined to a quantum dot ( qd ) is a promising candidate for a quantum bit @xcite .
owing to the zero dimensionality of qds , the electronic orbital states are quantized and the electron spin states are very stable due to a substantial suppression of spin - flip mechanisms @xcite .
progress in nanotechnology has allowed the fabrication of qds with desirable electronic and spin properties @xcite .
however , only recently it has been possible to measure the spin of an electron in a qd .
a single electron spin has been detected by magnetic resonance force microscopy @xcite and the readout of an individual electron spin in a qd via pulsed relaxation measurements @xcite and optical orientation experiments @xcite have been reported . in these experiments ,
an external magnetic field was used to distinguish spin - up and spin - down states split by the zeeman energy .
spin relaxation measurements between zeeman levels in a qd @xcite confirm the theoretical predictions that spin - flip relaxation in a qd is suppressed with respect to a bulk structure @xcite .
indeed , very long single - spin relaxation times have been observed : up to @xmath0ms in two - dimensional ( 2d ) gaas qds @xcite , and up to @xmath1ms in self - assembled gainas qds @xcite .
the spin relaxation is expected to be dominated by hyperfine interactions with the nuclei at magnetic fields below @xmath2 t @xcite and by spin - orbit ( so ) interactions for magnetic fields of about @xmath3tesla ( see ref . ) and for higher magnetic fields ( see ref . ) . in general , the so interaction consists of two distinct contributions : the dresselhaus so coupling @xcite which is due to bulk inversion asymmetry of the lattice and the rashba so coupling @xcite which is due to structure inversion asymmetry along the growth direction .
both of these so terms result in the splitting of electron energy levels and in the mixing of the electron spin states .
the latter makes spin - flip relaxation between zeeman levels possible , for example , due to the phonon scattering .
note that usually it is not simple to separate these two so mechanisms and estimate the relative contributions of each so term . in experiments , to obtain information about one of the so couplings , normally the other is neglected @xcite .
this leads to a lack of precision in estimates of the so coupling strength and to a neglect of the effects of the interplay of the rashba and the dresselhaus so couplings @xcite .
hence , it is very important to find a way to separate these so mechanisms , to increase our understanding of the so relaxation processes , and to improve predictions of the spin properties of nanostructures .
it is well known @xcite that for 2d quantum wells the different so couplings can be distinguished experimentally @xcite via detection of the associated anisotropy of the spin splitting in the conduction band .
in contrast , such a detection is not possible in qds since the spin splitting of the levels , being quadratically in the so coupling , is isotropic .
still , as we point out now , the so couplings in qds can be distinguished via their associated spin relaxation rates since they strongly differ due to different level mixing properties . in this paper , the electron energy spectrum and the spin relaxation for a 2d qd in magnetic fields perpendicular to the qd surface are studied .
level anticrossing @xcite ( due to the so coupling ) , at a point of accidental level degeneracy ( due to the interplay between the orbital and magnetic confinement ) , is analytically investigated .
this anticrossing is caused by the rashba so term only , leading to a cusp structure in the magnetic - field dependence of the spin relaxation rate , whereas the spin relaxation rate due to the dresselhaus so coupling is a monotonic function of magnetic field in this region .
this qualitative difference in the spin relaxation for different so couplings can serve to extract the different contributions in so coupling .
we consider a 2d isotropic qd with parabolic lateral confinement potential .
an external magnetic field is applied perpendicularly to the surface of the qd .
the hamiltonian of this system reads @xmath4 where @xmath5 , @xmath6 is the vector potential in the symmetric gauge , @xmath7 is the characteristic confinement frequency , and @xmath8 is the vector of the pauli matrices . the so interaction is taken into account by adding the linear dresselhaus @xcite and rashba @xcite terms for conduction band electrons in a [ 001 ] two - dimensional electron gas ( 2deg ) , @xmath9 the axes @xmath10 , @xmath11 , and @xmath12 are aligned along the principal crystallographic axes of gaas . it is convenient to introduce new phase coordinates ( @xmath13 ) which connected to the previous ones ( @xmath14 ) by the following formula @xcite : @xmath15 where @xmath16 here @xmath17 is the cyclotron frequency . in the new phase
coordinates , @xmath18 has the canonical form @xmath19 in this case , @xmath18 can be considered as the hamiltonian of two independent harmonic oscillators with hybrid frequencies @xmath20 .
therefore , the energy spectrum and eigenstates of electrons in a qd without the so coupling are given by @xmath21 where @xmath22 , @xmath23 is the electron - spin projection on the @xmath12-axis , @xmath24 is the zeeman frequency , and @xmath25 are oscillator functions .
let us consider the three lowest levels : @xmath26 the first level is the ground state .
in the case of weak magnetic confinement ( @xmath27 ) , the second level is lower than the third one ( @xmath28 ) .
however , at high magnetic fields , when the magnetic confinement is much stronger than the lateral confinement ( @xmath29 ) , @xmath30 , because @xmath31 . the condition for a crossing of the levels @xmath32 and @xmath33 is given by @xmath34 .
in other words , this level crossing takes place when the magnetic length @xmath35 is equal to @xmath36^{1/4}$ ] , where @xmath37 is the characteristic lateral size of a qd and @xmath38 .
note that the level crossing occurs at accessible magnetic fields for qds with lateral size @xmath39 nm .
now we take so coupling into account and find the energy spectrum and eigenstates of electrons in a qd . for a gaas
qd the so lengths are @xmath40 m @xcite and are much larger than the hybrid orbital length @xmath41 of a qd ( @xmath42 ) .
therefore , the so terms can be considered as small perturbations .
first we consider the dresselhaus so coupling [ see eq .
( [ eq : hdr ] ) ] .
it is important to note that in first - order perturbation theory there is no so interaction between the levels @xmath32 and @xmath33 due to the dresselhaus term ( @xmath43 ) .
hence we can apply standard perturbation theory for nondegenerate levels .
thus , in first - order perturbation theory , we get @xmath44 , latexmath:[\ ] ] 99 _ semiconductor spintronics and quantum computing _ , edited by d.d .
awschalom , d. loss , and n. samarth ( springer , new york , 2002 ) .
d. loss and d.p .
divincenzo , phys .
a * 57 * , 120 ( 1998 ) .
d. pines , j. bardeen , and c.p .
slichter , phys .
* 106 * , 489 ( 1957 ) .
a.v . khaetskii and y.v .
nazarov , phys .
b * 61 * , 12639 ( 2000 ) . j.a . folk , s.r . patel , k.m .
birnbaum , c.m .
marcus , c.i .
duruz , and j.s .
harris , jr .
, phys . rev .
lett . * 86 * , 2102 ( 2001 ) .
t. fujisawa , y. tokura , and y. hirayama , physica b * 298 * , 573 ( 2001 ) .
t. fujisawa , d.g .
austing , y. tokura , y. hirayama , and s. tarucha , nature ( london ) * 419 * , 278 ( 2002 ) .
r. hanson , b. witkamp , l.m.k .
vandersypen , l.h.willems van beveren , j.m .
elzerman , and l.p .
kouwenhoven , phys .
lett . * 91 * , 196802 ( 2003 ) .
r. hanson , l.h .
willems van beveren , i.t .
vink , j.m .
elzerman , w.j.m .
naber , f.h.l .
koppens , l.p .
kouwenhoven , and l.m.k .
vandersypen , cond - mat/0412768 ( unpublished ) .
d. rugar , r. budakian , h.j .
mamin , and b.w .
chui , nature ( london ) * 430 * , 329 ( 2004 ) .
elzerman , r. hanson , l.h .
willems van beveren , b. witkamp , l.m.k .
vandersypen , and l.p .
kouwenhoven , nature ( london ) * 430 * , 431 ( 2004 ) .
m. kroutvar , y. ducommun , d. heiss , m. bichler , d. schuh , g. abstreiter , and j.j .
finley , nature * 432 * , 81 ( 2004 ) .
golovach , a. khaetskii , and d. loss , phys .
lett . * 93 * , 016601 ( 2004 ) .
erlingsson and y.v .
nazarov , phys .
b * 66 * , 155327 ( 2002 ) .
khaetskii , d. loss , and l. glazman , phys .
88 * , 186802 ( 2002 ) .
coish and d. loss , phys . rev .
b * 70 * , 195340 ( 2004 ) .
g. dresselhaus , phys . rev . * 100 * , 580 ( 1955 ) .
bychkov and .i .
rashba , pisma zh .
. fiz . * 39 * , 66 ( 1984 ) [ jetp lett . * 39 * , 78 ( 1984 ) ] .
j. luo , h. munekata , f.f .
fang , and p.j .
stiles , phys .
b * 41 * , 7685 ( 1990 ) . l.j .
cui , y.p .
zeng , b.q .
wang , z.p .
zhu , l.y .
lin , c.p .
jiang , s.l .
guo , and j.h .
chu , appl .
. lett . * 80 * , 3132 ( 2002 ) .
zumbhl , j.b .
miller , c.m .
marcus , k. campman , and a.c .
gossard , phys .
lett . * 89 * , 276803 ( 2002 ) .
ganichev , v.v .
belkov , l.e .
golub , e.l .
ivchenko , p. schneider , s. giglberger , j. eroms , j. de boeck , g. borghs , w. wegscheider , d. weiss , and w. prettl , phys .
92 * , 256601 ( 2004 ) .
i.l . aleiner and v.i .
falko , phys .
* 87 * , 256801 ( 2001 ) .
de andrada e silva , phys .
b * 46 * , r1921 ( 1992 ) .
b. jusserand , d. richards , g. allan , c. priester , and b. etienne , phys .
b * 51 * , r4707 ( 1995 ) .
miller , d.m .
zumbhl , c.m .
marcus , y.b .
lyanda - geller , d. goldhaber - gordon , k. campman , and a.c .
gossard , phys .
90 * , 076807 ( 2003 ) .
destefani , s.e .
ulloa , and g.e .
marques , phys .
b * 69 * , 125302 ( 2004 ) .
we neglect the cubic dresselhaus term which is important for 2d quantum wells @xcite , but for a few - electron qd with the lateral size greater than the height ( @xmath151 ) the cubic term is smaller than the linear one by a factor of @xmath152 .
galkin , v.a .
margulis , and a.v .
shorokhov , phys .
b * 69 * , 113312 ( 2004 ) .
let us consider the difference between the rashba and dresselhaus so couplings .
we can express the so terms in the form @xmath153 , where @xmath154 is an effective magnetic field .
as shown in refs . , , and , @xmath154 has different symmetries in momentum space for rashba and dresselhaus coupling .
it is this difference which is responsible that in qds the ground state mixes with different excited states : the dresselhaus term mixes the spin - down ground state @xmath155 with @xmath156 [ see eq . ( [ eq : state2d ] ) ] whereas the rashba term mixes @xmath155 with @xmath157 and @xmath158 [ see eq .
( [ eq : state2r ] ) ] .
therefore , for dresselhaus coupling the spin - down ground state is orthogonal to the first excited state @xmath157 , whereas the rashba term leads to the mixing ( and thus anticrossing ) of these states .
thornton , t. ihn , p.c .
main , l. eaves , and m. henini , appl .
phys . lett . *
73 * , 354 ( 1998 ) .
t. fujisawa , y. tokura , and y. hirayama , phys .
b * 63 * , 081304(r ) ( 2001 ) .
i. uti , j. fabian , and s. das sarma , rev .
phys . * 76 * , 323 ( 2004 ) and references therein .
k. blum , _ density matrix theory and applications _ ( plenum press , new york , 1996 ) . in the case of the rashba so coupling , our treatment is valid only if @xmath159 ( for a gaas qd with @xmath56mev and for the rashba so coupling @xmath57 m , @xmath160 ) .
otherwise the secular approximation @xcite breaks down . due to the time - reversal symmetry of the hamiltonian
the behavior of the spin relaxation rate @xmath114 does not change with the sign change of the magnetic field .
but the inversion of the sign of the @xmath161-factor leads to opposite effects : the dresselhaus so coupling now gives rise to anticrossing and to the cusplike feature of @xmath114 , while the rashba term does not . in the case of anisotropic 2d qds characterized by @xmath162 ,
there is also mixing of the states due to dresselhaus coupling , which depends on the strength of anisotropy . for weak anisotropy
@xmath163 $ ] , this leads to a narrow cusp in the spin relaxation which can be neglected with respect to that due to rashba coupling .
note that anisotropy effects in qds in strong magnetic fields are studied in ref .
falko , and b.l .
altshuler , and o. tsyplyaev , cond - mat/0501046 ( unpublished ) .
one of the possible experimental separations of the so constants @xmath164 and @xmath165 could be the following : from the width of the cusp of @xmath114 the rashba constant @xmath164 can be found . then taking into account that the relaxation rate is just a sum of the two so contributions , we can find the dreesselhaus so coupling @xmath165 in the region of magnetic field away from the cusp .
|
the spin - orbit splitting of the electron levels in a two - dimensional quantum dot in a perpendicular magnetic field is studied .
it is shown that at the point of an accidental degeneracy of the two lowest levels above the ground state the rashba spin - orbit coupling leads to a level anticrossing and to mixing of spin - up and spin - down states , whereas there is no mixing of these levels due to the dresselhaus term .
we calculate the relaxation and decoherence times of the three lowest levels due to phonons .
we find that the spin relaxation rate as a function of a magnetic field exhibits a cusplike structure for rashba but not for dresselhaus spin - orbit interaction .
|
bose - einstein condensates ( becs ) with attractive two - body interaction are dynamically unstable in a trap if the number of condensed atoms reaches a critical value @xcite .
generally , hefty becs will collapse following a switch of the s - wave scattering length from a positive value ( or zero ) to a negative one via the use of a feshbach resonance @xcite .
however it was found that in such case the number of remnant condensate atoms can be above the critical number @xcite .
this was explained by the generation of bright solitary waves with repulsive interactions @xcite .
these repulsive interactions have been used to indirectly infer a repulsive relative phase @xmath0 @xcite between neighbor solitons @xcite .
origin of such phases could be quantum mechanical phase fluctuations of the bosonic field operators @xcite due to the nonlinear modulation instability . on the other hand , quantum and thermal excitations , which can not be taken into account by simple mean - field theory , play a crucial role in the collapse of becs @xcite .
these effects have been studied by effective quantum field method @xcite .
an interesting result arising from these studies is the absence of @xmath1-phase relation as revealed recently by @xcite .
briefly , a relative phase of @xmath1 is introduced in mean - field theory between neighbor bright solitons @xcite .
usually one expects a @xmath1 relative phase to support repulsive interaction between solitons . however , there are indications that the repulsive interaction can come from quantum ( thermal ) noise rather than the repulsive phase @xcite . in addition one
also doesnot get a fixed @xmath1-phase relation within mean - field theory if becs is developed from a nonuniform initial state @xcite .
it is therefore interesting to ask what a meaningful phase would entail .
a solid starting point would be mean field theory , in which @xmath1-phase is assumed @xcite .
collisional dynamics of bright solitons in attractive becs confined in a trap is usually based on it @xcite . yet
other phase relation is argued in @xcite as well .
consequently ascertaining the phase becomes important .
in this paper , we propose a method to measure the relative phase directly rather than to infer it indirectly from soliton interactions . in the experiment @xcite , two bright solitary waves are separated in coordinate space .
this makes it hard to measure the phase by overlapping the density @xcite
. however interference appears if the wave function is transformed into momentum space @xcite .
this indicates that we can extract the relative phase from the dynamic structure factor .
the dynamic structure factor provides important information about the spectrum of collective excitations and momentum distribution @xcite . stimulated
two - photon bragg scattering has been used to measure the dynamic structure factor at high momentum transfer and provided high resolution and sensitivity @xcite .
measuring phase in momentum space is proposed by pitaevskii et al .
@xcite , who suggest to measure the coherence of two spatially separated becs confined in a double - well potential in this way .
this method can be applied for the problems we consider here as our two bright solitons are well separated in the trap @xcite .
we calculate the dynamic structure factors for two solitons with @xmath1-phase and in - phase ( @xmath2 ) using the impulse approximation @xcite .
our results show that the dynamic structure factors have distinguishable peaks corresponding to these phase relations .
the paper is organized as follows . in sec.ii
, we show that the momentum density exhibits interference due to the relative phase between two bright solitary waves in a trapped bec . the phase carried by solitary waves
is successfully recovered by a maximal response of the dynamic structure factor .
only when @xmath2 appears a single peak , which represents a large occupation of scattered atoms with the same energy as the photon .
in contrast , two peaks would be the sign of @xmath3 . in sec.iii ,
the dynamic structure factor is discussed for a case when two solitary waves move symmetrically in a trap with a small velocity .
this is the general case as they are trapped in a harmonic potential and interact repulsively .
although the shapes of the dynamic structure factor are modified by the nonzero velocity , they provide distinguishable information of the concrete values of the phase also in this case . in sec.iv
, we conclude the paper .
the gross - pitaevskii equation ( gpe ) is very effective in providing both qualitative and quantitative predictions for static and dynamical properties of an attractive bec @xcite .
we consider n weakly interacting bosons trapped in a quasi - one - dimensional potential @xmath4 , where @xmath5 and @xmath6 are the transverse and the longitudinal trap frequencies accordingly .
the dynamics of the condensate is governed by the gpe @xmath7\psi \label{3dgp}\ ] ] here @xmath8 is the effective coupling constant representing two - body interaction strength . @xmath9 and @xmath10 are s - wave scattering length and mass of atoms . the wave function of the condensate is normalized as @xmath11 . the gpe is widely accepted for studying dynamics of interacting becs at temperature @xmath12 .
explicit analytical solutions are only available in a few situations . for instance , bright soliton solutions can be derived in free space using the inverse - scattering method @xcite . vortex solutions in rotating becs are also available for analytical treatment @xcite .
it is hard to obtain the exact soliton solution in the presence of harmonic trap . in this paper
, we consider the soliton solution of a bec with a gaussian form as it is confined in a harmonic trap .
the wave function of a bec confined in an elongated axially symmetric harmonic trap is given by @xcite @xmath13 where @xmath14 is the width of condensate in the @xmath15-direction and @xmath16 are the harmonic lengths correspondingly .
note that the gaussian wave function is close to the exact soliton solution when @xmath17 is comparable with @xmath18 @xcite . with this gaussian ansatz
, the energy of each atom is obtained readily from the gpe @xmath19 here we have used @xmath20 and @xmath21 denoting the zero - point energy in a harmonic trap and two - body interaction energy . a homogeneous condensate with attractive interaction in free space is unstable while the zero - point kinetic energy can prevent the collapse of a trapped condensate with the finite number of particles @xcite .
@xmath22 where @xmath23 is a constant of order @xmath24 and @xmath25{a_xa_ya_z}$ ] is the mean harmonic length .
it implies that on one hand , at given trap frequencies and s - wave scattering length , a condensate containing a finite number of particles is stabilized by the balance of kinetic energy and negative interaction energy . on the other hand ,
the interaction energy is limited by a small value too .
this will turn out to be helpful in calculating the dynamic structure factor . in the presence of two bright solitary waves ,
the wave function is taken as @xmath26 with relative phase @xmath27 . because of repulsive interactions between two solitary waves , they will be spatially separated in a harmonic trap @xcite .
this makes it difficult to measure the phase in coordinate space , where overlapping density profiles could induce a fringe pattern @xcite .
this obstacle can be overcome if we transform the wave function into momentum space , where two spatially separated condensates will exhibit interference @xcite . and the momentum distribution can be measured by stimulated two - photon bragg scattering @xcite giving the dynamic structure factor of the system . at large momentum transfer @xmath28 , we can use the impulse approximation to calculate the dynamic structure factor @xcite @xmath29 where @xmath28 and @xmath30 are momentum and energy transfered from the laser pulse to the condensate and @xmath31 is the momentum density of the bec @xmath32 where @xmath33 is momentum distribution obtained by fourier transformation of wave function @xmath34 @xmath35 in order to obtain the dynamic structure factor , at first we need to know the momentum density distribution .
we use the gaussian ( [ gaussianform ] ) to calculate the momentum distribution . for two solitary waves moving in the @xmath36-direction ,
the wave function of the system is @xmath37 .
@xmath38 is the distance between them .
the corresponding wave function in momentum space is @xmath39 where @xmath40 reads @xmath41 \label{singlemomentum}\end{aligned}\ ] ] , dashed , @xmath3 ; dotted , a single condensate .
other parameters are , @xmath42 . here
@xmath43 and @xmath44 . ] ) but with @xmath45 relative phase deviation from @xmath46 .
dynamic structure factors shift with respect to @xmath47 .
thick curves show the result of @xmath48 and the thin ones show @xmath49 . ] for the above wave function , the momentum density exhibits interference @xmath50n_0(\textbf{p } ) \label{fringe}\ ] ] where @xmath51 .
the period of the interference pattern is @xmath52 .
obviously momentum density distribution depends on the phase @xmath27 .
for example , @xmath53 at @xmath54 but has a maximum at @xmath55 .
this implies that the relative phase @xmath27 displaces the momentum interference . in the following , we study the dynamic structure factor of the system by applying bragg spectroscopy @xcite . here
interaction effects on the relative shift of the peak are negligible ( eq .
( [ energy ] ) ) .
weak interaction is particular useful because it makes using of impulse approximation feasible at high momentum transfer @xcite . by applying a bragg pulse in the @xmath36-direction
, we obtain the dynamic structure factor from eq .
( [ simpa ] ) @xmath56 with @xmath57 . substituting eq .
( [ fringe ] ) into the above equation , we get @xmath58s_s(q , e ) \label{dsf1}\ ] ] here @xmath59 of a single solitary wave is @xmath60 eq .
( [ sdynamic ] ) has a gaussian form . like the momentum density , eq .
( [ dsf1 ] ) also shows an interference pattern .
the period of cosine function is @xmath61 at a given distance @xmath38 and optical momentum @xmath62 , the relative phase @xmath27 will shift the interference pattern of the dynamic structure factor .
consequently , we can measure the relative phase from the fringe of the dynamic structure factor .
in the experiment @xcite , the maximal distance of two solitons is about @xmath63 and the width of each soliton is about @xmath64 .
since the size @xmath17 of an attractive condensate is small , the wave length of the laser field must be far smaller than @xmath17 .
here we assume the momentum of the laser field is @xmath65 . in fig .
( [ fringegaussian ] ) we show the dynamic structure factor for two different phase relations @xmath46 .
when @xmath2 , @xmath66 is close to that of a single condensate @xcite .
in contrast to the maximum located at @xmath67 ( @xmath68 ) for @xmath2 , @xmath66 for @xmath3 it is zero there due to the interference .
but two peaks emerge symmetrically around energy @xmath67 .
these results show that the dynamic structure factor exhibits contrasting interferences for @xmath2 and @xmath3 .
as we introduced previously , the relative phase plays a crucial role in collapsed condensates @xcite . in a bose gas with attractive two - body interaction , only a finite number of atoms can be condensed @xcite .
this is true for a single condensate . however , in a fragmented condensate , for example , several bright solitary waves interacting repulsively support a bigger system . in this case , the repulsive interaction among neighboring peaks comes from the special relative phase @xmath69 according to mean field theory @xcite . on the other hand , when including quantum noise , truncated wigner quantum field simulations give @xmath2 rather than @xmath3 @xcite . since the experiments did not directly monitor the phase , a measurement proposal based on the dynamic structure factor
is established to distinguish the relative phase .
we also study the dynamic structure factor if @xmath27 is slightly deviating from @xmath2 and @xmath1 .
the deviation is @xmath70 around them . for two bright solitons ,
a relative phase in the entire range @xmath0 can stabilize the system .
such a phase deviation could be possible in the experiment .
consequently , we d like to see how a slight phase deviation affects momentum interference and whether or not we could distinguish the phase .
the results are reported in fig .
( [ fringeerror ] ) .
the dynamic structure factor remains similar except a slight displacement of the peak position . for phase shifts around @xmath2
, it only moves the maximum along the energy axis .
the dynamic structure factors are not symmetric about @xmath71 any more for phase shift about @xmath3 .
a maximum appears in the lower energy part for a shift @xmath72 while other smaller peaks appear in the high energy part .
vice versa for @xmath73 . despite these shifts ,
the resulting momentum interference could give clearly distinguishable fringes for in - phase and @xmath1-phase .
these studies demonstrate the robustness of the dynamic structure factor against phase shift around @xmath3 and @xmath2 . , dashed , @xmath3 .
other parameters are , @xmath74 . ]
thus far we considered the situation when the two solitary waves have zero velocity and are well separated .
however , the observed solitary waves move in the trap and the velocity can not be always zero .
actually , the above results assume that the two solitary waves are separated by the maximal or minimal distance . in experiment @xcite ,
two solitons move symmetrically about the trap center .
the frequencies of the oscillation depend on the external harmonic trap frequencies . in this section
we discuss the case when the two solitary waves have a small velocity oscillating in trap . the wave function for a moving solitary wave
is @xmath75 where @xmath76 is the center of mass .
this is a moving wave packet with momentum @xmath77 . transforming it into momentum space
, we get @xmath78\nonumber\end{aligned}\ ] ] where @xmath79 .
hereafter , we assume there are two waves oscillating symmetrically in the @xmath36-direction with momentum @xmath80 with respect to the trap center .
the dynamic structure factor for the two moving solitary waves can be obtained as @xmath81 where @xmath82 is the momentum density of eq .
( [ velocitymomentum ] ) in the @xmath36-directions and @xmath83 characterizes the interference of the dynamic structure factor . ) but with @xmath45 relative phase deviation from @xmath46 .
dynamic structure factors shift with respect to @xmath47 .
thick curves show the result of @xmath48 and the thin ones show @xmath49 .
other parameters are the same as in fig .
( [ velocity ] ) . ] for a rough estimate , the maximal velocity for a particle in a harmonic trap is @xmath84 , where @xmath85 is the distance to the trap center . in the experiment @xcite ,
one can get the maximal velocity if it can reach the bottom of the potential , @xmath87 .
the velocity should be less than @xmath88 because of the separation of two solitary waves with repulsive interactions .
considering this point , the dynamic structure factor with nonzero velocity is plotted in fig .
( [ velocity ] ) .
we can see that the interferences are different from the zero velocity cases ( fig .
( [ fringegaussian ] ) ) .
the maximal momentum distribution at @xmath67 is lowered for @xmath2 and the two peaks are bigger for @xmath3 relatively .
but one can still distinguish the two situations with phase @xmath89 or @xmath1 .
again , we have calculated the situation when the phase deviates from @xmath90 .
these results are shown in fig .
( [ velocityerror ] ) , a slight shift of the maximal response will not obviously reduce the contrast of the dynamic structure factor . thus one can get the phase relation from measured data .
in this paper , we have discussed the dynamic structure factor for two bright solitary waves with different relative phases in attractive becs . in mean field theory
, the phase @xmath3 protects the condensate from collapse by providing a repulsive force between solitary waves .
however , both in - phase and @xmath1-phase relations can achieve this once quantum effects are considered @xcite .
our study shows that momentum interference is modified by the phase .
the dynamic structure factor displays distinguishable fringes at @xmath2 and @xmath1 respectively . at the proper photon energy and momentum
, it has a single peak located at @xmath67 for @xmath2 while there are double peaks for @xmath3 .
the distinguishable fringes are robust even when the solitary waves have a small velocity .
@xmath91 phase deviations from @xmath2 and @xmath3 are also discussed .
our result offers possible way to directly measure the relative phase between two neighboring solitary waves in attractive becs @xcite
. an experiment using bragg scattering could provide more detail information on the relative phase .
w. li thanks d. m. stamper - kurn , m. haque , m. hussein and xiaotao xie for useful discussions as well as j. brand and s. wster for many important comments .
|
we study the dynamic structure factor of two coherent bright solitary waves in attractive bose - einstein condensates confined in a harmonic trap .
we demonstrate that the wave function of the two solitary waves with a fixed relative phase shows interference in momentum space .
the fringes are shifted depending on the values of the phase .
this momentum interference can be extracted from the dynamic structure factor of the system using stimulated two - photon bragg scattering .
thus our method provides a way to measure the relative phase directly .
|
the kondo problem the question of how an impurity local moment becomes screened at low temperatures by the conduction electrons of a host metal was highly influential in stimulating the development of theoretical and numerical methods for treating strongly correlated condensed matter @xcite .
perturbative treatments of the spin - flip scattering between local and delocalized spins necessarily break down below a characteristic kondo temperature scale , giving rise to a complex many body problem .
nonetheless , much valuable understanding of the kondo problem has come from perturbative renormalization - group ( rg ) @xcite and perturbative scaling @xcite approaches .
these were distilled into their simplest form in the poor man s scaling of anderson @xcite . in poor man
s scaling , electron states far from the fermi energy are progressively eliminated as the effective bandwidth is reduced with a compensating adjustment of a dimensionless measure of the effective impurity - band exchange coupling .
the evolution of this coupling to ever larger values with decreasing bandwidth is suggestive of approach to a fully screened strong - coupling fixed point , although the scaling approach breaks down once the effective bandwidth drops below the order of the kondo temperature .
more sophisticated but generally less intuitive methods ( the first historically being the numerical renormalization group or nrg @xcite ) were devised to confirm that the infra - red fixed point indeed corresponds to infinite exchange @xcite .
poor man s scaling was subsequently extended to the anderson model with impurity coulomb interaction @xmath5 @xcite , and the @xmath6-channel kondo model @xcite , where it correctly predicts the existence of a stable rg fixed point at an intermediate value of the exchange coupling that lies within the perturbative domain for @xmath7 .
more recently , there has been much interest in kondo physics in settings where the band density of states has a power - law variation @xmath0 in the vicinity of the fermi energy @xmath8 .
pseudogapped or soft - gap systems described by band exponents @xmath1 can be found in a variety of materials including heavy - fermion unconventional superconductors @xcite , zero - gap bulk @xcite and engineered @xcite semiconductors , and various ( quasi ) two - dimensional systems like graphite @xcite and graphene @xcite .
several theoretical techniques that have proved powerful for describing quantum impurities in metallic hosts , including bethe ansatz , bosonization , and conformal field theory , can not be applied for a power - law density of states .
however , power - law variants of the kondo impurity model and the corresponding anderson model have been extensively studied using a variety of techniques including perturbative scaling @xcite , large-@xmath9 approaches @xcite , the nrg @xcite , the perturbative rg @xcite , and the local - moment approach @xcite . due to the depletion of the conduction - band density of states near the fermi energy
, these pseudogap models feature a quantum phase transition @xcite between a local - moment phase for weak impurity - band couplings , in which the impurity spin survives unscreened down to zero temperature , and a kondo or strong - coupling phase in which the impurity spin undergoes complete or partial many - body screening ( depending on the presence or absence of particle - hole symmetry ) @xcite .
this paper addresses the question of how well poor man s scaling captures the phase boundaries in quantum impurity problems that feature both ( i ) more than one independent coupling that flows under the reduction of the effective bandwidth , and ( ii ) unstable critical points arising from competing flows in the multi - dimensional parameter space of effective couplings . to this end
, we extend the poor man s scaling approach to treat the anderson model ( for arbitrary on - site repulsion @xmath10 ) and the particle - hole - symmetric kondo model with easy - axis or easy - plane anisotropy of the impurity - band exchange coupling . in each case , the scaling equations are valid for any density of states @xmath0 , whether @xmath11 is positive , negative , or zero .
we obtain analytical expressions for the locations of phase boundaries for different regimes of the pseudogap ( @xmath1 ) anderson and power - law ( @xmath12 ) anisotropic kondo models .
comparison with nonperturbative nrg results shows that poor man s scaling correctly reproduces the power - law relations between model parameters along the phase boundaries and generally also reproduce the boundaries with good quantitative accuracy .
the rest of the paper is organized as follows .
section [ sec : anderson ] treats the anderson model with a power - law density of states .
section [ subsec : model : a ] defines the model and summarizes the phase diagram that has been established through previous work .
the poor man s scaling equations are derived in sec . [ subsec : poor man : a ] .
section [ subsec : r>1 ] compares analytic approximations for the phase boundary with nrg results for superlinear ( @xmath3 ) densities of states and various regimes of the other model parameters , while sec .
[ subsec : r<1 ] does the same for @xmath13 .
limitations of the poor man s scaling approach are identified in sec .
[ subsec : limitations : a ] .
the anisotropic kondo model is the subject of sec .
[ sec : kondo ] .
section [ subsec : scaling : k ] presents the poor man s scaling equations along with a preliminary analysis .
phase boundaries are analyzed for @xmath14 and @xmath15 in secs .
[ subsec : pseudogap : k ] and [ subsec : divergent : k ] , respectively .
section [ sec : summary ] contains a brief summary .
the anderson impurity model is described by the hamiltonian @xcite @xmath16 where @xmath17 with @xmath18 ( or @xmath19 , @xmath20 ) describes a noninteracting conduction band having dispersion @xmath21 ; @xmath22 with @xmath23 and @xmath24 describes an impurity having level energy @xmath25 and on - site coulomb interaction @xmath10 ; and the hybridization term @xmath26 accounts for impurity - band tunneling .
@xmath27 is the number of unit cells in the host metal and , hence , the number of inequivalent @xmath28 values . without loss of generality , we take the hybridization matrix element @xmath29 to be real and non - negative . for compactness of notation , we drop all factors of the reduced planck constant @xmath30 , boltzmann s constant @xmath31 , and the impurity magnetic moment @xmath32 . a mapping to an energy representation where @xmath33 shows that the conduction - band dispersion @xmath34 and the hybridization matrix element @xmath29 affect the impurity degrees of freedom only in combination through the hybridization function @xcite @xmath35 to focus on the most interesting physics of the model , we assume a simplified form @xmath36 where @xmath37 is the heaviside function
. the primary focus of this work is cases @xmath1 in which the hybridization function exhibits a power - law pseudogap around the fermi energy .
we will also briefly discuss @xmath38 , representing a conventional metallic host .
one way that a hybridization function of the form of eq . can arise is from a purely local hybridization matrix element @xmath39 combined with a density of states ( per unit cell , per spin orientation ) varying as @xmath40 with @xmath41 , in which case @xmath42 .
however , all results below apply equally to situations in which the @xmath28 dependence of the hybridization contributes to the energy dependence of @xmath43 .
the assumption that @xmath43 exhibits a pure power - law dependence over the entire width of the conduction band is a convenient idealization .
more realistic hybridization functions in which the power - law variation is restricted to a region around the fermi energy exhibit the same qualitative physics , with modification only of nonuniversal properties such as critical couplings and kondo temperatures .
schematic @xmath25-@xmath44 phase diagrams of the pseudogap anderson model [ eqs .
] for fixed @xmath45 and band exponents ( a ) @xmath46 , ( b ) @xmath47 .
generically , the system falls into either a local - moment phase ( lm ) or one of two asymmetric strong - coupling phases ( asc@xmath48 ) .
however , there is also a symmetric strong - coupling phase ( the line labeled ssc ) that is reached only for @xmath14 under conditions of strict particle - hole symmetry ( @xmath49 ) and for sufficiently large hybridization widths @xmath44 . ] in the metallic ( @xmath38 ) anderson model , any value @xmath50 places the system in its strong - coupling phase , where the impurity degrees of freedom are completely quenched at absolute temperature @xmath51 .
the situation for pseudogapped hybridization functions ( @xmath1 ) is much richer , as summarized in the phase diagrams shown in fig .
[ fig : pam_phase ] for cases @xmath45 of onsite coulomb repulsion .
the most notable feature is the existence within a region @xmath52 , @xmath53 of a local - moment ( lm ) phase within which the impurity retains an unquenched spin degree of freedom down to @xmath51 .
there are also three different strong - coupling phases , distinguished by their ground - state electron number @xmath54 ( measured from half filling ) : a symmetric strong - coupling ( ssc ) phase with @xmath55 , reached only for @xmath14 under the condition @xmath49 for strict particle - hole symmetry ; and a pair of asymmetric strong - coupling phases asc@xmath56 and asc@xmath57 having @xmath58 and @xmath59 , respectively . for a compact summary of the ground - state properties in each phase of the pseudogap anderson impurity model and the nature of the quantum phase transitions separating those phases , the reader is referred to sec .
ii b1 of ref . .
this section presents a poor man s scaling treatment of the anderson hamiltonian with a power - law hybridization function .
jefferson @xcite and haldane @xcite provided scaling treatments of the metallic case @xmath38 valid in the limit @xmath60 .
these were subsequently extended to general values of @xmath10 ( ref . ) , although the analysis neglected the renormalization of @xmath10 .
reference presented scaling equations for the pseudogap case @xmath1 with @xmath5 .
below , the scaling analysis is generalized to arbitrary values of @xmath11 and @xmath10 .
two of us have previously presented a similar poor man s scaling analysis of the anderson - holstein impurity model with a power - law hybridization @xcite .
the treatment of the anderson model is somewhat simpler , and as we will see , the resulting scaling equations are amenable to approximate integration is several physically interesting limits .
we start with the anderson hamiltonian written in the form @xmath61 where @xmath62 and @xmath63 are as defined in eqs . and , respectively , but with @xmath64 in eq . replaced by @xmath65 d^{\dag}_{{\sigma } } c^{{\phantom{\dag}}}_{{\mathbf{k}}{\sigma } } + \text{h.c . }
\bigr\},\end{aligned}\ ] ] with hybridization functions @xmath66 for @xmath67 , @xmath68 having the same power - law dependence as @xmath43 defined in eq . . at the bare hamiltonian level
, one expects the hybridization @xmath69 between the empty and singly occupied impurity configurations to be identical to that @xmath70 between the singly occupied and doubly occupied impurity configurations .
however , this degeneracy might be broken under the scaling procedure .
following haldane @xcite , we focus on many - body states @xmath71 , @xmath72 , and @xmath73 formed by combining the conduction - band ground state ( having @xmath27 electrons of energy @xmath74 ) with one of the four possible configurations of the impurity level . neglecting for the moment the effect of the hybridization ( @xmath75 ) ,
the energies of these states are denoted @xmath76 , @xmath77 , and @xmath78 .
we now consider the effect of an infinitesimal reduction in the half - bandwidth from @xmath79 to @xmath80 , where @xmath81 .
the goal is to write a new hamiltonian @xmath82 similar in form to @xmath83 but retaining only conduction - band degrees of freedom having energies @xmath84 and with parameters @xmath85 , @xmath86 , and @xmath87 adjusted to account perturbatively for the band - edge states that have been eliminated .
let @xmath88 be the set of wave vectors @xmath28 describing particle - like states having energies @xmath89 , and @xmath90 be the set of wave vectors describing hole - like state with energies @xmath91 .
virtual tunneling of an electron from a @xmath90 state into the empty impurity level transforms the state @xmath71 to @xmath92 with energy @xmath93 here , @xmath94 schematically represents all processes involving a product of least @xmath6 factors @xmath95 .
similarly , virtual tunneling of an electron from the doubly occupied impurity level into a @xmath88 state transforms @xmath96 to @xmath97 with energy @xmath98 finally , virtual tunneling of an electron into the singly occupied impurity from a @xmath90 state or from the singly occupied level into a @xmath88 state transforms @xmath99 to @xmath100 with energy @xmath101 + o(v^3).\end{aligned}\ ] ] the @xmath102 terms in each of the above states @xmath103 include terms to enforce normalization , i.e. , @xmath104 . the modified energies can be used to define effective hamiltonian parameters @xmath105 and @xmath106 . at the same time , for each @xmath28 in the retained portion of the band ( i.e. , satisfying @xmath84 ) , @xmath69 must be replaced by @xmath107 - \sqrt{n_k } \ : \langle \tilde{{\sigma}}| c_{{\mathbf{k}}{\sigma}}^{\dag } { \hat{h}}'_a|\tilde{0}\rangle & \text{for\ , } { { \varepsilon}_{{\mathbf{k}}}}<0 , \end{cases}\ ] ] and @xmath70 must be replaced by @xmath108 \sigma\sqrt{n_k } \ : \langle \tilde{2}| c_{{\mathbf{k}},-{\sigma}}^{\dag } { \hat{h}}'_a|\tilde{{\sigma}}\rangle & \text{for\ , } { { \varepsilon}_{{\mathbf{k}}}}<0 .
\end{cases}\ ] ] it is straightforward to show that @xmath109 the leading corrections to @xmath87 involve numerous terms arising from the @xmath110 terms in the states @xmath103 . since these corrections are too small to be of much practical importance , we shall not evaluate them here .
the infinitesimal band - edge reduction described in the previous paragraphs can be carried out repeatedly to reduce the half - bandwidth by a finite amount from @xmath79 to @xmath111 .
equations and indicate that during this process , the impurity level energy evolves according to the scaling equation @xmath112 + o(v^3 ) , \ ] ] where @xmath113 is the rescaled hybridization function at the reduced band edges @xmath114 .
taking into account eq .
as well , one sees that the on - site repulsion follows @xmath115 + o(v^3 ) .\end{aligned}\ ] ] the band - edge values @xmath113 of the hybridization functions
@xmath116 rescale both due to the replacement of @xmath79 by @xmath117 in eq . and due to the perturbative corrections to @xmath118 in eq .
, leading to the scaling equation @xmath119 the scaling equations have been derived to lowest order in nondegenerate perturbation theory , and are strictly valid only so long as @xmath120 for each @xmath28 such that @xmath121 .
equation shows that the band - edge values of the hybridization functions @xmath116 are irrelevant ( in the rg sense ) for @xmath1 and at most marginally relevant or marginally irrelevant for @xmath38 . for the particle - hole - symmetric bare hybridization functions considered in this work , it is an excellent approximation to set @xmath122 , leading to the simplified scaling equations @xmath123 , \\
\label{u : scaling } \frac{d{\tilde{u}}}{d{\tilde{d } } } & \simeq \frac{2{\tilde{{\gamma}}}}{\pi } \biggl [ \frac{1}{{\tilde{d}}+{\tilde{{\varepsilon}}_d } } - \frac{1}{{\tilde{d}}-{\tilde{{\varepsilon}}_d } } \notag \\ & \qquad + \frac{1}{{\tilde{d}}-{\tilde{u}}-{\tilde{{\varepsilon}}_d } } - \frac{1}{{\tilde{d}}+{\tilde{u}}+{\tilde{{\varepsilon}}_d } } \ , \biggr ] .\end{aligned}\ ] ] equations with initial conditions @xmath124 , @xmath125 , and @xmath126 represent the main results of this section . the equations respect particle - hole symmetry in that @xmath127 so bare couplings satisfying @xmath49 inevitably lead to rescaled couplings that satisfy @xmath128 .
for @xmath38 , eqs. reproduce the scaling equations for the metallic anderson problem @xcite , while for @xmath1 in the limit @xmath129 of extreme particle - hole asymmetry , eqs . and reduce to ones presented previously @xcite for pseudogapped systems .
equation clearly has the solution @xmath130 substituting this expression for @xmath131 into eqs . and creates a pair of coupled differential equations for @xmath85 and @xmath86 .
analytical or numerical integration of these differential equations allows one to follow the evolution of the rescaled couplings under reduction of @xmath117 until one of the following conditions is met , signaling entry into a low - energy regime governed by a simpler effective model than the full pseudogap anderson model : \(1 ) if @xmath132 , the system enters the empty - impurity region of the asc@xmath57 strong - coupling phase , in which the ground - state impurity occupancy approaches zero . in this case , @xmath133 marks the crossover scale into a low - energy regime of ( for @xmath1 , generalized ) fermi - liquid behavior .
\(2 ) if @xmath134 , the system enters the full - impurity region of the asc@xmath56 strong - coupling phase , in which the ground - state impurity occupancy approaches two . here ,
@xmath135 marks crossover into the asymptotic fermi - liquid regime .
\(3 ) if @xmath136 , the system crosses over into an intermediate - energy local - moment regime .
( this is distinct from the lm phase , in which an uncompensated spin-@xmath137 degree of freedom survives at asymptotically low energies . ) on entry to this regime , the empty and doubly occupied impurity configurations are effectively frozen out , and one can perform a generalization @xcite of the schrieffer - wolff transformation @xcite to map the pseudogap anderson model to a pseudogap kondo model @xmath138 c_{{\mathbf{k}}'{\sigma}'}^{{\phantom{\dag } } } \,\ ] ] where @xmath62 is as given in eq . with the power - law density of states specified in eq .
, @xmath139 is the spin-@xmath137 operator for the impurity , @xmath140 is a vector of pauli matrices , the ( isotropic ) exchange coupling @xmath141 satisfies @xmath142 and the potential scattering @xmath143 satisfies @xmath144 for metallic hosts ( @xmath38 ) the system always lies in the strong - coupling phase of the kondo model , which constitutes another region of the strong - coupling phase of the anderson model . in pseudogap cases ,
the asymptotic low - energy behavior depends on the values of @xmath141 and @xmath143 : the system may lie in one of three kondo phases that are associated with the ssc ( for @xmath145 ) , asc@xmath57 ( for @xmath146 ) , or asc@xmath56 ( for @xmath147 ) phases of the anderson model ; or it may be fall in the lm phase of both the kondo and anderson models , in which the impurity retains a free two - fold spin degree of freedom down to absolute zero . in any of these cases , the energy scale @xmath148 for crossover into the asymptotic low - energy regime
is generally much smaller than the scale @xmath149 for entry into the local - moment regime . on approach to a strong - coupling ground state
, @xmath148 is just the kondo temperature @xmath150 .
\(4 ) if @xmath151 ( a situation that arises only if the bare @xmath10 is negative ) , then the system enters the intermediate - energy local - charge regime . at this point
, one can perform a generalized schrieffer - wolff transformation to a pseudogap charge - kondo model .
the system may lie in a strong - coupling phase of the charge - kondo model ( yet another region of an anderson - model strong - coupling phase ) or in the local - charge phase of both models , where the impurity retains a free two - fold charge degree of freedom down to absolute zero .
similarly to the situation in ( 3 ) , @xmath152 .
\(5 ) if @xmath153 and/or @xmath154 , then the system enters a mixed - valence regime where the states @xmath155 , @xmath156 , and @xmath157 are no longer all well - defined .
the scaling method is unable to determine whether further reduction of the effective half - bandwidth will bring the system into the mixed - valence regime of the strong - coupling phase or it will instead reach one or other of the phases with a residual localized degree of freedom @xcite . in the remainder of sec .
[ sec : anderson ] , we specialize to ranges of the band exponent @xmath1 and the bare parameters @xmath10 ( henceforth taken to be positive , representing on - site coulomb repulsion ) , @xmath25 , and @xmath44 for which it possible to make analytical predictions for the location of boundaries between lm and strong - coupling phases .
we compare these predictions with results obtained using the non - perturbative numerical renormalization group ( nrg ) method @xcite , as adapted to treat systems containing a pseudogap density of states @xcite . throughout the paper ,
we have used wilson s discretization parameter @xmath158 and kept up to 600 many body states after each iteration of the nrg .
analysis of band exponents in the range @xmath3 is simplified because eq .
means that @xmath159 decreases monotonically under reduction of the half - bandwidth . in the physically most relevant regime @xmath160 ,
this decrease in @xmath161 rules out the possibility of entry into the mixed - valence regime under condition ( 5 ) of sec.[subsec : poor man : a ] .
moreover , the decrease of @xmath131 is so rapid that any entry to the local - moment regime and subsequent mapping to the pseudogap kondo problem [ via eqs . and ]
will yield a sub - critical exchange coupling @xcite . under these circumstances , the upper critical level energy @xmath162 separating the asc@xmath57 phase ( in which @xmath163 is satisfied at sufficiently low @xmath117 ) from the lm phase ( in which one eventually reaches @xmath164 ) is effectively determined by the condition @xmath165 that places the fully renormalized impurity level precisely at the the fermi energy .
this picture of the quantum phase transition as arising from a renormalized level crossing is consistent with the observation of first - order behavior for @xmath3 @xcite . by particle - hole symmetry ,
the boundary between the lm and asc@xmath56 phases is at the lower critical level energy @xmath166 [ see fig.[fig : pam_phase](b ) ] .
the aforementioned boundary between the lm and asc@xmath57 phases can be located by performing an approximate integration of eqs . and eq . using eq . .
for uniformity of presentation , we express our result in the form of a critical hybridization width @xmath167 .
we will consider bare parameters satisfying @xmath168 and ( for reasons that will become clear shortly ) @xmath169 .
cases @xmath170 and @xmath60 will be considered separately . if we suppose that @xmath170 , then so long as @xmath173 , eq .
can be approximated by @xmath174 where use has been made of eq . .
this differential equation can be integrated to yield @xmath175 , \ ] ] which for @xmath176 describes a very weak downward renormalization of @xmath86 with decreasing @xmath117 . during the same initial phase of the scaling ,
. can be approximated by @xmath177 and hence @xmath178 .\ ] ] equations and imply that @xmath179 in the case of present interest where @xmath180 , the level energy scales upward in absolute terms by one - quarter the amount that the on - site interaction scales down , but @xmath25 experiences a much greater fractional shift than @xmath10 .
equations , , and remain valid until @xmath181 rises to approach unity , a condition that occurs [ for the assumed ordering of the bare parameters , and for the weak renormalization of @xmath10 that holds for @xmath169 ] at @xmath183 , at which point @xmath184}{(r-1)\pi d } .\ ] ] in the regime @xmath185 , the doubly occupied impurity configuration is essentially frozen out . now eq .
can be approximated by @xmath186 which has the solution
@xmath187 \biggr[1 + o\biggl ( \frac{{\tilde{{\varepsilon}}_{d,1}}}{u } \biggr)\biggr].\ ] ] using eq .
, this gives @xmath188.\ ] ] a more careful treatment of scaling over the range of @xmath117 in which @xmath189 [ invalidating the nondegenerate perturbation theory used to derive eqs .
] would likely modify the numerical prefactor of @xmath190 on the right - hand side of eq . .
with this caveat , the equation should capture the scaling of the impurity level energy until @xmath191 grows to reach 1 at some reduced half - bandwidth @xmath192 .
for @xmath193 , the system crosses over into the low - energy regime of the asc@xmath57 phase ( for @xmath194 ) or that of the lm phase ( for @xmath195 ) .
the only exception occurs for a combination of bare parameters that places the system precisely on the boundary between the two phases , in which case @xmath165 . recalling that we are considering cases @xmath3 , eq .
shows that the boundary location @xmath196 is primarily determined by initial phase of scaling ( @xmath197 ) , and to leading order in @xmath198 and @xmath199 satisfies @xmath200 this relation can be recast as @xmath201 for @xmath202 .
the phase boundary between the lm and asc@xmath57 phases at @xmath167 can be determined to the desired accuracy by performing successive nrg runs to refine the value of @xmath203 using the method of bisection . at the end of each run ,
the zero - temperature limit of @xmath204 ( temperature times the impurity contribution to the static magnetic susceptibility ) @xcite is used to determine whether the system is in the lm phase ( @xmath205 ) or in the asc@xmath57 phase ( @xmath206 ) , and thus to modify the range of @xmath44 values within which @xmath203 must lie .
critical hybdrization width plotted as @xmath207 vs @xmath198 , comparing nrg data ( symbols ) with the scaling predictions of eq .
( dashed lines ) and eq .
( dotted lines ) .
results are for @xmath208 and @xmath209 and for three values of @xmath210 shown in the legend . ]
figure [ fig : r_gt_1 ] shows the critical hybridization width plotted as @xmath211 vs @xmath198 for @xmath212 and @xmath209 .
nrg results ( symbols ) calculated for three different fixed ratios @xmath213 are compared with the scaling expression in eq .
( dashed lines ) .
the numerics confirm the predicted linear dependence of @xmath203 on @xmath25 .
the @xmath214 variation of @xmath203 is also well supported for @xmath215 , but for @xmath208 is only approached in the limit of very small @xmath198 . this deviation from the scaling theory
likely arises because the latter band exponent lies close to the value @xmath216 that acts as an upper critical dimension for the pseudogap anderson and kondo models and at which there are logarithmic corrections to simple power - law behaviors @xcite .
results for @xmath217 ( not shown ) indicate that increasing @xmath11 leads to a continuous improvement in the accuracy with which eq .
reproduces the nrg data . if the bare parameters of the anderson hamiltonian instead satisfy @xmath60 , then eq
. applies from the outset of scaling , and @xmath85 satisfies eq . with @xmath219 and @xmath220 , i.e. , @xmath221 .\ ] ] now the condition @xmath165 places the lm - asc@xmath57 phase boundary at @xmath222 figure [ fig : r_gt_1 ] compares the prediction of eq .
( dotted lines ) with nrg results obtained for @xmath223 .
the scaling approach reproduces the numerical results very well for @xmath224 , and ( in contrast to the behavior found for @xmath170 ) there is no significant difference between @xmath208 and @xmath215 in the accuracy of the analytical results . for @xmath226 , eq .
implies that @xmath227 .
the system flows to mixed valence [ under condition ( 5 ) in sec.[subsec : poor man : a ] ] at a reduced half - bandwidth @xmath228 provided that @xmath229 and @xmath230 both remain smaller than @xmath231 . however , the system flows to a different low - energy regime if @xmath191 or @xmath232 reaches @xmath68 at some @xmath233 .
we first consider cases @xmath49 where the system exhibits strict particle - hole symmetry , and focus on the universal ( large - bandwidth ) limit @xmath234 .
so long as @xmath235 , eq .
can again be approximated by eq .
, which can be integrated to yield @xmath236 .\ ] ] eq . can be re - expressed as @xmath237 in terms of new variables @xmath238 and @xmath239 that allow eq . to be recast exactly in the form @xmath240 equation shows that with increasing @xmath241 ( or decreasing @xmath117 ) , @xmath242 initially rises , before peaking at @xmath243 , and then dropping off exponentially for @xmath244 .
the system will enter its local - moment regime [ under condition ( 3 ) in sec .
[ subsec : poor man : a ] ] if there exists a reduced half - bandwidth @xmath245 such that @xmath246 .
the approximate scaling equation is valid only so long as @xmath247 .
equation predicts that @xmath86 experiences a stronger downward renormalization once @xmath248 approaches @xmath209 , a range in which the nondegenerate perturbation theory used to derive eqs. also begins to break down .
however , in this range of @xmath242 , physically one expects renormalization to slow to a halt as charge fluctuations are progressively frozen out .
therefore , in the spirit of haldane @xcite , we apply eq .
all the way to the point where @xmath249 , and we seek @xmath250 defined to be the smallest solution of @xmath251 ratio @xmath252 of the critical coupling found using nrg to the scaling prediction given by eq .
, plotted as a function of @xmath198 for @xmath253 and band exponents @xmath254 , @xmath255 , and @xmath256 . ] for @xmath257 , eq .
has no real solution , so the system necessarily crosses over to mixed valence for @xmath258 . for @xmath259 , by contrast , eq .
has a solution @xmath260 satisfying @xmath261 . since @xmath262 , eq .
gives @xmath263 , meaning that at @xmath264 the system satisfies condition ( 3 ) for crossover into its local - moment regime .
equation gives @xmath265^{1/(1-r ) } \ge e^{-1/(1-r)}\ ] ] since @xmath266 .
this implies , at least for @xmath267 , that the rescaled on - site interaction @xmath268 remains of the same order as @xmath10 . a schrieffer - wolff transformation performed at @xmath264 yields a pseudogap kondo model with [ eqs . and ]
@xmath269 it is known that for @xmath270 , the critical exchange coupling @xmath271 separating the kondo ( @xmath272 ) and lm ( @xmath273 ) phases satisfies @xmath274 where @xmath275 for @xmath276 [ refs . and ] and @xmath277 for @xmath278 [ ref . ] . combining this information with eq .
, one arrives at the prediction that the boundary between the lm and ssc phases is determined by the condition @xmath279 .
then , eq . gives @xmath280 $ ] and , hence , @xmath281 $ ] .
this means that the lm phase occupies the parameter range @xmath282 , where @xmath283}{f(r ) } \ : \frac{4{\gamma}}{\pi d } \ : \exp\biggl(-\frac{4{\gamma}}{\pi d}\biggr ) \biggr\}^{1/(1-r ) } .\ ] ] for @xmath284 , one can invert eq . to deduce that the lm phase occupies the parameter range @xmath285 , with @xmath286 } \ : \biggl(\frac{u}{2d}\biggr)^{1-r}.\ ] ] if one were to neglect the rescaling of the on - site coulomb interaction and set @xmath287 in eq .
, one would instead obtain an equation for @xmath203 identical to eq . apart from the omission of the factor of @xmath288 $ ] from the right - hand side .
thus , the leading effect of many - body renormalization of the on - site coulomb repulsion is the reduction of @xmath203 by a factor that can be approximated as @xmath289 for @xmath267 .
figure [ fig : r<1:phsym ] plots the ratio of the critical hybridization width @xmath290 found using nrg to the scaling prediction @xmath291 given by eq . . for band exponents
@xmath254 , @xmath255 , and @xmath256 , we find that this ratio is well converged for @xmath292 which implies that the scaling analysis correctly captures the @xmath293 dependence of @xmath203 at the lm - ssc phase boundary .
the absolute value of @xmath294 falls as @xmath11 decreases , and seems likely to approach unity as @xmath295 .
we infer that eq .
describes the nrg results apart from a correction factor that depends solely on the band exponent @xmath11 .
( color online ) ratio @xmath296 of the mixed - valence threshold hybridization found using nrg and that given by poor man s scaling .
results for fixed @xmath297 are plotted vs band exponent @xmath11 with @xmath298 defined using four different values of @xmath299 ( see text for details ) . for @xmath300 [ @xmath301
, it proves impossible to find kondo - regime behavior for @xmath302 [ @xmath303 . ]
poor man s scaling can not only find the lm - ssc phase boundary at @xmath304 , but can also locate a crossover within the ssc phase at @xmath305 between a kondo regime , in which only the singly occupied impurity configurations have significant occupation at low temperatures , and a mixed - valence regime having significant ground - state occupancy of the empty and/or doubly occupied impurity configuration(s ) .
we have seen [ after eq . ] that the system reaches mixed valence for @xmath306 ( equivalent to @xmath307 , a mixed - valence threshold hybridization ) , and argued [ before eq . ]
that it enters the lm phase for @xmath308 $ ] .
therefore , the system exhibits fully developed kondo physics ( i.e. , enters its local - moment regime at intermediate values of @xmath117 and then crosses over to the ssc ground state for @xmath309 ) only for @xmath310 , equivalent to the condition @xmath311 with @xmath312 / f(r ) .\ ] ] the kondo regime has a width @xmath313 that narrows rapidly with increasing @xmath11 and turns out to be restricted to @xmath314 for @xmath315 . for band exponents in the range @xmath316 , the ssc phase can be accessed only from mixed valence , while for @xmath47 this phase disappears altogether @xcite . within the nrg approach
, we can define the mixed - valence threshold hybridization by examining the temperature dependence of the impurity contribution to the magnetic susceptibility @xmath317 .
we can identify the anderson model as being in its local - moment regime if @xmath318 where @xmath299 is a ( somewhat arbitrary ) cutoff chosen to lie between the value @xmath319 corresponding to a free spin-@xmath137 degree of freedom and the high - temperature or mixed - valent limiting value @xmath320 . with this criterion ,
the system is in the kondo regime of the ssc phase if with decreasing @xmath321 , @xmath204 first rises above @xmath299 before dropping towards its ssc value @xcite of @xmath322 .
we therefore define @xmath323 as the smallest hybridization prefactor for which @xmath324 at all temperatures .
figure [ fig : r<1:ph - mv ] shows the ratio @xmath325 between the mixed - valence threshold coupling found using nrg and the scaling prediction of eq .. the ratio is plotted vs band exponent @xmath11 for fixed @xmath326 and four different cutoffs : @xmath327 and @xmath328 .
as one would expect , increasing the value of @xmath299 creates a more stringent criterion for the identification of kondo physics , reduces the range of exponents @xmath11 over which kondo - regime behavior is found , and for given @xmath11 reduces the value of @xmath329 .
however , the ratio @xmath330 is nearly independent of @xmath11 except in the case @xmath331 .
this confirms that the condition for reaching mixed valence is correctly captured by eq .
apart from a multiplicative factor that depends on the value of the cutoff @xmath299 .
we now turn to the limit @xmath168 of strong particle - hole asymmetry on the impurity site . in order to locate the boundary between the lm and asc@xmath57 phases
, we will perform an approximate integration of eqs . and eq . using eq . .
below we consider @xmath172 .
the situation where @xmath332 will be considered in the next section .
so long as @xmath173 , eq .
can once more be approximated by eq .
, leading to eq . , and eq . can again be approximated by eq . , which yields @xmath333 .\
] ] equations and imply that @xmath334 equations , , and remain valid until either @xmath161 reaches @xmath68 at @xmath335 or @xmath181 reaches @xmath68 at @xmath336 . by writing @xmath337 , then employing eqs . , , and
, the latter condition can be recast as @xmath338^{1-r } = 1,\ ] ] with @xmath241 and @xmath339 as defined in eqs . and , and @xmath340 \simeq 1 .\ ] ] given eq . ,
the conditions @xmath341 are satisfied provided that eq .
has a real solution @xmath342 .
such solutions exist for @xmath343^{1-r}$ ] . for @xmath344 ,
there is a monotonic decrease in @xmath345 with increasing @xmath11 , from @xmath346 to @xmath347 , while the solution to eq .
satisfies @xmath348 . in the regime @xmath185 , entered with @xmath349 , the doubly occupied impurity configuration is essentially frozen out . now eq .
can be approximated by eq .
, again yielding eq . .
this second phase of the scaling continues until one of the following conditions is met : ( a ) @xmath163 , signalling crossover into the empty - impurity region of the asc@xmath57 phase ; ( b ) @xmath350 , marking entry into the mixed - valence region of the asc@xmath57phase ; ( c ) @xmath164 , marking entry into the local - moment regime . in case ( c )
, the system may be mapped onto the pseudogap kondo hamiltonian described by eqs .
, which may lie in ( c)(i ) the asc@xmath57 phase , or ( c)(ii ) the lm phase . integrating the poor man s scaling equations with sufficient accuracy to distinguish among all these possibilities is in general a formidable challenge .
progress on locating the lm - asc@xmath57 phase boundary can be made in the limit @xmath351 of very weak impurity - band hybridization , where @xmath352 and @xmath353 $ ] .
focusing for simplicity on @xmath354 , one finds @xmath355 and hence [ via eq .
] @xmath356 .\ ] ] in this limit of small @xmath339 , one expects only a small fractional change in the bare level energy @xmath25 to be required to drive the system from case ( a ) to case ( c)(ii ) of the previous paragraph . under these circumstances , just as was done with greater rigor for @xmath3 , one can approximate the location of the phase boundary by the condition @xmath165 , leading to @xmath357 eq .
can be inverted such that the system is in the lm phase if @xmath358 , where the critical coupling is given by @xmath359 for @xmath60 , the evolution of @xmath85 with @xmath25 is as described by eq .. for @xmath351 , arguments similar to those given at the end of the previous section lead to the conclusion that the lm - asc@xmath57boundary is given by eq . .
( color online ) critical coupling @xmath203 found using nrg ( symbols ) plotted as @xmath360 against @xmath198 for @xmath361 and @xmath362 and for @xmath363 , and @xmath256 .
also plotted are the poor man s scaling prediction of eq . for @xmath170 ( dashed lines ) and
eq . for @xmath60 ( dotted lines ) . ]
figure [ fig : r<1:phasym ] shows the critical hybridization width plotted as @xmath364 vs @xmath198 for band exponents @xmath363 and @xmath256 and for two values of the ratio @xmath365 listed in legend .
nrg data ( symbols ) are compared with the poor man s scaling predictions . for @xmath170 ,
@xmath366 exhibits a @xmath367 dependence that is described very well by eq .
( dashed lines ) apart from an overall multiplicative factor that grows with increasing @xmath11 . for @xmath60 , is almost ( but not quite ) a constant as predicted by eq .
( dotted lines ) .
these behaviors show that the poor man s scaling approach provides a good account of the phase boundary in the limit of strong particle - hole asymmetry on the impurity site .
the results presented above demonstrate that poor man s scaling successfully captures many aspects of the pseudogap anderson model .
however , it would be remiss to gloss over two limitations of the approach .
first , a reliable calculation of the critical hybridization based on poor man s scaling alone is possible for all @xmath1 only for @xmath368 ( on the asc@xmath57 side ) or @xmath369 ( on the asc@xmath56 side ) . near the particle - hole symmetric point @xmath49
, the method breaks down for @xmath370 .
this is clear to see for @xmath371 because @xmath167 diverges as @xmath372 [ see fig.[fig : pam_phase](b ) ] and therefore any phase boundary lies outside the perturbative regime . for @xmath373 , @xmath167 remains finite for all @xmath52 [ see fig . [
fig : pam_phase](b ) ] but , as discussed in ref . and in sec .
[ subsec : r<1 ] above , the strong - coupling phases are accessed directly from mixed valence , and in such cases we have been unable to find a scaling criterion for locating the phase boundaries .
second , poor man s scaling does not appear to be capable of reproducing the full rg fixed - point structure of the model as identified using the nrg @xcite .
scaling eq . and
its counterpart @xmath374 for the potential scattering in the pseudogap kondo model both indicate that particle - hole asymmetry is an irrelevant perturbation about the symmetric plane @xmath128 .
this is consistent with nrg results band exponents on the range @xmath375 , where a single particle - hole - symmetric quantum critical point ( qcp ) governs the physics all over the phase boundary between the lm and strong - coupling phases shown in fig.[fig : pam_phase](a ) .
however , there also exists a range @xmath376 in which the boundaries between the lm phase and each strong - coupling phase ( ssc , asc@xmath57 , and asc@xmath56 ) is governed by a different qcp . within this second range of band exponents , poor man
s scaling can not reproduce the fact that particle - hole asymmetry is a relevant perturbation that causes flow from the symmetric qcp to one or other of the two asymmetric qcps [ as illustrated schematically for the pseudogap kondo model in fig .
16(b ) of ref . ] .
this is a quite subtle aspect the pseudogap kondo and anderson models that even much more sophisticated rg treatments are unable to fully capture @xcite .
in this section , we present a poor man s scaling analysis of the phase boundary between the kondo and local - moment ( lm ) phases of the kondo model with distinct longitudinal and transverse spin - flip couplings between the impurity and a power - law - vanishing or -divergent density of states .
the model is described by the hamiltonian @xmath377 where @xmath62 is as given in eq . with the density of states specified in eq . , and @xmath139 and @xmath378 ( with @xmath140 being a vector of pauli matrices ) are , respectively , the spin-@xmath137 operators for the impurity and for conduction band electrons at the impurity site .
the properties of the model are invariant under @xmath379 , but for notational simplicity we will consider only @xmath380 .
our focus is primarily on pseudogap cases @xmath1 , which can arise , for example , due to the low - temperature freeze - out of charge fluctuations in the anderson - holstein model with a power - law density of states @xcite .
however , in sec .
[ subsec : divergent : k ] we briefly consider the range @xmath15 describing bands with a generalized van hove singularity at the fermi energy @xcite . by generalizing anderson s poor man s scaling treatment of the conventional ( @xmath38 ) kondo problem @xcite ,
it is straightforward to extend withoff and fradkin s analysis of the pseudogap kondo problem to anisotropic exchange . under progressive reduction of the half - bandwidth from @xmath79 to @xmath381
, the exchange couplings @xmath382 evolve to @xmath383 according to [ aniso : scaling ] @xmath384 and @xmath385 on the right - hand side of each of these equations , the first term reflects the change in the density of states at the band edge ( a single - particle effect ) while the second term reflects the lowest - order many - body effects .
these equations neglect all contributions beyond second - order in the exchange , and are therefore restricted in validity to the regime where @xmath386 and @xmath387 .
equations can be combined to obtain @xmath388 which can be integrated to yield @xmath389 one sees that exchange anisotropy is irrelevant for @xmath1 ( pseudogapped systems ) , marginal for @xmath38 ( conventional metals ) , and relevant for @xmath4 ( describing a power - law divergence of the host density of states at the fermi energy ) .
equation can be inserted into eq . to obtain @xmath390 after the completion of the work reported in this paper , we learned of a recent poor man s scaling formulation of the power - law kondo model with a more general anisotropic exchange coupling @xmath391 @xcite .
for the case @xmath392 considered here , the scaling equations of ref . reduce to eqs . and yield scaling trajectories fully equivalent in appearance to those plotted in figs .
[ fig : flows : r>0 ] and [ fig : flows : r=-0.1 ] of this paper . for @xmath1 , eqs .
have two stable fixed points , both isotropic as expected from eq . : the weak - coupling or lm fixed point @xmath393 , and the strong - coupling or kondo fixed point @xmath394 ( which lies beyond the regime of validity of the equations but is known to exist from nonperturbative studies ) .
there is also a critical fixed point @xmath395 that lies on the boundary between the basins of attraction of the stable fixed points .
the goal of this subsection is to map out the location of this boundary away from the point of su(2 ) spin symmetry . in light of eq .
, it is clear that any starting point on the boundary flows under eqs . to the isotropic critical point , which therefore governs the low - energy physics . for @xmath396 , one can factorize out the variation of @xmath397 arising from pure density - of - states effects [ i.e. , the effect of the @xmath398 term on the right - hand - side of eq . ] through the substitution @xmath399 which converts eq .
to @xmath400 \rho_0 j_z \ , e^{-rl}\ ] ] with the initial condition @xmath401 .
for any antiferromagnetic bare exchange @xmath402 , eq .
yields @xmath403 and @xmath404 for all @xmath405 .
if @xmath406 remains finite as @xmath407 , then @xmath397 vanishes as @xmath408 and the system must lie in the lm phase . on the other hand , we can associate the divergence of @xmath406 at some value @xmath409 with entry into the kondo regime around temperature @xmath410 .
the boundary between the two phases is determined by the divergence of @xmath411 only at @xmath412 . for a ferromagnetic bare exchange @xmath413 ,
any @xmath414 is sufficient to ensure that @xmath415 for all @xmath416 . in this case
, the system enters the kondo regime if @xmath406 changes sign and reaches @xmath417 for some finite @xmath418 . for the purposes of more detailed analysis , it proves convenient to parameterize the anisotropy of the bare exchange couplings in terms of the variable @xmath419,\ ] ] which can range from @xmath420 ( for @xmath421 ) to @xmath422 ( for @xmath423 ) to @xmath424 ( for @xmath425 ) .
then eq . can be rewritten @xmath426 solutions of this equation will be examined in the next two subsections . in cases where @xmath427 , @xmath428 defined in eq .
is positive and eq .
has the solution @xmath429 .\ ] ] for antiferromagnetic bare exchange ( @xmath402 ) , the kondo phase occupies the region of parameter space in which there is a solution @xmath430 of the equation @xmath431 , i.e. , @xmath432 thus , the kondo phase extends over @xmath433 where @xmath434 for @xmath435 ( weak anisotropy ) , @xmath436 which reduces for @xmath437 to the standard result @xcite @xmath438 . for @xmath439 ( strong anisotropy ) ,
@xmath440 which implies that the kondo phase occupies the region @xmath441 where @xmath442 for ferromagnetic bare exchange ( @xmath413 ) , the condition for entry into the kondo regime becomes @xmath443 , which is met for some finite @xmath418 provided that @xmath444 therefore , the kondo phase extends over the region @xmath445 where @xmath446 for @xmath447 ( weak anisotropy ) , @xmath448 while for @xmath439 ( strong anisotropy ) , @xmath449 so the kondo phase spans @xmath450 where @xmath451 for @xmath452 , @xmath428 defined in eq .
is negative and the solution of eq .
is @xmath453 .\ ] ] for antiferromagnetic bare exchange ( @xmath402 ) , the kondo phase spans the region in which @xmath454 i.e. , the region @xmath455 where @xmath456 for @xmath457 ( weak anisotropy ) , @xmath458 while for @xmath459 ( strong anisotropy ) , @xmath460 for @xmath461 , @xmath411 is a monotonically decreasing function of @xmath462 and the system always lies in the lm phase . in the special case @xmath464 of pure-@xmath463 bare exchange coupling , the scaling in eq .
can be replaced by @xmath465 which converts eq .
to @xmath466 with initial condition @xmath467 .
the equation has solution @xmath468.\ ] ] in the kondo phase , there must be an @xmath418 ( @xmath469 ) such that @xmath470 , a condition that is satisfied for @xmath441 where @xmath471 as one would expect , this result coincides with the limits @xmath472 of eqs . and .
the preceding results for the location of the phase boundary as a function of @xmath428 and the sign of @xmath473 can be re - expressed as the statement that for any value of @xmath473 , the kondo phase occupies the region @xmath474 , where @xmath475 is a monotonically decreasing function of @xmath473 that has the following limiting forms : [ j_pc : summary ] @xmath476 & & \text{for } 1 \gg -\rho_0 j_z \gg r\pi , \\[1ex ] \label{j_pc : limit2 } \rho_0 j_{\!{\perp},c } & \simeq r\pi/2 - 2 \rho_0 j_z/\pi & & \text{for } \rho_0 |j_z| \ll r , \\[1ex ] \label{j_pc : limit3 } \rho_0 j_{\!{\perp},c } & \simeq { \displaystyle}r - { { \textstyle\frac{1}{2}}}(\rho_0 j_z - r ) & & \text{for } |\rho_0 j_z\!-\!r|\ll r/3 , \\[1ex ] \label{j_pc : limit4 } \rho_0 j_{\!{\perp},c } & \simeq 2 \rho_0 j_z \exp(-\rho_0 j_z / r ) & & \text{for } r \ll \rho_0 j_z\ll 1 .\end{aligned}\ ] ] in the limit @xmath477 , these expressions reproduce the standard result @xcite @xmath478 .
the purpose of this section is to test these statements based on poor man s scaling against nonperturbative nrg calculations . scaling trajectories for the pseudogap kondo model with ( a ) @xmath254 , ( b ) @xmath479 , calculated via numerical iteration of eqs . .
arrows indicate the direction of flow under reduction of the half - bandwidth @xmath79 .
thick lines show trajectories that flow to the critical fixed point , thereby defining the boundary between the lm and kondo phases .
circles represent points on the phase boundary as determined using the nrg , with all @xmath473 and @xmath480 values for given @xmath11 rescaled by the same multiplicative factor , chosen so that the isotropic boundary point is located at @xmath481 . ]
scaling trajectories for the pseudogap kondo model , calculated via numerical iteration of eqs . with different starting parameters ,
are plotted in figs .
[ fig : flows : r>0](a ) and [ fig : flows : r>0](b ) for @xmath254 and @xmath479 , respectively .
solid lines show trajectories that flow to the fixed points of the model .
arrows on some of the trajectories show the direction of flow of the couplings under reduction of the half - bandwidth @xmath79 .
the phase boundary ( thick line ) separating the basins of attraction of the lm fixed point ( @xmath482 ) and the kondo fixed point ( @xmath483 ) was found by ( a ) reversing the flow of eqs . and ( b ) choosing starting parameters very close to the critical coupling @xmath438 and lying on either side of the the trajectory @xmath484 . for comparison , nrg data for the phase boundary ( circles )
are shown , with all values of @xmath473 and @xmath475 rescaled by the multiplicative factor that places the isotropic critical point at @xmath485 .
this @xmath11-dependent multiplicative factor is introduced to account both for a known reduction in hybridization arising from the nrg discretization @xcite and for the effect of higher - order terms omitted from the poor man s scaling equations , which shift the isotropic critical point from @xmath486 to @xmath487 @xcite .
figure [ fig : flows : r>0 ] shows that poor man s scaling does an excellent job of reproducing the shape of the phase boundary over the entire region of couplings @xmath488 , @xmath489 .
a more rigorous test of the poor man s scaling is provided by figs.[fig : limit1][fig : limit4 ] , which compare @xmath490 vs @xmath491 calculated for one of the limiting cases in eqs .
( solid lines ) with their nrg counterparts ( symbols ) .
the nrg results are again scaled so that the isotropic critical point is at @xmath485 ) .
pseudogap kondo model phase boundary @xmath490 vs @xmath491 for @xmath254 , comparing nrg results ( symbols ) with the poor man s scaling prediction for @xmath492 as given in eq .
( line ) . ]
figure [ fig : limit1 ] plots the critical coupling @xmath490 for @xmath254 over a range of ferromagnetic exchange couplings @xmath493 . although the perturbative scaling analysis is not strictly valid for @xmath494 , eq . captures surprisingly well the variation of @xmath495 at least as far as @xmath496
for @xmath479 , the restriction @xmath497 rules out the applicability of eq .
anywhere within the range of validity of the scaling equations , so no results are shown for this case .
pseudogap kondo model phase boundary plotted as @xmath498 vs @xmath499 for @xmath254 and @xmath479 , comparing nrg results ( symbols ) with the poor man s scaling prediction for @xmath500 as given in eq .
( lines ) . ] in fig .
[ fig : limit2 ] , the critical coupling is plotted as @xmath498 versus @xmath499 near @xmath501 .
the nrg results are well reproduced by the poor man s scaling prediction eq . over its entire range of validity @xmath502 .
figure [ fig : limit3 ] focuses on the vicinity of the isotropic critical point at @xmath485 , plotting the critical coupling as @xmath503 vs @xmath504 . the scaling prediction in eq .
closely reproduces the nrg results over the range @xmath505 .
pseudogap kondo model phase boundary plotted as @xmath503 vs @xmath504 for @xmath254 and @xmath479 , comparing nrg results ( symbols ) with the poor man s scaling prediction for @xmath506 as given in eq .
( lines ) . ]
pseudogap kondo model phase boundary plotted as @xmath507 vs @xmath491 for @xmath254 and @xmath479 , comparing nrg results ( symbols ) with the poor man s scaling prediction obtained via numerical solution of eq .
( solid lines ) and the asymptotic form for @xmath508 from eq .
( dashed line ) . ]
lastly , fig .
[ fig : limit4 ] plots the critical coupling as @xmath509 versus @xmath510 for @xmath511 .
we find that the nrg results closely follow the asymptotic form for @xmath512 [ eq .
, dotted lines ] over the range @xmath513 for @xmath254 and over @xmath514 for @xmath479 .
there are minor deviations from the asymptotic form as @xmath491 nears @xmath68 due to perturbative effects beyond second order .
we have also plotted the poor man s scaling prediction obtained via numerical solution of eq .
( solid lines ) , which can be seen to describe correctly the deviation of @xmath515 near @xmath464 from its @xmath516 asymptote .
the overall conclusion from figs .
[ fig : flows : r>0][fig : limit4 ] is that the poor man s scaling approach provides an excellent account of the location of the boundary between the kondo and local - moment phases of the spin - anisotropic pseudogap kondo model under conditions of strict particle - hole symmetry . scaling trajectories for the power - law kondo model with @xmath517 , representing a divergence of the density of states at the fermi energy .
arrows indicate the direction of flow under reduction of the half - bandwidth @xmath79 .
thick lines show trajectories that flow to the critical point , defining the phase boundary between the ferromagnetic and kondo phases . ]
the poor man s scaling analysis of the anisotropic kondo model can also be applied to cases @xmath4 where the the density of states in eq .
[ rho : def ] diverges in power - law fashion at the fermi energy .
examination of eqs .
show that the poor man s scaling trajectories for @xmath4 can be obtained from those for band exponent @xmath518 through the simple replacements @xmath519 , @xmath520 .
this mapping implies that the scaling trajectories for @xmath4 should be reflections of those for @xmath1 about the axis @xmath464 with reversal of the direction of flow arrows .
this is illustrated fig .
[ fig : flows : r=-0.1 ] , which plots the scaling trajectories for a representative case @xmath517 over the range of exchange couplings @xmath521 and @xmath522 .
arrows indicate the direction of flow of couplings with decreasing effective half - bandwidth .
the model has three stable fixed points : a ferromagnetic fixed point at @xmath523 where the impurity is locked into a many - body spin triplet with the conduction band , the symmetric strong - coupling fixed point at @xmath524 , and an intermediate coupling fixed point at @xmath525 .
the phase boundary ( thick lines ) separating the ferromagnetic and strong - coupling phase is given by the condition @xmath526 , which is consistent with nrg studies of the model .
in this work , we have extended the poor man s scaling method to analyze phase boundaries in variants of the anderson and kondo impurity models that feature a power - law vanishing or divergence of the host density of states at the fermi energy .
the predicted locations of the phase boundaries are generally in excellent qualitative and good quantitative agreement with those obtained using the numerical renormalization group ( nrg ) , which remains the most reliable technique for treating power - law quantum impurity problems .
the scaling approach has the advantages that it is much more intuitive than the nrg and it can clarify algebraically the functional dependence of the critical impurity - host coupling on other model parameters .
thus , poor man s scaling retains considerable value even for quantum impurity problems where two or more competing rg flows give rise to different possible infrared - stable fixed points .
this work has been supported in part by nsf - dmr grants no .
0710540 ( m.c .
, k.i . ) , 1107814 ( t.c . , k.i . ) , and 1508122 ( t.c . , k.i . ) .
a.m. was supported by the university of florida reu site in materials physics under nsf grant no.dmr-1156737 with additional support from the us department of defense .
we thank e. kogan for bringing ref . to our attention and for useful discussions .
p. w. anderson , j. phys .
c * 3 * , 2436 ( 1970 ) .
k. g. wilson , rev .
* 47 * , 773 ( 1975 ) .
j. h. jefferson , j. phys .
c * 10 * , 3589 ( 1977 ) .
f. d. m. haldane , phys .
* 40 * , 416 ( 1978 ) .
nozires and a. blandin , j. phys .
( paris ) * 41 * , 193 ( 1980 ) .
m. sigrist and k. ueda , rev .
phys . * 63 * , 239 ( 1991 ) .
t. timusk and b. statt , rep .
phys . * 62 * 61 ( 1999 ) . for reviews ,
see g. nimtz , b. schlict , `` narrow - gap lead salts , '' in _ narrow - gap semiconductors , _ springer tracts in modern physics , vol .
98 , ed .
g. hhler ( springer - verlag , berlin , 1983 ) , p. 1 ; r. dornhaus and g. nimtz , `` the properties and applications of the hg@xmath527cd@xmath528te alloy system , '' _ ibid .
_ , p. 119 .
g. w. semenoff , phys .
lett . * 53 * , 2449 ( 1984 ) .
a. h. castro neto , f. guinea , n. m. r. peres , k. s. novoselov , and a. k. geim , rev .
phys . * 81 * , 109 ( 2009 ) .
d. withoff and e. fradkin , phys .
lett . * 64 * , 1835 ( 1990 ) .
k. ingersent , in proceedings of physical phenomena at high magnetic fields - ii " , edited by z. fisk et al .
( world scientific publishing company , singapore , 1996 ) , pp .
179-184 . c. r. cassanello and e. fradkin , phys .
b * 53 * , 15079 ( 1996 ) .
m. vojta , phys .
* 87 * , 097202 ( 2001 ) .
a. polkovnikov , phys .
b * 65 * , 064503 ( 2002 ) .
k. chen and c. jayaprakash , j. phys . : condens .
matter * 7 * , l491 ( 1995 ) .
k. ingersent , phys .
b * 54 * , 11936 ( 1996 ) .
r. bulla , th .
pruschke , and a. c. hewson , j. phys . : condens .
matter * 9 * , 10463 ( 1997 ) . c. gonzalez - buxton and k. ingersent , phys .
b * 57 * , 14254 ( 1998 ) .
m. vojta and r. bulla , phys . rev .
b * 65 * , 014511 ( 2001 ) .
k. ingersent and q. si , phys .
lett . * 89 * , 076403 ( 2002 ) .
j. h. pixley , s. kirchner , k. ingersent , and q. si , phys .
lett . * 109 * , 086403 ( 2012 ) .
t. chowdhury and k. ingersent phys .
b * 91 * , 035118 ( 2015 ) .
m. kiran and m. vojta , phys .
b * 69 * , 174421 ( 2004 ) .
m. vojta and l. fritz , phys .
b * 70 * , 094502 ( 2004 ) .
l. fritz and m. vojta , phys .
b * 70 * , 214427 ( 2004 ) .
d. e. logan and m. t. glossop , j. phys . : condens
. matter * 12 * , 985 ( 2000 ) .
r. bulla , m. t. glossop , d. e. logan and t. pruschke , j. phys . : condens .
matter * 12 * , 23 ( 2000 ) .
m. t. glossop and d. e. logan , j. phys . : condens .
matter * 15 * , 7519 ( 2003 ) .
p. w. anderson , phys .
rev . * 124 * , 41 ( 1961 ) .
the hybridization _
function _ is denoted by the symbol @xmath529 to distinguish it from the hybridization _ prefactor _
@xmath44 appearing on the right - hand side of eq .
m. cheng and k. ingersent , phys .
b * 87 * , 075145 ( 2013 ) .
j. r. schrieffer and p. a. wolff , phys .
* 149 * , 491 ( 1966 ) .
reference states incorrectly that meeting the conditions for case ( 5 ) necessarily results in entry into a strong - coupling phase .
r. bulla , t. a. costi , and th .
pruschke , rev .
phys . * 80 * , 395 ( 2008 ) .
the impurity contribution to a physical property is defined in the standard manner ( see , e.g. , ref . ) as the difference between the total value of the property for the system including the impurity and the total value of the same property for the host alone .
|
we use the poor man s scaling approach to study the phase boundaries of a pair of quantum impurity models featuring a power - law density of states @xmath0 that gives rise to quantum phase transitions between local - moment and kondo - screened phases . for the anderson model with a pseudogap ( i.e. , @xmath1 )
, we find the phase boundary for ( a ) @xmath2 , a range over which the model exhibits interacting quantum critical points both at and away from particle - hole symmetry , and ( b ) @xmath3 , where the phases are separated by first - order quantum phase transitions .
for the particle - hole - symmetric kondo model with easy - axis or easy - plane anisotropy of the spin exchange , the phase boundary and scaling trajectories are obtained for both @xmath1 and @xmath4 ( the later case describing a density of states that diverges at the fermi energy ) .
comparison with nonperturbative results from the numerical renormalization group shows that poor man s scaling correctly describes the shape of phase boundaries expressed as functional relations between model parameters .
|
oort ( 1950 ) postulated the existence of a cloud of comets orbiting the sun with typical semi - major axes , @xmath2 , of @xmath3au .
this cloud is required to explain the existence of long period comets and the fact that a considerable fraction of these comets have orbital energies concentrated in a narrow range , corresponding to @xmath4 au@xmath5 . however , to date no direct observation of objects in the oort cloud exist .
dynamical simulations suggest that the oort cloud was formed by the ejection of icy planetesimals from the jupiter - neptune region by planetary perturbations ( e.g. , duncan et al 1987 ; dones et al .
2004 ) . on orbital
time scales , galactic tides and passage of nearby stars raised the perihelia of these comets above the region of the giant planets influence .
comets with semi - major axes above @xmath6au are loosely bound to the sun , and tides by the galactic potential ( e.g. , heisler & tremaine 1986 ) as well as impulse by passing nearby stars , may send these comets to the inner solar system where they reveal themselves as long period comets .
in contrast , oort cloud objects with @xmath7au ( `` inner oort cloud '' ; e.g. , hills 1981 , bailey 1983 ) have more stable orbits .
we note that the inner `` boundary '' of the oort cloud , @xmath8 , is set by perturbations due to the giant planets and the primordial stellar neighborhood of the sun .
this @xmath8 is estimated to be in the range of 1,000au to 3,000au ( duncan et al .
1987 ; dones et al . 2004
; fernndez 1997 ) .
below @xmath8 the number of comets steeply falls due to planetary influence .
dynamical simulations suggest that the spatial density of oort cloud objects in the 3,00050,000au range falls as @xmath9 , with @xmath10 ( e.g. , duncan et al . 1987 ) .
the current flux of long period comets calibrated by dynamical simulations suggest that the outer oort cloud ( i.e. , @xmath11au ) contains @xmath12 comets with nuclei absolute planetary magnitude brighter than 16mag ( e.g. , heisler 1990 ; weissman 1996 ) .
this magnitude roughly corresponds to objects with radius of 2 km ( assuming @xmath13 geometric albedo ) .
the dynamically more stable inner oort cloud contains about 2100 more comets than its outer counterpart ( e.g. , hills 1981 ; duncan et al .
1987 ; fernndez & brunini 2000 ; dones et al . 2004 ; brasser et al .
2008 ) . detecting oort cloud objects , and estimating their size distribution ,
will improve our understanding of the mass of the solar - system planetary accretion disk , and the dynamical processes in the young solar system .
moreover , it may bear some clues to the stellar density at the region in which the sun was born ( e.g. , fernndez 1997 ; fernndez & brunini 2000 ; brasser et al .
however , these objects are beyond the direct reach of even the currently largest planned telescopes , and some indirect methods ( e.g. , astrometric microlensing ; gaudi & bloom 2005 ) . bailey ( 1976 )
suggested that small kuiper belt objects ( kbos ) as well as oort cloud objects could be detected by occultations of background stars .
indeed several surveys are looking for kbo occultations ( e.g. , roques et al .
2006 ; lehner et al . 2009 ; chang et al . 2006 , 2007 ; bianco et al . 2009 ; schlichting et al .
however , to date there is only one reported occultation of a kuiper belt object ( schlichting et al .
2009 ) . in this letter
we discuss the potential of space missions designed to search for transiting exo - planets ( e.g. , _ kepler _ , _ corot _ ) to detect occultations of background stars by small oort cloud and kuiper belt objects .
space telescope designed to look for exo - planet transits usually have exposure times , @xmath14 , longer by orders of magnitude than the typical duration of an occultation of a background star by a small solar system object , @xmath15 .
the relative flux decrement of such an occultation @xmath16 , integrated over the exposure time , is diluted by a factor @xmath17 .
however , these satellites have superb photometric accuracy .
their noise level per exposure is @xmath18 , where @xmath19 is a normalization parameter ( @xmath20 ) for 1s integration of a 12th magnitude star .
this sensitivity parameter depends on the star magnitude @xmath21 as @xmath22 . ] .
therefore , we may be able to detect these minute occultations . here ,
we adopt a detection criterion for an occultation of @xmath23 , where @xmath24 is the signal - to - noise ratio . throughout the paper
we adopt @xmath25 . for an occultation by objects at a given distance from earth , @xmath26 , the maximal value of @xmath16 ( obtained for an optimal impact parameter . ]
@xmath27 ) increases monotonically with the object size .
we define @xmath28 as the radius of the smallest object that can be detected , and correspondingly @xmath29 .
furthermore , we can define the sum of impact parameters for which objects with a radius @xmath30 are detectable , @xmath31db$ ] , where @xmath32 is heaviside step function . according to these definitions @xmath33 .
the size distribution of kbos and oort cloud objects per unit solid angle is parametrized by a power - law , @xmath34 , or by a broken power - law . here
, @xmath35 is the total number of objects with @xmath36 km per unit area , and @xmath37 is the size distribution power - law index . for oort cloud objects @xmath38 , where @xmath39 is the surface area of the celestial sphere .
for kbos larger than @xmath40 km @xmath41 ( bernstein et al .
2004 ; fuentes & holman 2008 ; fraser et al .
2008 ) , and @xmath42 for smaller kbos ( e.g , farinella & davis 1996 ; pan & sari 2005 ; schlichting et al . 2009 ) . for oort cloud objects @xmath37 is unconstrained by observations ( see however goldreich et al .
2004 ) .
the number of events in a survey that observes @xmath43 stars for a duration @xmath44 is then @xmath45 , where @xmath46 is the angular velocity of the occulting objects on the sky as seen from earth . here
all objects are at the same distance @xmath26 , all stars have the same magnitude and angular radii and @xmath46 is constant . when this is not the case , it is straightforward to integrate over the distribution of these properties .
the functional forms of @xmath47 , @xmath16 , @xmath48 and thus @xmath49 depend on the ratios between the three angular scales in the problem .
these are the object angular radius , @xmath50 ; the stellar angular radius , where @xmath51 and @xmath52 are the effective temperature and bolometric magnitude of the star , respectively . ] @xmath53 ; and the angular fresnel scaleau and @xmath54 , f=10.6 km and @xmath55 arcsec . ] @xmath56 , with @xmath57 , and @xmath58 the wavelength of observation .
the duration of the eclipse is @xmath59 , where @xmath60 .
table [ tab : rmin ] gives the functional form of @xmath48 in the different `` asymptotic '' cases ( denoted by a
e ) , where the parameters are normalized to those of a typical oort cloud object and the _ kepler _ sensitivity . the cases where @xmath61 ( case a ) and @xmath62 ( cases b c ) represent geometric occultations . in these cases
we approximate @xmath63 for any @xmath64 . here , the @xmath37-dependent factor ( i.e. , @xmath65/13 $ ] ) is a result of the assumed circular shape of the object .
we obtained this approximate correction factor by numerically integrating geometrical occultations by circular shaped objects . in the diffractive regime ( @xmath66 ;
cases d e ) we numerically calculated diffractive light curves using eq .
b1 in roques et al .
we find that @xmath67 can be well approximated by @xmath68 .
\label{eq : deltadiffractive}\ ] ] the scaling of eq .
[ eq : deltadiffractive ] can be understood as follows .
the occultation `` shadow '' pattern of an object can be described as a set of bright and `` dark '' concentric rings with the central ring ( around the `` poisson peak '' ) being dark with an angular radius of @xmath69 .
the area in all the rings is constant ( @xmath70 ) so the angular width of the rings is inversely proportional to its angular distance from the center ( e.g. , fig . 1 in nihei et al 2007 ) .
the depth of the decrement in the flux in the dark rings is @xmath71 so the value of @xmath16 for the central ring occultation is @xmath72 .
this decrement remains similar in all the dark rings as long as the ring width is smaller than @xmath73 .
therefore @xmath74 . for each of the five cases we use the approximations described above and give formula ( table [ tab : nev ] ) for the expected number of events as a function of the parameters @xmath24 , @xmath19 , @xmath14 , @xmath75 , @xmath26 , @xmath58 and @xmath76 .
here @xmath77 is the relative velocity between the occulting object and earth , projected on the plane perpendicular to the earth - object direction .
since the object size distribution is steep , the integration over @xmath30 is dominated by the smallest objects , and is carried to infinity ( rather than the radius at which the condition defining the `` case '' is violated ) .
llll@l@l@l@l@l a & @xmath78 & @xmath79 & @xmath80 & @xmath81 & @xmath82 & & & + b & @xmath83 & @xmath84 & @xmath85 & @xmath86 & @xmath87 & @xmath88 & @xmath89 & + c & @xmath90 & @xmath91 & @xmath85 & @xmath86 & @xmath87 & @xmath88 & @xmath89 & + d & @xmath92 & @xmath93 & @xmath94 & @xmath95 & @xmath96 & @xmath97 & & @xmath98 + e & @xmath99 & @xmath93 & @xmath94 & @xmath95 & @xmath96 & @xmath97 & & @xmath98 [ tab : rmin ] ll@l@l@l@l@l@l a & @xmath100 & @xmath101 & @xmath102 & @xmath103 & @xmath104 & & + b & @xmath105 & @xmath106 & @xmath107 & @xmath108 & @xmath109 & @xmath110 & + d & @xmath111 & @xmath112 & @xmath113 & @xmath114 & @xmath115 & & @xmath116 + e & @xmath117 & @xmath106 & @xmath107 & @xmath108 & @xmath118 & @xmath119 & @xmath120 [ tab : nev ]
to estimate the observed rate of events we need to take into account the radial distance distribution ( discussed in
[ sec : intro ] ) , the relative velocity ( @xmath77 ) probability distribution , the ecliptic latitude distribution , and the stellar angular size and magnitude co - distributions . in figure
[ fig : sat_starsangraddist ] we present the estimated angular radii distribution of target stars observed by _
kepler _ and _ corot_. to estimate this we queried the guide star catalog ( gsc ; version 2.2 ; lasker et al .
2008 ) for all stars within 2deg from the center of the _ kepler _ ( _ corot _ ) field , that have @xmath30-band magnitudes between 9 and 14 ( 12 and 16 ) , and @xmath121 colors between 0.59 and 3.4mag .
this color range roughly corresponds to the spectral range of _ kepler _ targets .
we converted the color index of each star to an effective temperature and a bolometric correction by fitting the color indices with those obtained from synthetic photometry of black - body spectra with different temperatures ( e.g. , ofek 2008 ) .
the synthetic photometry ( poznanski et al . 2002 ) was preformed using the transmission curves of the filters used in the gsc ( moro & munari 2000 ) .
[ fig : sat_starsangraddist],width=321 ] given the unit vector in the direction of a target @xmath122 , the speed of the observer relative to the occulting object @xmath123 , and the heliocentric velocity unit vector of the observer @xmath124 , the approximate relative velocity , of an observer in an earth trailing orbit , projected on the plane perpendicular to @xmath122 is @xmath125 where @xmath58 and @xmath126 are the ecliptic longitude and latitude of the observed star , @xmath127 is the ecliptic longitude ( as measured from the sun ) of the observer , and @xmath128kms@xmath5 is the heliocentric speed of the observer . finally , assuming circular orbit ( i.e. , the probability distribution of @xmath127 is uniform ) the probability distribution of @xmath77 is @xmath129 .
the oort cloud is predicted to have a roughly spherical shape .
however , the kuiper belt and the asteroids main belt are concentrated toward the ecliptic plane . using the inclination distribution of kbos estimated by elliot et al .
( 2005 ) the expectation probability to find a kbo at an ecliptic latitude of @xmath130 is about @xmath131 times smaller than that at the ecliptic .
this relative probability was estimated by convolving the inclination probability distribution of kbos with the ecliptic latitude distribution for a given inclination ( i.e. , elliot et al . 2005
40 and 34 , respectively ) .
we note , however , that this number is highly uncertain . to estimate the rate of occultation events we integrate the appropriate formulas in table [ tab : nev ] over @xmath77 , @xmath26 , @xmath75 and @xmath19 co - distributions .
for each combination of parameters we integrate over the most appropriate occultation channel ( i.e. , cases a
e ) . in table
[ tab : rates ] we present the estimated rates predicted for _ kepler _ and _ corot _ for the oort cloud ( upper block ) , kuiper belt ( middle block ) , and main belt ( lower block ) object occultations .
the predicted rate for oort cloud objects in table [ tab : rates ] in case of @xmath132 and @xmath133au is lower by about an order of magnitude compared to the pre - factors in table [ tab : nev ] .
this is mostly because table [ tab : nev ] is normalized for all objects being at @xmath134 au while in table [ tab : rates ] the same number of objects is spread between @xmath8 and @xmath135 , with a spatial density distribution proportional to @xmath136 .
lllcc @xmath137 & @xmath138 & @xmath139 & 30 & 0.1 + @xmath137 & @xmath138 & @xmath140 & 8 & @xmath141 + @xmath137 & @xmath138 & @xmath142 & 0.7 & @xmath143 + @xmath137 & @xmath144 & @xmath140 & 60 & 0.03 + @xmath137 & @xmath145 & @xmath140 & 3 & @xmath146 + @xmath147 & @xmath148 & @xmath139 & @xmath149 & @xmath150 + @xmath151 & @xmath148 & @xmath140 & @xmath152 & @xmath141 + @xmath153 & @xmath148 & @xmath154 & @xmath155 & @xmath156 + @xmath157 & @xmath158 & @xmath159 & @xmath160 & 0.2 [ tab : rates ] these rate estimates suggest that if our current best guess regarding the properties of the oort cloud are correct , then it may be detectable by _
kepler_. regardless , _
kepler _ will be able to put the best constraint ( so far ) on the content of the oort cloud .
we note that the rate estimates in table [ tab : rates ] are an approximation to the actual rate .
this is mainly because our formulas represent asymptotic approximations , while in some cases two out of the three angular scales , @xmath161 , @xmath69 , @xmath75 , maybe be similar .
contrary to transits and grazing transits by extra - solar planets , solar system occultations will appear in a single photometric measurement and they happen only once .
therefore , verification of each event by itself will require a thorough understanding of the noise properties of the photometer .
_ kepler _ will have about @xmath162 15-min independent photometric measurements .
assuming purely gaussian noise , the 8-@xmath163 detection threshold we adopted corresponds to a probability of @xmath164 .
however , even small deviations from a gaussian noise may impair our confidence in the astrophysical nature of the events . here
we present two tests that can be used to check if the occultations signature are due to real events or due to some sort of uncharacterized noise .
in addition , these tests can be used to statistically measure @xmath37 and to distinguish between oort cloud objects and kbos .
the occultation rate depends on the relative velocity @xmath77 , with a somewhat different dependency in each regime . therefore , if a large number of occultations are detected , a simple sanity test is to look for the dependency of the events rate on @xmath77 .
we note that since _ kepler _ is observing at high ecliptic latitude the @xmath77 modulation will be small ( eq . [ eq : vrel ] ) compared to the maximal modulation which will be seen by an instrument observing the ecliptic .
moreover , in all the cases ( a e ) , the function @xmath165 depends on @xmath37 .
therefore , the amplitude of the modulation is related to @xmath37 and the occultation type ( i.e. , a
knowing the occultation regime ( see below ) , we can measure @xmath37 ( i.e. , removing the degeneracy between @xmath37 and @xmath165 ) .
furthermore , as shown in table [ tab : nev ] the occultations rate depends on @xmath19 , which depends on the stellar magnitude and on @xmath75 , which in turn is a function of the star magnitude and effective temperature .
we expect bright and angularly small stars to have higher chance of producing detectable occultations .
moreover , this dependency is different for the various occultation cases .
this can be used to further test the nature of the occultations , by comparing the distribution of magnitudes and effective temperatures of the stars showing occultations to those of the entire observed stellar population .
most importantly , for a _ kepler_-like observatory , kbo occultations will be mostly in the diffractive regime , while oort cloud object occultations will be in the geometric regime ( fig . [
fig : sat_starsangraddist ] ) .
therefore , the magnitude and @xmath75 of occulted stars by kbos and oort cloud objects will follow different co - distributions .
therefore , we can statistically distinguish between oort cloud and kbo occultation events .
however , quantifying these effects and estimating the number of occultations needed to carry out the suggested analysis requires a more realistic estimate of the events rate ( i.e. , using full numerical integration of light curves ) and a full exploration of the parameters space . to summarize ,
we show that _ kepler _ and similar space missions can detect deca - kilometer objects in the oort cloud .
we present statistical methods to verify that the occultations are real rather than due to uncharacterized noise .
in addition we suggest that it will be possible to statistically measure their size distribution ( @xmath37 ) , and the dominant occultation regime , and therefore differentiate between oort cloud objects and kbos .
considering _ kepler _ capabilities and reasonable oort cloud parameters , we find that _ kepler _ may detect 0 to @xmath166 stellar occultations by oort cloud objects .
we find that _ kepler _ is unlikely to detect kuiper belt objects ( mainly because of its high ecliptic latitude pointing ) .
moreover , _
corot _ is unlikely to detect kuiper belt or oort cloud objects .
however , the exact properties of the oort cloud and the kuiper belt are not well known .
therefore , such searches are warranted .
the analytical treatment we present in this paper is useful in order to maximize the efficiency of existing ( and future , e.g. , _
plato _ ) experiments to detect trans neptunian objects .
however , this treatment is accurate only in the asymptotic cases , and numerical calculations are needed in order to give more precise predictions .
we thank reem sari , hilke schlichting , orly gnat and michael busch for many valuable discussions .
support for program number hst - ar-11766.01-a was provided by nasa .
eoo is supported by an einstein fellowship .
en is partially supported by an irg grant .
|
the size distribution and total mass of objects in the oort cloud have important implications to the theory of planets formation , including the properties of , and the processes taking place in the early solar system .
we discuss the potential of space missions like _ kepler _ and _ corot _ , designed to discover transiting exo - planets , to detect oort cloud , kuiper belt and main belt objects by occultations of background stars . relying
on published dynamical estimates of the content of the oort cloud , we find that _ kepler _ s main program is expected to detect between @xmath0 and @xmath1 occultation events by deca - kilometer - sized oort cloud objects .
the occultations rate depends on the mass of the oort cloud , the distance to its `` inner edge '' , and the size distribution of its objects .
in contrast , _ kepler _ is unlikely to find occultations by kuiper belt or main belt asteroids , mainly due to the fact that it is observing a high ecliptic latitude field .
occultations by solar system objects will appear as a photometric deviation in a single measurement , implying that the information regarding the time scale and light - curve shape of each event is lost .
we present statistical methods that have the potential to verify the authenticity of occultation events by solar system objects , to estimate the distance to the occulting population , and to constrain their size distribution .
our results are useful for planning of future space - based exo - planet searches in a way that will maximize the probability of detecting solar system objects , without hampering the main science goals .
|
the electromagnetic form factors of the nucleons , @xmath5 and @xmath6 , depend on @xmath7 . in the space - like region
they are relatively real .
they have been measured using @xmath8 elastic scattering where @xmath9 is the is the absolute value of the four - momentum transfer from the incoming lepton to the nucleon .
the techniques used are : 1 .
rosenbluth separation : measurements made at at least two angles and fixed @xmath7 .
+ @xmath10 2 .
polarized beam and polarized target 1 .
target polarization in scattering plane @xmath11 @xmath12 vector : + @xmath13 2 .
target polarization @xmath14 to @xmath12 vector : + @xmath15 3 .
polarized beam and polarization of recoil nucleon : 1 .
recoil polarization in scattering plane @xmath11 @xmath12 vector : + @xmath16 2 .
recoil polarization @xmath14 to @xmath12 vector : + @xmath17 the first method gives only the absolute value of the form factors . combining measurements
2a and 2b gives the relative sign thru : @xmath18 .
combining measurements 3a and 3b gives the relative sign thru : @xmath19 .
all three methods have been used in the space like region for both the proton and the neutron .
@xcite in the time - like region @xmath20 is positive and the form factors are complex , so it is necessary to measure @xmath21 , @xmath22 and the phase difference .
these can be determined using the process @xmath2 .
the momentum transfer @xmath23 where @xmath24 is the square of the center of mass energy .
the @xmath25 are in an l=0 ( @xmath26 ) of l=1 ( @xmath27 ) state with @xmath28 and @xmath29 . at threshold @xmath30 and thus @xmath31 and @xmath32=0 $ ] .
the vmd model of dubnicka , dubnickova and strizenec @xcite gives predictions of the magnitude and phases of the form factor and predict significant non zero phase difference in the pep - n energy region .
the following experiments are possible .
1 . unpolarized beam particles
rosenbluth separation : measurements made at at least two angles and fixed @xmath7 .
+ @xmath33 $ ] 2 .
recoil polarization @xmath11 to scattering plane : + @xmath34/\sqrt{\tau } \over |g_e|^2\sin^2(\theta)/\tau + |g_m|^2(1+\cos^2\theta)}$ ] 2 .
one longitudinally polarized beam particle 1 .
recoil polarization @xmath14 to baryon : + @xmath35 2 .
recoil polarization in scattering plane @xmath11 to baryon : + @xmath36/\sqrt{\tau } \over |g_e|^2\sin^2(\theta)/\tau + |g_m|^2(1+\cos^2\theta)}$ ] method 1b allows us to measure the phase difference once the absolute values have been determined using the rosenbluth separation ( 1a ) . the measurement can be carried out at all scattering angles simultaneously .
the polarization of the recoil nucleon can be measured in a polarimeter similar to the one used recently at jlab @xcite .
the method is to precisely measure the incoming trajectory , re - scatter the recoil nucleon on a carbon target , and then measure the outgoing angle to get the angular dependence of the re - scatter .
the number of counts @xmath37 is given by : @xmath38 where @xmath39 is the polarization to be measured , @xmath40 is the effective analyzing power , and @xmath41 is the second scattering angle .
this may require a separate detector unless we can embed a precise position measuring system within the multi - layered shower counter .
the @xmath42 form factors could be measured by using the self analyzing power of the decay .
corrections must be made for the precession of the baryon spin by the field of the vertex magnet .
we estimate that there will be about 200 @xmath25/day .
i assume that we have an analyzing power similar to that at jlab which is 0.5 at 230 mev and 0.1 at a few gev .
the probability of scattering @xmath43 in the polarimeter is 0.01 to 0.1 giving the figure of merit @xmath44 .
the error on the nucleon polarization is given by @xmath45 where @xmath46 is the number of nucleons going into the polarimeter .
with these parameters , a 100 day run yields @xmath47 .
@xmath48 and @xmath49 and their relative phase can be measured at pep - n using unpolarized beams and measuring the recoil polarization of the nucleon .
99 gerassimos g. petratos , nucl .
a666&6667 ( 2000 ) 61 . and references therein .
a. z. dubnickova , s. dubnicka and strizenec , ll nuovo cimento a106 , ( 1993 ) 1253 .
a. z. dubnickova , s. dubnicka and m. rekalo ll nuovo cimento a109 ( 1996 ) 241 . b. milbrath _ et .
al . , _ phys .
lett . 80 ( 1998 ) 452 .
m. k. jones _ et .
al . , _ phys .
84 , ( 2000 ) 1398 , physics archives nucl - ex/9910005 ; http://hallaweb.jlab.org/equipment/detectors/fpp_overview.html ,
|
the nucleon electromagnetic form factors @xmath0 and @xmath1 are complex quantities in the time - like region .
the absolute values can be determined by measuring the angular distribution of the nucleons in @xmath2 .
the complex phase can only be determined by measuring one or more polarizations of the initial or final state . for pep - n
, we can use unpolarized @xmath3 and @xmath4 and measure the polarization of one of the outgoing nucleons .
|
spintronics and graphene are exciting , vigorous , and rather new areas of research with a rapid pace of new discoveries
. books and a multitude of review articles are available for those that want to learn about one of these fields [ ] .
the intersection of both provides fertile soil for an abundance of fascinating physics and its fruits have been partially reaped by previous reviews [ ] . here
, we put the cherry on the cake and review specifically spintronics in graphene quantum dots . while classical electronics relies on the charge for information processing , spin - based electronics or _
spintronics _ is the paradigm of an advanced technology where the spin degree of freedom complements or even replaces charge as the carrier of information . the expected benefits of spintronics encompass non - volatile data storage , faster and more energy efficient data processing , increased data density , and many more . in order to fully exploit the potential , efficient generation , transport , transfer , manipulation , and detection of spin polarization
is required .
all these requirements are closely connected to material properties .
therefore , progress in spintronics seems intrinsically connected to the quest for new materials with appropriate characteristics [ ] .
c nuclear spin via the hyperfine interaction . ]
graphene is a novel material that has undeniably stirred up the solid state community since its isolation more than a decade ago [ ] .
it is a monatomically thin , quasi two - dimensional layer of carbon atoms arranged in a honeycomb lattice with two sublattices , a and b. due to its atomic structure , it has the lowest surface mass density of all solids [ ] . at the same time , its in - plane mechanical strength is second to no other material [ ] .
the hexagonal structure is the consequence of sp@xmath2-hybridized carbon atoms .
the remaining p - orbitals form the @xmath3-bands responsible for transport .
electrons move quasi - relativistically with the fermi velocity @xmath4 , where @xmath5 is the vacuum speed of light , and with high mobilities [ ] .
spin - orbit couplings are small and nuclear spins are sparse , making it a potentially good host material for spin [ ] .
in addition to the real electron spin , the occupation of sublattices a and b can be described by the sublattice pseudospin . moreover , in momentum space , the occupation of the two inequivalent dirac points where valence and conduction bands touch may be denoted by the valley isospin [ ] .
electron spin currents can be injected optically in graphene with rashba - type spin - orbit interaction [ ] . this allows to leave away ferromagnetic contacts , thus eliminating a possible source of spin scattering .
magnetic behavior has been observed at the edges of graphene [ ] .
these and further properties make graphene a very promising material for spintronics .
quantum dots ( qds ) , in graphene as well as in other materials , are quasi zero - dimensional regions to which charge carriers can be confined [ ] . with electrostatic gate electrodes that can be used to adjust the confinement potential and with electric contacts for transport , qds allow in principle for full control over the individual electron .
in particular , qds provide a controllable playground to investigate the behavior of spins as well as sources of decoherence and methods to increase coherence times [ ] . in general
, there are nuclear spins as well as electron spins in a solid . unless explicitly stated otherwise , we always refer to electron spin .
in this review , we focus on what we believe to be the most immediate questions related to spintronics with localized electron spins in graphene .
however , we point out that there are a several related systems that make use of carbon nanotubes or transition metal dichalcogenides ( tmdcs ) and where the electron spin degree of freedom might be substituted or complemented by a pseudospin or valley isospin [ ] .
the review is organized as follows . in the next section ,
we give an overview over spintronics in quantum dots and motivate the use of graphene . in sec .
[ t1sec ] , we discuss various graphene quantum dots and their associated spin relaxation times . closely related to the concept of qds are localized states due to defects in graphene nanoflakes , which we treat in sec .
[ gnf ] . in sec .
[ nspins ] , we address the influence of nuclear spins on spintronics in graphene .
recent experimental achievements and the current status are the topic of sec .
finally , we give a summary and perspective in sec .
[ cando ] .
quantum effects become increasingly relevant as components of information processing devices shrink to the few - nanometer scale .
this poses a challenge but also offers the chance to exploit quantum mechanics for data processing [ ] .
the spin is the canonical example of a quantum mechanical degree of freedom with two eigenstates . in solids
, it is possible to use spin - orbit coupling and many other spin dependent effects to manipulate the spin with electric fields and hence much faster than with external magnetic fields . in order to prepare , manipulate , and detect spin in an active and controlled way
, spintronic devices may take advantage of any such effects [ ] .
in contrast to electronics where carrier mobilities and lifetimes matter , spin mobilities and coherence times are relevant parameters for spintronics .
the coherence of electron spins is mainly determined by the spin - orbit interaction ( soi ) , the abundance of nuclear spins in the host material , and magnetic behavior .
these characteristics depend on the material and interlink spintronics with material science . due to their lower magnetic moment , nuclear spins couple weakly to the environment and exhibit typical coherence times orders of magnitude longer than for electron spins [ ] .
spintronics promises new applications and substantially better performance than standard micro- and nanoelectronics by exploiting the spin degree of freedom .
hard drive read heads are a multibillion dollar industry and magnetoresistive random - access memory ( mram ) is still waiting in the wings .
both technologies are based on the giant magnetoresistant effect , whose discovery in the late 1980s may be considered the birth of spintronics [ ] . beyond classical computing ,
spintronics has the potential for key components of quantum computing [ ] .
quantum computing can be viewed as a coherent superposition of classical computations running in parallel , with algorithms like the one by shor exploiting this [ ] . obviously , such algorithms require appropriate quantum hardware [ ] .
as proposed by loss and divincenzo , spin states of coupled qds can be used to implement a universal set of one- and two - qubit gates ( _ qubit _ is short for quantum bit ) .
spin qubit gates can be realized with magnetic fields and , using spin - dependent effects , also with electric fields , or a combination of both [ ] . such a solid state approach to quantum computing benefits from the scalability known from standard electronics and from long coherence times [ ] . the dot regions to which an integer number of electrons can be confined are typically on the length scale of 10 - 100 nanometers .
the eigenstate spectrum is discrete and quantum dots are often referred to as _ artificial atoms_. coulomb repulsion and the exchange coupling also play a role [ ] . also for spins that are confined to the qd , there remains a coupling to the environment . at finite temperatures , spins and phonons in the host material couple via the admixture mechanism , which involves spin - orbit interaction and electron - phonon coupling , or directly , via spin - orbit interaction only [ ] .
moreover , the hyperfine interaction mediates a coupling between the electron spin and nuclear spins in its vicinity [ ] .
these mechanisms lead to a finite spin lifetime , an important figure of merit for applications like the loss - divincenzo quantum computer where it should exceed the clock time by orders of magnitude [ ] . the theoretical prediction and explanation of spin lifetimes ( in graphene qds ) is part of this review .
its many outstanding properties make graphene , a monatomically thin layer of carbon atoms , an interesting material for a variety of applications , including spintronics .
the low spin - orbit coupling and a small abundance of nuclear spins are beneficial for long spin coherence times [ ] . in table
[ t1 ] , we compare the main sources of decoherence for different materials .
electron mobilities in graphene are typically high and allow for spin transport over micrometer length scales [ ] . while graphene is naturally non - magnetic , it can exhibit magnetic behavior under certain conditions [ ] .
more recently , there has been an effort to stack graphene and other two - dimensional materials like hexagonal boron nitride ( hbn ) and tmdcs on top of each other , held together by van der waals forces [ ] .
it is envisioned that by combining different materials in such van der waals heterostructures , one may tailor physical properties .
the concept is akin to bandgap engineering in semiconductor alloys and might once again highlight the connection between spintronics and materials science .
the hexagonal structure of graphene comes from the sp@xmath2-hybrid formed by three of the four outer - shell electrons in carbon .
the resulting @xmath6-bands are energetically far away from the charge neutrality point and play no role in transport .
the remaining p@xmath7 orbitals form the @xmath3-bands responsible for transport . in the low - energy approximation , they obey the quasi - relativistic dirac hamiltonian [ ] @xmath8 where @xmath9 labels the dirac valleys @xmath10 ( @xmath11 ) and @xmath12 ( @xmath13 ) in reciprocal space , @xmath14 are pauli matrices in standard representation that describe the sublattice , and @xmath15 is the wave vector measured w.r.t .
@xmath10 or @xmath12 , respectively .
the sublattice splitting @xmath16 vanishes for ideal graphene but may occur due to a substrate [ ] .
an electrostatic potential may be considered via the term @xmath17 .
the eigenspectrum of eq .
( [ dirac ] ) with @xmath18 is given by @xmath19 and has no gap if @xmath20 is possible and @xmath21 .
in that case , it is not possible to confine electrons electrostatically because of klein tunneling [ ] . in sec .
[ t1sec ] , we will discuss special cases where the energy spectrum becomes gapped , thus enabling electrostatic confinement .
.sources of decoherence in different materials .
if two spin - orbit splittings are listed , the upper ( lower ) number corresponds to the conduction ( valence ) band .
data on the different isotopes is taken from ref .
[ ] . + @xmath22 data for a ( 4,4 ) armchair carbon nanotube .
the spin - orbit coupling scales inversely with the nanotube radius [ ] .
+ @xmath23 the isotopes @xmath24mo , @xmath25mo , @xmath26mo , @xmath27mo , and @xmath28mo have nuclear spin 0 and a combined natural abundance of 75% .
[ cols="<,<,<",options="header " , ] [ t3 ] in graphene , we consider two mechanisms for the electron - phonon coupling , namely the deformation potential @xmath29 and the bond - length change @xmath30 , @xmath31 in a minimal coupling picture , the form of @xmath32 , where @xmath33 , highlights the equivalence of strain and a valley - dependent magnetic field , see eq .
( [ dirac ] ) and refs . [ ] .
the strain tensor is given by @xmath34 and the coupling constants are @xmath35 and @xmath36 [ ] .
we point out that all possible in - plane strains @xmath37 ( @xmath38 ) are considered by eq .
( [ epc ] ) .
the admixture mechanism creates a lowest - order contribution for in - plane deformations @xmath39 but not for out - of - plane deformations @xmath40 since a local tilt of the graphene lattice does not give rise to finite @xmath41 .
out - of - plane phonons can still induce a spin flip via the intrinsic spin - orbit interaction @xmath42 , given in eq .
( [ soi ] ) , or in higher order , where stretching effects lead to nonzero @xmath41 . in the former case a small local tilt @xmath43 , where @xmath44 , changes the mutual orientation of pseudospin @xmath14 and real spin @xmath45 . as a consequence ,
the intrinsic spin - orbit coupling acquires spin - flip terms , @xmath46 and thus gives rise to direct spin - phonon coupling @xmath47 [ ] .
the origin of this mechanism is illustrated in fig .
[ pic2 ] ( b ) .
below , we discuss the relaxation times of the electron spin in specific graphene qds where the electron spectrum is gapped and where the valley degeneracy is lifted . to flip its spin via a virtual transition to a different orbital state @xmath48 .
both electron - phonon coupling @xmath49 and rashba - type spin - orbit interaction @xmath50 couple the different orbits but only @xmath50 leads to a spin flip , thus effectively enabling a spin flip within the orbital state @xmath51 .
( b ) due to out - of - plane phonons , the local coordinate frame of the graphene lattice , @xmath52 , that describes pseudospin @xmath53 may differ from the laboratory frame @xmath54 that fixes the quantization axis of the electron spin @xmath45 in an external magnetic field . in the local lattice frame , @xmath55 may depend on all components of @xmath45 , as indicated by the red arrows .
this changes the intrinsic spin - orbit coupling to the form of eq .
( [ directspc ] ) , where the primes have been dropped . ] in graphene nanoribbons ( gnrs ) , electrons are naturally confined within the quasi one - dimensional structure and can move freely only along the longitudinal direction ( @xmath56-axis in fig .
[ pic3 ] ) . and @xmath57 .
the resulting bandgap allows for electrostatic confinement along the @xmath56-axis .
a potential @xmath58 defines the barrier regions b1 and b2 ( shaded ) as well as the dot region d in between .
( b ) for @xmath59 , there is one or more bound state(s ) inside the quantum dot ( green lines ) .
a continuum of extended states exists for energies above the confinement potential ( pink shade ) .
both bound and extended states provide virtual intermediate levels for the admixture mechanism responsible for spin relaxation . ] while the longitudinal wavenumber @xmath60 may vary continuously , the transverse wavenumber @xmath61 is discrete and depends on the ribbon width @xmath62 as well as the specific boundary conditions . in gnrs with lateral armchair boundaries ,
all components of the wavefunction spinor must vanish on the outermost atoms , which implies @xmath63 where @xmath64 labels the transverse excitations .
the width of the armchair gnr , @xmath65 , is determined by @xmath66 and @xmath67 , see fig .
[ pic3 ] ( a ) and ref .
[ ] . for a ribbon with @xmath68 , all transverse wavenumbers @xmath61 will be finite , thus leading to a gapped phonon spectrum with @xmath69 . for @xmath70
, this amounts to @xmath71 .
the gap allows for electrostatic confinement of electrons along the longitudinal direction by virtue of an electrostatic potential @xmath58 .
analytical solutions for bound states are known if the potential has the piecewise constant form @xmath72 , where @xmath73 is the longitudinal dimension of the quantum dot , see fig .
[ pic3 ] ( a ) and ref .
at least one bound state exists if the confinement potential does not exceed the bandgap , @xmath59 . depending on @xmath74 , there may be several bound states and independent of @xmath74 , there is always a continuum of extended states energetically above @xmath74 . since @xmath75 is typically of the order of @xmath74 , we consider both bound and extended states for virtual transitions within the admixture mechanism . due to the boundary conditions ,
the electronic states become superpositions of @xmath10 and @xmath12 with equal weight on both valleys .
this has important consequences .
first , the valley degeneracy is lifted , thus circumventing the necessity of a complicated exchange interaction . nevertheless , the orbital system is symmetric under time reversal and the spin states in a certain orbit form a kramers pair at zero magnetic field , which suppresses spin relaxation due to van vleck cancellation .
furthermore , the direct spin - phonon coupling @xmath47 does not allow for spin relaxation as the according matrix element is proportional to the valley index @xmath9 and hence vanishes after summing up the contributions from @xmath10 and @xmath12 .
the spin may still relax via @xmath41 and the admixture mechanism .
( [ fermi ] ) , phonon emission [ absorption ] is proportional to @xmath76 [ @xmath77 , where the phonon number @xmath78 is determined by the bose - einstein distribution . here , we assume phonon vacuum , i.e. , pure phonon emission .
the spin relaxation depends on the confinement potential @xmath74 , the qd aspect ration @xmath79 , and the applied magnetic field @xmath80 , which is assumed to be perpendicular to the graphene plane .
moreover , the mechanical boundary conditions determine the phonon spectrum and hence also the relaxation rate . in fig .
[ pic4 ] , we plot the spin relaxation rate @xmath81 in blue as calculated with eq .
( [ fermi ] ) for @xmath82 , @xmath83 , and free lateral boundaries [ ] . in agreement with eq .
( [ t1estimate ] ) , the doubly logarithmic scale highlights the behavior @xmath84 for @xmath85 , where @xmath86 is the dimensionless frequency of in - plane phonons with @xmath87 and @xmath88 as young s modulus and the mass density of graphene , respectively [ ] . due to destructive interference of processes originating from @xmath89 and @xmath32 ,
the relaxation rate features two dips for @xmath90 .
a van hove singularity of the phonon density of states @xmath91 leads to a divergence around @xmath92 a common effect for quasi one - dimensional systems [ ] . of the qd ground state for @xmath82 , @xmath83 , and free mechanical boundaries [ ] .
the doubly logarithmic scale highlights the behavior @xmath84 for @xmath85 , in agreement with eq .
( [ t1estimate ] ) .
the dimensionless phonon frequency @xmath93 is explained in the main text .
destructive interference of processes originating from @xmath89 and @xmath32 causes two dips of @xmath94 for @xmath90 and a van hove singularity of the phonon density of states leads to a divergence around @xmath92 . ]
results for different values of @xmath74 and @xmath79 , or for fixed mechanical boundary conditions can be found in ref .
as one would expect , virtual transitions to extended states gain importance as @xmath95 decreases and even dominate over virtual transitions to bound states for @xmath96 .
fixed mechanical boundaries lead to a gapped phonon spectrum which excludes spin relaxation via single - phonon processes for @xmath97 ( or @xmath98 for @xmath70 ) . as a quasi one - dimensional system ,
the armchair gnr is particularly interesting for spintronics - based quantum computing as it leaves two spatial dimensions for system periphery in a scaled - up device with multiple qds along the gnr .
the wave functions of bound states also extend to neighboring qds and thus allow for multiple - qubit operations .
moreover , the small bandgap allows the coupling of non - neighboring qds via detuning of the intermediate qds , which allows for more efficient quantum computation since shuttling of qubits can be avoided and thus raises the threshold for fault - tolerant quantum computing [ ] . gate - tunable qds with an electrostatic confinement potential can exist in 2d bulk mono- and bilayer graphene .
analytical solutions exist if the systems have circular symmetry in the graphene plane and the confinement potential has a rectangular profile , @xmath99 , where @xmath100 is the radial coordinate and @xmath101 is the radius of the disk - shaped quantum dot . due to the absence of intervalley scattering
, the orbital degeneracy between valleys @xmath10 and @xmath12 can be broken in a controlled way by the inclusion of a mass term a sublattice splitting for monolayer graphene or a layer bias in the case of bilayer graphene in combination with a magnetic field .
in particular , the splitting of orbits in opposite valleys can be tuned with the applied magnetic field .
in addition to a valley splitting , the mass term also gives rise to the required bandgap [ ] .
for magnetic fields , where the magnetic length @xmath102 is much smaller than @xmath101 , the discrete qd spectrum converges to the bulk landau levels .
below , we outline the analytical solution of the eigensystem both for mono- and for bilayer graphene .
we first treat monolayer graphene disks and bilayer graphene afterwards . as is obvious from eq .
( [ spectrum ] ) , a sublattice potential @xmath16 will open up a bandgap of @xmath103 .
a sublattice potential arises when graphene lies on a substrate but the two sublattices couple differently to the substrate material as , e.g. , for graphene on hbn , where @xmath104 has been observed [ ] .
while electron spin is neglected , here , the perpendicular component of the magnetic field is taken into account via minimal coupling to the motion of the charge carriers .
pseudospin can be included in the total angular momentum , @xmath105 , which commutes with the hamiltonian . as a result of the circular symmetry
, the azimuthal dependence can be separated away , leaving a differential equation for the radial dependence .
both for the dot region ( @xmath106 ) and for the barrier region ( @xmath107 ) , the solutions are found to be hypergeometric functions .
the matching condition for @xmath108 leads to a discrete spectrum of bound states .
the valley degeneracy of @xmath10 and @xmath12 is lifted for finite magnetic field and exceeds the zeeman splitting for experimentally accessible parameters [ ] . for the ground state of a circular qd in monolayer graphene and a perpendicular magnetic field .
the qd radius is @xmath109 and here , the bandgap @xmath103 is twice the confinement potential , @xmath110 .
the coupling constants @xmath111 and @xmath112 indicate the associated mechanisms , @xmath89 and @xmath32 , respectively .
both la and ta in - plane modes disperse linearly while the out - of - plane mode has a transitions from quadratic ( za ) to linear ( za ) dispersion .
the sum of all relaxation channels is represented by the red dotted , blue dashed , and black solid lines .
a quadratic ( linear ) dispersion of the out - of - plane modes is assumed for the red ( blue ) curve , and a crossover between them for the black curve .
the inset shows the universal dependence of @xmath113 on the inclination angle @xmath114 of the magnetic field [ ]
. aps . ] as for spins in armchair gnr quantum dots , relaxation via the admixture mechanism and via direct spin - phonon coupling will be considered in the following .
assuming @xmath115 , no continuum of extended states has been considered in ref .
[ ] . as a result of the broken valley degeneracy
, the spin states do not form a kramers pair and van vleck cancellation does not occur in circular qds in monolayer graphene .
the spectrum of acoustic phonons consists of three branches , transverse ( ta ) and longitudinal ( la ) in - plane modes , and one out - of - plane mode ( za ) . near the center of the brillouin zone ,
the in - plane modes disperse linearly and the za modes quadratically with a crossover to linear behavior .
only the la mode changes the size of the unit cell and is thus the only mode where the deformation potential ( @xmath89 ) is nonzero .
both la and ta deform the unit cell and hence generate a finite coupling via the bond - length change ( @xmath32 ) .
za modes give rise to direct spin - phonon coupling ( @xmath47 ) .
the according spin relaxation times for an electron in the qd ground state and phonon vacuum have been calculated in ref . [ ] with specific parameters .
the results are shown in fig .
[ pic5 ] . for excited qd states
, the relaxation times decrease quickly . in a two - dimensional material ,
the minimal coupling prescription involves the perpendicular components of the @xmath80-field .
but the zeeman energy , which is transformed to lattice excitations by the mentioned couplings , depends on the entire magnetic field strength . as a result
, all relaxation rates also depend on the inclination angle @xmath114 between the vector of the magnetic field and the normal vector of the graphene plane . for purely in - plane magnetic fields ,
the relaxation rates @xmath94 are reduced by 50% compared with the case of a perpendicular field , see inset in fig .
[ pic5 ] . in bilayer graphene , a gap opens if the electrostatic potential is different for the two layers .
then , an electrostatic confinement potential can be used to confine charge carriers in a quantum dot .
such systems are studied in experiment [ ] . with appropriate top and bottom gates ,
the bandgap as well as the confinement potential can be tuned electrically and independently [ ] . in a simple description ,
the system consists of two individual graphene layers , where one sublattice of the upper layer couples to one sublattice of the lower layer with an interlayer hopping amplitude @xmath116 [ ] .
the inversion symmetry of the bilayer system is broken by the potential difference between the distinct layers .
as for the case of monolayer graphene above , the problem can be solved analytically for a potential of the form @xmath99 . due to the contribution of the pseudospin of both layers ,
the total angular momentum is an integer here .
the analytic solution of the eigensystem follows the same procedure as for the monolayer case but is more complicated because of the two coupled layers .
the radial parts of the wavefunctions again turn out to be hypergeometric functions both inside and outside the dot region , and the matching condition at @xmath108 leads to a discrete spectrum of bound states .
the valley degeneracy for @xmath117 can be lifted with a finite magnetic field perpendicular to the graphene plane and for @xmath118 , the energy spectrum approaches the landau level structure .
in particular , there are no kramers pairs for a finite magnetic field perpendicular to the graphene plane . to our knowledge
, the spin relaxation time @xmath0 of a bound state inside the qd has not been estimated yet . for acoustic phonons neat the center of the brillouin zone
, the atoms within a unit cell do not vibrate against each other .
moreover , the interlayer van der waals bonds of bilayer graphene are much less rigid than the intralayer @xmath6-bond . as a consequence ,
the change of the interlayer distance can be neglected and the individual layers can be treated completely independently for small wavenumbers . the deformation potential and the bond - length change within each layer are thus given by two copies of eq .
( [ epc ] ) , one for each layer . using the spin - orbit coupling of bilayer graphene , @xmath94 can in principle
be calculated via the admixture mechanism as for monolayer graphene [ ] . as for monolayer graphene ,
out - of - plane modes change the mutual orientation of real spin @xmath45 and lattice related spins ( pseudospin @xmath14 and for bilayer graphene also the layer spin @xmath119 [ ] ) .
the resulting direct spin - phonon coupling might turn out to be more complicated than that for monolayer graphene . for coupling to the in - plane modes of bilayer graphene , eqs .
( [ t1estimate ] ) and ( [ epc ] ) suggest a behavior @xmath120 or @xmath121 , depending on the order of the dipole approximation for the specific mechanism .
both ferromagnetic and antiferromagnetic materials are important for data processing and storage ( e.g. mram ) [ ] . in contrast to typical magnetic elements like fe , ni , co , and rare earths , carbon does not contain occupied d- or f - orbitals and pristine graphene is strongly diamagnetic [ ] . nevertheless , the prospects of magnetic ordering at room temperature and unmatched data density in monatomically thin layers motivate intense research efforts aimed at magnetism in graphene .
magnetic moments may appear in graphene because of vacancies [ ] , light or heavy adatoms [ ] , edge effects [ ] , or molecular doping [ ] .
the resulting magnetic structures may be 0d , 1d , or 2d , depending on the origin of magnetism ( e.g. , isolated defects , gnr edges , or molecular doping of bulk graphene , respectively ) .
the experimental detection of magnetic moments is possible via squid ( superconducting quantum interference device ) magnetometry [ ] , spin transport measurement [ ] , or spin - sensitive stm ( scanning tunneling microscopy ) [ ] . in the following ,
we focus on magnetic moments induced by vacancies or light adatoms . according to lieb s theorem ,
the magnetic moment of the ground state of the hubbard model for graphene is given by @xmath122 , where @xmath123 and @xmath124 are the numbers of sublattice sites [ ] .
carbon atoms can be removed from the graphene lattice by means of irradiation with electrons or argon ions [ ] .
alternatively , light adatoms like h and f passivate the p@xmath7 orbital of the carbon atom to which they bond , thus effectively removing this carbon atom from its sublattice .
a hbn substrate may be used to stabilize the hydrogen adsorption on one sublattice and to suppress migration of the adatoms [ ] .
curie temperatures for the ferromagnetic ground state can exceed @xmath125 [ ] . the above defects , i.e. , vacancies and light adatoms , are similar to quantum dots in that they lead to localized states . in ref . [ ] , we have studied the exchange coupling between two spin states localized at two separate yet nearby vacancies in a graphene nanoflake ( gnf ) .
the need for a bandgap does not play a role in gnfs since electrons are naturally confined within such quasi zero - dimensional structures . specifically , we consider gnfs with hexagonal symmetry as found for flakes grown by chemical vapor deposition ( cvd ) [ ] and either zigzag or armchair terminations .
two atoms are removed at sites @xmath126 w.r.t . the flake center such that the lattice retains the mirror symmetries @xmath127 and @xmath128 , see fig .
[ pic6 ] .
we describe the electronic system with a full tight - binding calculation that involves hoppings @xmath129 between lattice sites @xmath130 and @xmath131 up to third nearest neighbors ( @xmath132 ) . a perpendicular magnetic field
is included in the hoppings via the peierls phase .
the resulting spectrum is discrete and may or may not feature degeneracies , depending on the exact gnf configuration , @xmath80-field , and maximal order of nearest neighbor hoppings . .
( a ) antibonding eigenstate @xmath133 . here
, we only consider nearest neighbor hopping such that the ( anti-)bonding energy eigenstates @xmath134 and @xmath135 lie in the middle of the spectrum and have the same on - site probability densities .
( b ) localized vacancy state @xmath136 .
the probability density of @xmath137 looks similar yet mirrored about the @xmath138-axis [ ] . ]
the two nearby vacancies can be interpreted as a double quantum dot ( dqd ) , whose localized states form a bonding and an antibonding eigenstate within the gnf spectrum .
if the energy splitting @xmath16 between the bonding and antibonding eigenstates is much smaller than their energy differences to the remaining spectrum , the orbital part of the dqd can be described by @xmath139 where we employ the localized states @xmath140 as a basis , @xmath141 is the mean energy of the ( anti-)bonding states , and where the intervacancy hopping with @xmath142 may acquire a peierls phase due to the magnetic field .
symmetry allows us to find the explicit form of @xmath143 , see fig .
we complement this picture with an on - site coulomb repulsion @xmath144 and aim to calculate the exchange coupling @xmath145 , which energetically splits the singlet state @xmath146 from the triplet state @xmath147 and determines their mutual dynamics . within this model and for @xmath148 , we find that the singlet state is preferred , i.e. , antiferromagnetic ordering , and @xmath149 for the exchange coupling .
both @xmath150 and @xmath144 , and hence @xmath145 , depend on the perpendicular magnetic field that is included from the beginning .
typically , the magnetic field has a significant effect when the total magnetic flux through the gnf , @xmath151 , where @xmath152 is the surface area of the nanoflake , reaches one flux quantum , @xmath153 , where @xmath154 is planck s constant and @xmath155 is the unit charge .
( solid red line and left axis ) and the eigenenergies ( dashed black lines and right axis ) of the according ( anti-)bonding states are plotted against the perpendicular magnetic field .
the shown data belongs to a nanoflake with zigzag terminations and a total 598 atoms [ ] .
two vacancies are located at @xmath156 .
only nearest neighbor hopping is taken into account , such that the ( anti-)bonding energy eigenstates @xmath157 and @xmath158 lie in the middle of the spectrum .
one flux quantum passes through the flake at @xmath159 . ]
depending on the overall size of the gnf , the distance between the vacancies , and the gnf termination ( zigzag or armchair ) , the @xmath80-field dependence can be used to tune @xmath145 over several orders of magnitude . for some flake configurations ,
it is possible to tune the spectrum into a degeneracy , @xmath160 , thus switching off the antiferromagnetic order , see fig .
such an _ in - situ _ tunability of the exchange coupling is most promising for spintronics applications as it allows to change the magnetic behavior without preparing a new device . for flakes with armchair edges , we find that @xmath145 increases for gnfs of larger overall size and for smaller separation distances of the two vacancies .
bulk graphene can be viewed as the limit where the gnf diameter is much greater than the vacancy separation and where edge effects become negligible .
we expect that the former trend saturates in this limit .
the latter trend applies separately for two cases : ( i ) when the atom that has been removed for the vacancy at @xmath161 belongs the a sublattice and ( ii ) when it belongs to the b sublattice .
such trends are not obvious for zigzag terminated flakes , where edge effects play a significant role for gnfs with up to @xmath162 atoms [ ] .
nel temperatures , @xmath163 with the boltzmann constant @xmath164 , range from below @xmath165 to beyond @xmath125 and thus allow for experimental verification of these results , possibly via the techniques mentioned at the beginning of this section . including a non - local coulomb interaction might lead to ferromagnetic ordering , @xmath166 , and possibly enable an _ in - situ _ tunable crossover from antiferromagnetism to ferromagnetism [ ] .
our method relies on symmetry and hence the removal of an @xmath152 and a @xmath80 atom .
nevertheless , ferromagnetic ordering is expected if the two atoms are removed from the same sublattice and only on - site coulomb interaction is considered [ ] .
we also point out that the exchange coupling @xmath145 can be used for coherent two - qubit operations , as required for universal quantum computing [ ] .
naturally , only 1% of all carbon atoms belong to the isotope @xmath1c , which is the only stable carbon isotope with finite nuclear spin , see table [ t1 ] .
this abundance can be lowered by isotopic purification [ ] .
possibly , this is the reason why the combination of nuclear spins and graphene has so far received little attention from the scientific community .
however , there are at least three situations where nuclear spins become important for spintronics in graphene .
( i ) first , it has been found theoretically that @xmath0 relaxation times of electron spins in graphene qds can exceed @xmath167 for certain parameters and if nuclear spins are neglected , see figs . [ pic4 ] and [ pic5 ] or refs .
[ ] . in this case ,
the spin dephasing time @xmath168 might be limited by the coupling to nuclear spins of the @xmath1c nuclei .
( ii ) then , state - of - the - art graphene qds often employ a graphene ( bi-)layer encapsulated by bottom and top sheets of hexagonal boron nitride . neither boron nor nitrogen posses any stable isotopes without nuclear spin , see table [ t1 ] .
the spin of an electron in such an encapsulated graphene qd is in immediate contact with the nuclear spins of hbn .
( iii ) finally , the nuclear spin of an impurity , of an adatom , or of @xmath1c itself might be interesting for spintronics .
natural monolayer graphene contains one nuclear spin per @xmath169 surface area .
this density can be controlled by enriching or depleting @xmath1c .
possibly , control and coupling of nuclear spins in the graphene sheet or on top of it can be achieved .
ultimately , this could lead to 2d nuclear spintronics with similarities to nitrogen vacancy centers in diamond , where @xmath1c nuclear spins have been used for quantum operations [ ] .
the hamiltonian for an ensemble of nuclear spins in a solid has five different contributions [ ] , @xmath170 the term @xmath171 arises due to the zeeman energies of nuclear spins in a magnetic field @xmath80 .
the second term describes the hyperfine interaction between electron spins and nuclear spins .
it is the sum of two terms , @xmath172 .
the contact ( isotropic ) hyperfine interaction can be modeled as an overhauser field , @xmath173 , that acts on the electron spin @xmath45 in a similar way as an external magnetic field , @xmath174 here , @xmath175 is the spin operator for the nucleus at site @xmath176 and the coupling strength @xmath177 is proportional to the local probability density of the electron and the gyromagnetic ratio of the nucleus at site @xmath176 [ ] .
( [ overhauser ] ) vanishes for @xmath3-band electrons ( in particular those of graphene ) as their probability densities at the nuclear sites are zero due to odd symmetry . in this case ( in particular for graphene ) , the anisotropic hyperfine interaction , @xmath178 , and the coupling of nuclear spin and electron orbit , @xmath179 , dominate electron - nuclear coupling .
in addition to @xmath180 , also @xmath179 can mediate a coupling of nuclear and electron spins if spin - orbit interaction is included .
the nuclear dipole - dipole hamiltonian , @xmath181 , directly couples the magnetic moments of distinct nuclear spins . and
nuclei with spin @xmath182 ( thus excluding @xmath1c ) interact with an electric field gradient via the nuclear quadrupolar coupling , @xmath183 .
all these terms are discussed in detail in ref . [ ] .
spintronics with nuclear spins point ( iii ) in our list above would rely on the coupling of nuclear spins to external fields , which is achieved by @xmath171 and @xmath183 , as well as the coupling between different nuclear spins , which is mediated by @xmath181 . to our knowledge ,
the effect of nuclear spins from hbn on the electron spin in encapsulated graphene qds point ( ii ) have not been studied , yet . in the remainder of this section
, we review studies on point ( i ) , the decoherence of the electron spin due to hyperfine interaction with the nuclear spins of @xmath1c .
the contact hyperfine interaction vanishes for flat graphene and indirect coupling of electron and nuclear spins via @xmath179 and spin - orbit interaction is expected to be small and hence neglected .
consequently , the coupling between electron and nuclear spins in graphene is given by @xmath184 where @xmath176 indexes the sites of @xmath1c isotopes .
the coupling constants @xmath185 are proportional to the probability density of the electron at site @xmath176 and the respective coupling strength @xmath186 , with @xmath187 [ ] .
the @xmath188-axis is chosen to be perpendicular to the graphene sheet .
in contrast , the coupling strength of the ( contact ) hyperfine interaction in gaas ranges from @xmath189 to @xmath190 ( depending on the specific isotope ) , thus exceeding that of graphene by more than one order of magnitude [ ] .
the importance of the hyperfine interaction in graphene has been studied by comparing cvd - grown samples consisting entirely of @xmath191c and @xmath1c , respectively [ ] .
the experimental results indicate that in contrast to the cases of gaas or carbon nanotubes , hyperfine interactions with @xmath1c nuclear spins have a negligible effect on spin transport relaxation times in cvd - grown graphene [ ] . ) for various parameters on a doubly logarithmic scale .
the longitudinal hyperfine coupling strength in graphene is @xmath192 [ ] .
the polarization of the nuclear spin system is assumed to be @xmath193 for the solid lines and @xmath194 for the dashed ones .
for the thin black lines ( left and bottom axes ) , we vary the relative @xmath1c abundance but fix the total number of atoms to @xmath195 , which corresponds to a square qd with an approximate size of @xmath196 . for the thick orange lines ( right and top axes ) ,
we vary @xmath197 and fix the relative @xmath1c abundance to its natural value , @xmath198 . ]
the interaction of an electron spin in a graphene qd with a certain number of nuclear spins , @xmath199 , will affect the time evolution of its transversal component @xmath200 , where @xmath201 .
if @xmath202 , the nuclear spins form a bath and one can use the central limit theorem to obtain @xmath203 [ ] . in the presence of a perpendicular magnetic field @xmath204 , the electron zeeman energy is much greater than the transversal hyperfine coupling strength @xmath205 .
then , the characteristic decoherence time @xmath206 is given by @xmath207 where @xmath208 denotes the polarization of the nuclear spin system and @xmath197 is the total number of carbon atoms , i.e. , @xmath198 for natural carbon [ ] .
typical values for @xmath206 exceed @xmath209 , see fig .
[ pic8 ] . in a further study
, it was found that in the same situation , the longitudinal spin component @xmath210 is conserved up to small corrections , which oscillate with a frequency determined by the hyperfine interaction [ ] .
if less than roughly 10 nuclear spins interact with the electron spin in the graphene qd , the nuclear spin system can not be modeled as bath but allows for an exact treatment .
then , decoherence times depend on the exact position and orientation of the nuclear spins , and typically lie in the regime of @xmath211 [ ] .
spintronics with individual electron spins in semiconductors has evolved dramatically from basic concepts to universal quantum control of spin states and quantum registers with up to four qds [ ] .
isotopic purification enables coherence times beyond @xmath212 at low temperature in silicon and beyond @xmath211 at room temperature in diamond [ ] .
despite the challenge of efficient spin injection into graphene through tunneling barriers , remarkable improvements in spin transport have been achieved over the past few years [ ] .
in contrast to electron mobility , the spin lifetime seems to be rather insensitive to charged impurity scattering [ ] and can exceed @xmath213 at room temperature [ ] , even for cvd - grown graphene [ ] .
the small intrinsic spin - orbit interaction makes it practically impossible to observe the spin hall effect ( she ) in graphene [ ] .
but controlled hydrogenation can increase the spin - orbit interaction by three orders of magnitude and thus enable the observation of a she at room temperature and zero magnetic field [ ] .
last year , a quantum spin hall state has been observed in graphene with a large , tilted magnetic field [ ] .
graphene qds have been studied thoroughly w.r.t . their orbital spectrum [ ] , spin - filling sequence [ ] , and charge relaxation times [ ] .
moreover , high - quality bilayer qds , encapsulated in hbn and equipped with electric gates , allow for electrostatic confinement as discussed in subsec .
[ mbgds ] .
small qds in the @xmath214 range can be created by electroburning and exhibit coulomb blockade at room temperature due to addition energies as large as @xmath215 [ ] .
slightly larger graphene nanoflakes , epitaxially grown on ir(111 ) and intercalated with oxygen , are reported to exhibit a linear spectrum [ ] . despite these promising proceedings with graphene - based qds , pauli blockade a basic step towards spintronics [ ] and already achieved in carbon nanotubes [ ] has ( to our knowledge ) not been observed in graphene , yet .
gapped armchair gnrs , subsec .
[ t1gnr ] , and symmetrical graphene nanoflakes , sec .
[ gnf ] , rely on precise edges .
atomically accurate armchair gnrs can be produced bottom - up , with widths up to @xmath216 , from aromatic precursor molecules [ ] or top - down .
one top - down technique that leads to precise edges is the unzipping of carbon nanotubes with an abruptly expanding nitrogen gas inside the nanotubes [ ] .
another possibility is the electron - beam induced mechanical rupture of bulk graphene [ ] .
the latter method can be used to produce armchair edges as well as zigzag edges and can be applied in high vacuum , thus leading to minimally contaminated samples . beyond this , clean structuring methods for graphene include nano - etching of suspended graphene [ ] and the use of a silicon atom as a monatomic chisel [ ] .
we have given an overview of spintronics with electron spin states in graphene quantum dots and nanoflakes .
long coherence times can be expected because of weak sources of decoherence , i.e. , small coupling strengths of the spin - orbit and hyperfine interactions , and a marginal abundance of nuclear spins in carbon . we have explained spin relaxation via spin - orbit mediated coupling to acoustic phonons and we have introduced a generic formula for the spin relaxation time @xmath0 for graphene and other systems .
valley degeneracy and the lack of a bandgap in bulk graphene pose challenges to spin operations in graphene quantum dots .
specific systems where these challenges can be overcome have been reviewed and , if applicable , studies on their spin relaxation times have been discussed .
magnetism can be induced in graphene by several means . here , we have focused on magnetism in graphene nanoflakes and _ in - situ _ tunability of antiferromagnetism in symmetric nanoflakes with defects .
we have summarized studies on the interaction of electron and nuclear spins and finally , we have pointed out recent experimental achievements relevant for spintronics with graphene quantum dots .
we believe that graphene has the potential for a top - notch spintronics material that can outperform established materials like silicon or gaas for specific applications .
the observation of the pauli blockade represents a fundamental step towards spintronics with graphene quantum dots . in principle , the necessary architectures exist already but suffer from poor tunability of the source and drain barriers and inefficient spin injection .
similarly , the components required for electrostatic confinement of electrons in armchair gnr quantum dots are already available but have not been integrated into one device , yet . with view to the rapid progress of device fabrication ,
we anticipate significant progress in these fields within the next 2 - 3 years .
ultimately , we see spintronics in graphene as a building block for 2d - heterostructured spintronics devices composed of ( functionalized ) graphene and other 2d materials like tmdcs and hbn .
we thank the european science foundation and the deutsche forschungsgemeinschaft ( dfg ) for support within the eurographene project congran and the dfg for funding within sfb 767 and for 912 .
wolf , d.d .
awschalom , r.a .
buhrman , j.m .
daughton , s. von molnar , m.l .
roukes , a.y .
chtchelkanova , and d.m .
treger , science * 294 * , 1488 ( 2001 ) . d.d .
awschalom , d. loss , and n. samarth , semiconductor spintronics and quandum computation ( springer , berlin , heidelberg , 2002 ) .
i. uti , j. fabian , and s. das sarma , rev . mod .
76 * , 323 ( 2004 ) .
castro neto , f. guinea , n.m.r .
peres , k.s .
novoselov , and a.k .
geim , rev .
phys . * 81 * , 109 ( 2009 ) .
katsnelson , graphene carbon in two dimensions ( cambridge university press , new york , 2012 ) .
gl , p. potasz , m. korkusinski , and p. hawrylak , graphene quantum dots ( springer , berlin , heidelberg , 2014 ) .
awschalom , l.c .
bassett , a.s .
dzurak , e.l .
hu , and j.r .
petta , science * 339 * , 1174 ( 2013 ) .
j. sinova and i. uti , nat . mater .
* 11 * , 368 ( 2012 ) .
r. jansen , nat .
mater . * 11 * , 400 ( 2012 ) .
p. recher and b. trauzettel , nanotechnology * 21 * , 302001 ( 2010 ) .
d. pesin and a.h .
macdonald , nat .
mater . *
11 * , 409 ( 2012 ) .
w. han , r.k .
kawakami , m. gmitra , and j. fabian , nat .
nanotechnol .
* 9 * , 794 ( 2014 ) .
b.z . jang and w.c .
huang , united states patent 7071258 , filed 21/10/2002 , granted 4/7/2006 .
novoselov , a.k .
geim , s.v .
morozov , d. jiang , y. zhang , s.v .
dubonos , i.v .
grigorieva , and a.a .
firsov , science * 306 * , 666 ( 2004 ) .
a.k . geim and k.s .
novoselov , nat .
* 6 * , 183 ( 2007 ) .
n. savage , nature * 483 * , s30 ( 2012 ) .
brodie , phil .
trans . * 149 * , 249 ( 1859 ) . c. lee , x. wei , j.w .
kysar , and j. hone , science * 321 * , 385 ( 2008 ) .
wallace , phys .
rev . * 71 * , 622 ( 1947 ) .
divincenzo and e.j .
mele , phys .
b * 29 * , 1685 ( 1984 ) .
semenoff , phys .
* 53 * , 2449 ( 1984 ) .
novoselov , a.k .
geim , s.v .
morozov , d. jiang , m.i .
katsnelson , i.v .
grigorieva , s.v .
dubonos , and a.a .
firsov , nature * 438 * , 197 ( 2005 ) . y .-
lin , c. dimitrakopoulos , k.a .
jenkins , d.b .
farmer , h .- y .
chiu , a. grill , and p. avouris , science * 327 * , 662 ( 2010 ) .
b. trauzettel , d.v .
bulaev , d.loss , and g. burkard , nature phys .
* 3 * , 192 ( 2007 ) .
c.l . kane and e.j .
mele , phys .
lett . * 95 * , 226801 ( 2005 ) .
h. min , j.e .
hill , n.a .
sinitsyn , b.r .
sahu , l. kleinman , and a.h .
macdonald , phys .
b * 74 * , 165310 ( 2006 ) .
m. gmitra , s. konschuh , c. ertler , c. ambrosch - draxl , and j. fabian , phys .
b * 80 * , 235431 ( 2009 ) .
s. konschuh , m. gmitra , and j. fabian , phys .
b * 82 * , 245412 ( 2010 ) .
g. audi , o. bersillon , j. blachot , and a.h .
wapstra , nucl .
a * 729 * , 3 ( 2003 ) .
nair , p. blake , a.n .
grigorenko , k.s .
novoselov , t.j .
booth , t. stauber , n.m.r .
peres , and a.k .
geim , science * 320 * , 1308 ( 2008 ) .
j. rioux and g. burkard , phys .
b * 90 * , 035210 ( 2014 ) .
m. inglot , v.k .
dugaev , e.y .
sherman , and j. barna , phys .
b * 89 * , 155411 ( 2014 ) .
o.v . yazyev and l. helm , phys .
b * 75 * , 125408 ( 2007 ) .
yazyev , rep .
prog . phys .
* 73 * , 056501 ( 2010 ) . c. tao , l. jiao , o.v .
yazyev , y .- c .
chen , j. feng , x. zhang , r.b .
capaz , j.m . tour , a. zettl , s.g .
louie , h. dai , and m.f .
crommie , nature phys .
* 7 * , 616 ( 2011 ) .
s. tarucha , d.g .
austing , t. honda , r.j .
van der hage , and l.p .
kouwenhoven , phys .
lett . * 77 * , 3613 ( 1996 ) .
kouwenhoven , d.g .
austing , and s. tarucha , rep .
phys . * 64 * , 701 ( 2001 ) .
r. hanson , l.p .
kouwenhoven , j.r .
petta , s. tarucha , and l.m.k .
vandersypen , rev .
phys . * 79 * , 1217 ( 2007 ) .
petta , a.c .
johnson , j.m .
taylor , e.a .
laird , a. yacoby , m.d .
lukin , c.m .
marcus , m.p .
hanson , and a.c .
gossard , science * 309 * , 2180 ( 2005 ) .
r. winkler and u. zlicke , phys .
b * 91 * , 205312 ( 2015 ) .
a. plyi , p.r . struck , m. rudner , k. flensburg and g. burkard , phys .
lett . * 108 * , 206811 ( 2012 ) .
h. wang and g. burkard , phys . rev .
b * 90 * , 035415 ( 2014 ) .
a. plyi and g. burkard , phys .
. lett . * 106 * , 086801 ( 2011 ) .
n. rohling , m. russ , and g. burkard , phys .
. lett . * 113 * , 176801 ( 2014 ) .
x. xu , w. yao , d. xiao , and t.f .
heinz , nature phys .
* 10 * , 343 ( 2014 ) .
a. kormnyos , v. zlyomi , n.d .
drummond , and g. burkard , phys .
x * 4 * , 011034 ( 2014 ) .
r.p . feynman , engineering and science * 23 * , 22 ( 1960 ) .
moore , electronics 114 ( 1965 ) .
thompson and s. parthasarathy , mater . today * 9 * , 20 ( 2006 )
dyakonov and v.i .
perel , phys .
a * 35 * , 459 ( 1971 ) .
s. datta and b. das , appl .
lett . * 56 * , 665 ( 1990 ) . y.k .
kato , r.c .
myers , a.c .
gossard , and d.d .
awschalom , science * 306 * , 1910 ( 2004 ) .
d. bhowmik , l. you , and s. salahuddin , nat .
nanotechnol .
* 9 * , 59 ( 2014 ) .
morton , a.m. tyryshkin , r.m .
brown , s. shankar , b.w .
lovett , a. ardavan , t. schenkel , e.e .
haller , j.w .
ager , and s.a .
lyon , nature * 455 * , 1085 ( 2008 ) .
baibich , j.m .
broto , a. fert , f. nguyen van dau , and f. petroff , phys .
* 61 * , 2427 ( 1988 ) .
g. binasch , p. grnberg , f. saurenbach , and w. zinn , phys .
b * 39 * , 4828 ( 1989 ) .
park , j. wunderlich , x. mart , v. hol , y. kurosaki , m. yamada , h. yamamoto , a. nishide , j. hayakawa , h. takahashi , a.b .
shick , and t. jungwirth , nat .
* 10 * , 347 ( 2011 ) .
d. loss and d.p .
divincenzo , phys .
a * 57 * , 120 ( 1998 ) .
nielsen and i.l .
chuang , quantum computation and quantum information ( cambridge university press , new york , 2000 ) .
p. shor , proceedings of the 35th annual symposium on the foundations of computer science , santa fe , us ( ieee press , los alamitos , 1994 ) , pp.124 - 134 .
divincenzo and d. loss , superlattice microst .
* 23 * , 419 ( 1998 ) .
a.m. tyryshkin , s. tojo , j.j.l .
morton , h. riemann , n.v .
abrosimov , p. becker , h .- j .
pohl , t. schenkel , m.l.w .
thewalt , k.m .
itoh , and s.a .
lyon , nat .
mater . * 11 * , 143 ( 2012 ) .
shulman , o.e .
dial , s.p .
harvey , h. bluhm , v. umansky , and a. yacoby , science * 336 * , 202 ( 2012 ) .
t. takakura , a. noiri , t. obata , t. otsuka , j. yoneda , k. yoshida , and s. tarucha , appl .
lett . * 104 * , 113109 ( 2014 ) .
a.v . khaetskii and y.v .
nazarov , phys .
b * 64 * , 125316 ( 2001 ) .
m.s . rudner and e.i .
rashba , phys .
b * 81 * , 125426 ( 2010 ) .
coish and j. baugh , phys .
status solidi b * 246 * , 2203 ( 2009 ) .
r. barends , j. kelly , a. megrant , a. veitia , d. sank , e. jeffrey , t.c .
white , j. mutus , a.g .
fowler , b. campbell , y. chen , z. chen , b. chiaro , a. dunsworth , c. neill , p. omalley , p. roushan , a. vainsencher , j. wenner , a.n .
korotkov , a.c .
cleland , and j.m .
martinis , nature * 508 * , 500 ( 2014 ) . m.h.d .
guimares , a. veligura , p.j .
zomer , t. maassen , i.j .
vera - marun , n. tombros , and b.j .
van wees , nano lett .
* 12 * , 3512 ( 2012 ) .
w. han , j .-
chen , d. wang , k.m .
mccreary , h. wen , a.g .
swartz , j. shi , and r. kawakami , nano lett .
* 12 * , 3443 ( 2012 ) .
m. drgeler , f. volmer , m. wolter , b. terrs , k. watanabe , t. taniguchi , g. gntherodt , c. stampfer , and b. beschoten , nano lett .
* 14 * , 6050 ( 2014 ) .
kamalakar , c. groenveld , a. dankert , and s.p .
dash , nat .
commun . * 6 * , 6766 ( 2015 ) .
m. droth and g. burkard , phys .
b * 91 * , 115439 ( 2015 ) .
novoselov , v.i .
falko , l. colombo , p.r .
gellert , m.g .
schwab , and k. kim , nature * 490 * , 192 ( 2012 ) .
geim and i.v .
grigorieva , nature * 499 * , 419 ( 2013 ) .
b. hunt , j.d .
sanchez - yamagishi , a.f .
young , m. yankowitz , b.j .
leroy , k. watanabe , t. taniguchi , p. moon , m. koshino , p. jarillo - herrero , and r.c .
ashoori , science * 340 * , 1427 ( 2013 ) .
x. cui , g .- h .
lee , y.d .
kim , g. arefe , p.y .
huang , c .- h .
lee , d.a .
chenet , x. zhang , l. wang , f. ye , f. pizzocchero , b.s .
jessen , k. watanabe , t. taniguchi , d.a .
muller , t. low , p. kim , and j. hone , nat .
nanotechnol .
doi:10.1038/nnano.2015.70 ( 2015 ) .
o. klein , z. phys . * 53 * , 157 ( 1929 ) .
katsnelson , k.s .
novoselov , and a.k .
geim , nature phys .
* 2 * , 620 ( 2006 ) .
d. huertas - hernando , f. guinea , and a. brataas , phys .
b * 74 * , 155426 ( 2006 ) .
s. abdelouahed , a. ernst , j. henk , i.v .
maznichenko , i. mertig , phys .
b * 82 * , 125424 ( 2010 ) .
s. konschuh , m. gmitra , d. kochan , and j. fabian , phys . rev .
b * 85 * , 115423 ( 2012 ) .
liu , w. feng , and y. yao , phys .
lett . * 107 * , 076802 ( 2011 ) .
r. winkler , spin - orbit coupling effects in two - dimensional electron and hole systems ( springer , berlin , heidelberg , 2003 ) . k. komider , j.w .
gonzlez , and j. fernndez - rossier , phys .
b * 88 * , 245436 ( 2013 ) .
n. tombros , c. jozsa , m. popinciuc , h.t .
jonkman , and b.j .
van wees , nature * 448 * , 571 ( 2007 ) .
p. boross , b. dra , a. kiss , and f. simon , sci . rep .
* 3 * , 3233 ( 2013 ) .
m. diez and g. burkard , phys .
b * 85 * , 195412 ( 2012 ) .
* 12 * , 3512 ( 2012 ) .
j. balakrishnan , g.k.w .
koon , m. jaiswal , a.h .
castro neto , and b. zyilmaz , nature phys .
* 9 * , 284 ( 2013 ) .
fedorov , m. gradhand , s. ostanin , i.v .
maznichenko , a. ernst , j. fabian , and i. mertig , phys .
lett . * 110 * , 156602 ( 2013 ) .
a. barreiro , h.s.j .
van der zant , and l.m.k .
vandersypen , nano lett .
* 12 * , 6096 ( 2012 ) .
liu , d. hug , and l.m.k .
vandersypen , nano lett .
* 10 * , 1623 ( 2010 ) .
allen , j. martin , and a. yacoby , nat .
commun . * 3 * , 934 ( 2012 ) .
a.m. goossens , s.c.m .
driessen , t.a .
baart , k. watanabe , t. taniguchi , and l.m.k .
vandersypen , nano lett .
* 12 * , 4656 ( 2012 ) .
a. varlet , m .- h .
liu , v. krueckl , d. bischoff , p. simonet , k. watanabe , t. taniguchi , k. richter , k. ensslin , and t. ihn , phys .
lett . * 113 * , 116601 ( 2014 ) . z.b .
tan , d. cox , t. nieminen , p. lhteenmki , d. golubev , g.b .
lesovik , and p.j .
hakonen , phys .
lett . * 114 * , 096602 ( 2015 ) .
g. balasubramanian , p. neumann , d. twitchen , m. markham , r. kolesov , n. mizuochi , j. isoya , j. achard , j. beck , j. tissler , v. jacques , p.r .
hemmer , f. jelezko , and j. wrachtrup , nat .
* 8 * , 383 ( 2009 ) .
n. rohling and g. burkard , new j. phys .
* 14 * , 083008 ( 2012 ) .
ponomarenko , f. schedin , m.i .
katsnelson , r. yang , e.w . hill , k.s .
novoselov , and a.k .
geim , science * 320 * , 356 ( 2008 ) .
s. schnez , j. gttinger , m. huefner , c. stampfer , k. ensslin , and t. ihn , phys .
b * 82 * , 165445 ( 2010 ) .
a. de martino , l. dellanna , and r. egger , phys .
lett . * 98 * , 066802 ( 2007 ) .
pereira , f. guinea , j.m.b
. lopes dos santos , n.m.r .
peres , and a.h .
castro neto , phys .
lett . * 96 * , 036801 ( 2006 ) .
recher , j. nilsson , g. burkard , and b. trauzettel , phys .
b * 79 * , 085407 ( 2009 ) . j.h .
van vleck , phys .
rev . * 57 * , 426 ( 1940 ) .
we use upper case indices for graphene hamiltonians and lower case indices for general hamiltonians .
y. zhang , z. jiang , j.p . small , m.s .
purewal , y .- w .
tan , m. fazlollahi , j.d .
chudow , j.a .
jaszczak , h.l .
stormer , and p. kim , phys .
* 96 * , 136806 ( 2006 ) .
falkovsky , phys .
a * 372 * , 5189 ( 2008 ) .
m. droth and g. burkard , phys .
b * 84 * , 155404 ( 2011 ) . m. droth and g. burkard , phys .
b * 87 * , 205432 ( 2013 ) . p.r
. struck and g. burkard , phys .
b * 82 * , 125401 ( 2010 ) .
s. amasha , k. maclean , i.p .
radu , d.m .
zumbhl , m.a .
kastner , m.p .
hanson , and a.c .
gossard , phys .
lett . * 100 * , 046803 ( 2008 ) .
m. xiao , m.g . house , and h.w .
jiang , phys .
lett . * 104 * , 096801 ( 2010 ) .
hachiya , g. burkard , and j.c .
egues , phys .
b * 89 * , 115427 ( 2014 ) .
f. bloch , phys .
* 105 * , 1206 ( 1957 ) .
redfield , ibm j. res .
* 1 * , 19 ( 1957 ) .
bulaev , b.trauzettel , and d. loss , phys .
b * 77 * , 235301 ( 2008 ) .
h. suzuura and t. ando , phys .
b * 65 * , 235412 ( 2002 ) .
f. guinea , m.i .
katsnelson , and a.k .
geim , nature phys .
* 6 * , 30 ( 2009 ) .
l. brey and h.a .
fertig , phys .
b * 73 * , 235411 ( 2006 ) .
svore , b.m .
terhal , and d.p .
divincenzo , phys .
a * 72 * , 022317 ( 2005 ) .
m. braun , p.r . struck , and g. burkard , phys .
b * 84 * , 115445 ( 2011 ) .
j. sawiska , i. zasada , and z. klusek , phys .
b * 81 * , 155433 ( 2010 ) .
f. guinea , new .
j. phys .
* 2 * , 083063 ( 2010 ) .
nair , m. sepioni , i .-
tsai , o. lehtinen , j. keinonen , a.v .
krasheninnikov , t. thomson , a.k .
geim , and i.v .
grigorieva , nature phys .
* 8 * , 199 ( 2012 ) .
mccreary , a.g .
swartz , w. han , j. fabian , and r.k .
kawakami , phys .
lett . * 109 * , 186604 ( 2012 ) .
h. zhang , c. lazo , s. blgel , s. heinze , and y. mokrousov , phys .
lett . * 108 * , 056802 ( 2012 ) .
son , m.l .
cohen , and s.g .
louie , nature * 444 * , 347 ( 2006 ) .
j. hong , e. bekyarova , p. liang , w.a .
de heer , r.c .
haddon , and s. khizroev , sci . rep .
* 2 * , 624 ( 2012 ) .
nair , i .-
tsai , m. sepioni , o. lehtinen , j. keinonen , a.v .
krasheninnikov , a.h .
castro neto , m.i .
katsnelson , a.k .
geim , and i.v .
grigorieva , nat .
commun . * 4 * , 2010 ( 2013 ) .
r. decker , j. brede , n. atodiresei , v. caciuc , s. blgel , and r. wiesendanger , phys .
b * 87 * , 041403(r ) ( 2013 ) .
lieb , phys .
lett . * 62 * , 1201 ( 1989 ) . j. fernndez - rossier and j.j .
palacios , phys .
lett . * 99 * , 177204 ( 2007 ) .
a.v . krasheninnikov and f. banhart , nat .
* 6 * , 723 ( 2007 ) .
robertson , b. montanari , k. he , c.s .
allen , y.a .
wu , n.m . harrison , a.i .
kirkland , and j.h .
warner , acs nano * 7 * , 4495 ( 2013 ) .
s. hemmatiyan , m. polini , a. abanov , a. h.macdonald , and j. sinova , phys .
b * 90 * , 035433 ( 2014 ) .
giesbers , k. uhlov , m. konen , e.c .
peters , m. burghard , j. aarts , and c.f.j .
flipse , phys .
lett . * 111 * , 166101 ( 2013 ) .
hmlinen , z. sun , m.p .
boneschanscher , a. uppstu , m. ijs , a. harju , d. vanmaekelbergh , and p. liljeroth , phys .
lett . * 107 * , 236803 ( 2011 ) . s .-
phark , j. borme , a.l .
vanegas , m. corbetta , d. sander , and j. kirschner , acs nano * 5 * , 8162 ( 2011 ) .
g. burkard , d. loss , and d.p .
divincenzo , phys .
b * 59 * , 2070 ( 1999 ) .
yazyev , nano lett .
* 8 * , 1011 ( 2008 ) .
f. hilser and g. burkard , phys .
b * 86 * , 125204 ( 2012 ) .
bassett , j.f .
heremans , d.j .
christle , c.g .
yale , g. burkard , b.b .
buckley , d.d .
awschalom , science * 345 * , 1333 ( 2014 ) .
overhauser , phys .
b * 92 * , 411 ( 1953 ) .
khaetskii , d. loss , and l. glazman , phys .
lett . * 88 * , 186802 ( 2002 ) .
j. fischer , b. trauzettel , and d. loss , phys .
b * 80 * , 155401 ( 2009 ) .
m. wojtaszek , i.j .
vera - marun , e. whiteway , m. hilke , and b.j .
van wees , phys .
b * 89 * , 035417 ( 2014 ) . h.o.h .
churchill , a.j .
bestwick , j.w .
harlow , f. kuemmeth , d. marcos , c.h .
stwertka , s.k .
watson , and c.m .
marcus , nature phys . * 5 * , 321 ( 2009 ) .
churchill , f. kuemmeth , j.w .
harlow , a.j .
bestwick , e.i .
rashba , k. flensberg , c.h .
stwertka , t. taychatanapat , s.k .
watson , and c.m .
marcus , phys .
lett . * 102 * , 166802 ( 2009 ) .
m. fuchs , v. rychkov , and b. trauzettel , phys .
b * 86 * , 085301 ( 2012 ) .
m. fuchs , j. schliemann , and b. trauzettel , phys .
b * 88 * , 245441 ( 2012 ) .
young , j.d .
sanchez - yamagishi , b. hunt , s.h .
choi , k. watanabe , t. taniguchi , r.c .
ashoori , and p. jarillo - herrero , nature * 505 * , 528 ( 2014 ) .
j. gttinger , t. frey , c. stampfer , t. ihn , and k. ensslin , phys .
lett . * 105 * , 116801 ( 2010 ) . c. volk ,
c. neumann , s. kazarski , s. fringes , s. engels , f. haupt , a. mller , and c. stampfer , nat .
commun . * 4 * , 1753 ( 2013 ) .
w. jolie , f. craes , m. petrovi , n. atodiresei , v. caciuc , s. blgel , m. kralj , t. michely , and c. busse , phys .
b * 89 * , 155435 ( 2014 ) .
m. ciorga , a.s .
schrajda , p. hawrylak , c. gould , p. zawadzki , s. jullian , y. feng , and z. wasilewski , phys .
b * 61 * , r16315(r ) ( 2000 ) .
j. cai , p. ruffieux , r. jaafar , m. bieri , t. braun , s. blankenburg , m. muoth , a.p .
seitsonen , m. saleh , x. feng , k. mllen , and r. fasel , nature * 466 * , 470 ( 2010 ) h. huang , d. wei , j. sun , s.l .
wong , y.p .
feng , and a.h .
castro neto , sci .
* 2 * , 983 ( 2012 ) .
a. morelos - gmez , s.m .
vega - daz , v.j .
gonzlez , f. tristn - lpez , r. cruz - silva , k. fujisawa , h. muramatsu , t. hayashi , x. mi , y. shi , h. sakamoto , f. khoerunnisa , k. kaneko , b.g .
sumpter , y.a .
kim , v. meunier , m. endo , e. muoz - sandoval , and m. terrones , acs nano * 6 * , 2261 ( 2012 ) .
k. kim , s. coh , c. kisielowski , m.f .
crommie , s.g .
louie , m.l .
cohen , and a. zettl , nat .
commun . * 4 * , 2723 ( 2013 ) . b. sommer , j. sonntag , a. ganczarczyk , d. braam , g. prinz , a. lorke , and m. geller , sci . rep . * 5 * , 7781 ( 2015 )
wang , e.j.g .
santos , b. jiang , e.d .
cubuk , c. ophus , a. centeno , a. pesquera , a. zurutuza , j. ciston , r. westervelt , and e. kaxiras , nano lett .
* 14 * , 450 ( 2014 ) .
|
thanks to its intrinsic ability to preserve spin coherence , graphene is a prime material for spintronics . in this review article , we summarize recent achievements related to spintronics in graphene quantum dots and motivate this field from a spintronics and a materials science point of view .
we focus on theory but also discuss recent experiments .
the main sources of spin decoherence are interactions with lattice excitations and the hyperfine interaction with present nuclear spins .
we explain effective spin - phonon coupling in detail and present a generic power law for the spin relaxation time @xmath0 as a function of the magnetic field . for specific cases ,
we discuss spin relaxation in detail .
the heisenberg exchange interaction is paramount for coherent spin qubit operation and addressed in the context of magnetism in graphene nanoflakes .
nuclear spins in the host and surrounding material can be considered by several means and the influence of @xmath1c nuclei has been studied in detail .
impressive advances in general spintronics and the fabrication of graphene devices are likely to spark significant advances in spintronics with graphene quantum dots in the near future .
|
let @xmath1 , @xmath2 or @xmath3 be a bounded connected open set whose boundary @xmath4 is lipschitz .
let @xmath5 be a ( small ) nonempty open subset , and assume that @xmath6 .
we use the notation @xmath7 , @xmath8 and we denote by @xmath9 the outward unit normal to @xmath10 at any point @xmath11 . bold letters and
symbols denote vector - valued functions and spaces ; for instance @xmath12 is the hilbert space of the functions @xmath13 with @xmath14 for all @xmath15 .
this work is related to the null controllability problem for the non - stationary stokes system @xmath16 which describes a viscous incompressible fluid flow in the bounded domain @xmath10 .
we use as a control function the density of external forces @xmath17 concentrated in the arbitrary subdomain @xmath18 during the time interval @xmath19 ; @xmath20 is the vector field of the fluid velocity , and @xmath21 is the scalar pressure .
the real @xmath22 denotes the constant viscosity of the fluid .
the symbol @xmath23 stands for the characteristic function of @xmath18 .
we introduce the following spaces @xmath24 then , for any @xmath25 , and @xmath26 , there exists exactly one solution @xmath27 of ( [ stokes ] ) with the following regularity : @xmath28;\hvec\right)\cap l^2\left(0 , t;\vvec\right ) , \,\,\pi \in l^2(0,t ; u)\ ] ] ( see @xcite ) . the null controllability problem for ( [ stokes ] ) at time
@xmath0 is the following : _ for any @xmath29 , find @xmath30 such that the corresponding solution to satisfies @xmath31 _ the controllability properties of evolution pdes have attracted a lot of works in the last decades : some relevant references are @xcite . in particular , the stokes - and more generally the navier - stokes - system has received a lot of attention : we mention the references @xcite .
the following result is proved in @xcite ( see also @xcite ) by the way of carleman estimates .
[ th-1 ] the linear system _ ( [ stokes ] ) _ is null - controllable at any time @xmath6 . on the other hand ,
the ( numerical ) approximation of controls either distributed or located on the boundary for the stokes system has received much less attention , due to the underlying ill - posedness of the approximation . in practice ,
such approximation is usually addressed in the framework of an optimal control reformulation .
precisely , one seeks to minimize the quadratic functional @xmath32 over the non - empty set @xmath33 following @xcite , duality arguments allow to replace this constrained minimization by the unconstrained minimization of its conjugate function @xmath34 defined as @xmath35 over @xmath36 , where @xmath37 solves the adjoint backward stokes system associated with ( [ stokes ] ) : @xmath38 @xmath39 is the hilbert space defined as the completion of any smooth space functions included in @xmath40 for the norm @xmath41 . the control of minimal square - integrable norm is then given by @xmath42 where @xmath43 is associated with the unique minimizer @xmath44 in @xmath39 of @xmath34 ( see @xcite ) .
the difficulty , when one wants to approximate such control in any finite dimensional space , that is when one likes to minimize numerically @xmath34 , is that the space @xmath39 is huge , in particular , contains @xmath45 , and even elements that may not be distributions .
numerical experiments do suggest that the minimizer @xmath44 is very singular ( we refer to @xcite for a detailed analysis in the close case of the heat equation ) .
this phenomenon is independent of the choice of @xmath46 , but is related to the use of dual variables .
actually , the equality ( [ null_condition ] ) can be viewed as an equality in a very small space ( due to the strong regularization effect of the heat kernel ) .
accordingly , the associated multiplier @xmath47 belong to a large dual space , much larger than @xmath12 , that is hard to represent ( numerically ) in any finite dimensional space .
an alternative way of looking at these problems and avoiding the introduction of dual variables has been proposed in @xcite .
it is based on the following simple strategy .
instead of working all the time with solutions of the underlying state equation , and looking for one that may comply with the final desired state , one considers a suitable class of functions complying with required initial , boundary , final conditions and appropriate regularity , and seeks one of those that is a solution of the state equation .
this is in practice accomplished by setting up an error functional defined for all feasible functions , and measuring how far those are from being a solution of the underlying state equation .
one advantage of this variational approach is that the way to get closer to a solution of the problem is by minimizing a functional that can not get stuck on local minima because the only critical points of the error turn out to be global minimizers with zero error .
therefore a general strategy for approximation consists in using a descent algorithm for this error functional .
this approach which has the flavor of a least - squares type method has been employed successfully in our null controllability context for the linear heat equation in @xcite and for an hyperbolic system in @xcite .
we apply this approach to the stokes system . in section [ sec2 ] , we describe the ingredients of the variational approach for the system ( [ stokes ] ) and reduce the search of one controlled trajectory for the stokes system to the minimization of the quadratic functional @xmath48 defined by ( [ fonce ] ) over the affine space @xmath49 defined by ( [ spacea ] ) . in section [ sec3 ] , by a general - purpose lemma ( lemma [ basic ] ) , using the very specific structure of the functional @xmath48 , we prove that we may construct minimizing sequences for the error functional @xmath48 that do converge strongly to an extremal point for @xmath48 ( see proposition [ convergence ] ) .
section [ sec4 ] , we adapt the argument for the direct problem of the steady navier - stokes system .
results of this work were partially announced in the note @xcite .
following @xcite , we define the non - empty space @xmath50 these hypotheses on @xmath20 imply that it belongs to @xmath51 , \lvec^2(\omega))$ ] and give a meaning to the equalities @xmath52 , @xmath53 in @xmath12 .
note also that @xmath49 is defined in agreement with the regularity of any solution @xmath27 of the stokes system with a source term @xmath54 .
then , we define the functional @xmath55 by @xmath56 where the corrector @xmath57 is the unique solution in @xmath58 of the ( elliptic ) boundary value problem @xmath59 for any @xmath60 , the term @xmath61 belongs to @xmath62 so that the functional @xmath48 is well - defined in @xmath49 . the approach developed here is based on the following result .
[ prop1 ] @xmath27 is a controlled solution of the stokes system ( [ stokes ] ) by the control function @xmath63 if and only if @xmath64 is a solution of the extremal problem : @xmath65 proof- from the controllability of the stokes system given by theorem [ th-1 ] , the extremal problem ( [ extremal_pb ] ) is well - posed in the sense that the infimum , equal to zero , is reached by any controlled solution of the stokes system .
note that , without additional assumptions , the minimizer is not unique .
conversely , we check that any minimizer for @xmath48 is a solution of the ( controlled ) stokes system : let @xmath66 be arbitrary where @xmath67 the first variation of @xmath48 at the point @xmath64 in the admissible direction @xmath68 defined by @xmath69 exists , and is given by @xmath70 where the corrector @xmath71 , associated with @xmath68 , is the unique solution of @xmath72 multiplying the main equation of this system by @xmath57 ( recall that @xmath57 is the corrector associated with the minimizer @xmath73 ) , integrating by parts , and using the boundary conditions on @xmath57 and @xmath74 , we get @xmath75 where we have used that @xmath76_0^t\,d\xvec = \int\!\!\!\!\int_{q_t } \yvec\cdot\vvec_t\,d\xvec\,dt,\ ] ] and that @xmath77 therefore if @xmath64 minimizes @xmath48 , the equality @xmath78 for all @xmath66 implies that the corrector @xmath79 solution of ( [ corrector_unsteady ] ) satisfies the conditions @xmath80 but from the unique continuation property for the stokes system ( see @xcite ) , it turns out that @xmath81 in @xmath82 and that @xmath83 is a constant in @xmath82 . eventually , the relation ( [ eprime ] ) is then reduced to @xmath84 and then implies that this constant is zero . consequently ,
if @xmath60 is a minimizer for @xmath48 , then @xmath85 in @xmath82 , and the corrector @xmath57 is zero in @xmath82 , so that @xmath86 .
therefore , @xmath64 solves ( [ stokes ] ) , and since @xmath60 , the state @xmath20 is controlled at the time @xmath0 by the function @xmath87 which acts as a control distributed in @xmath18 .
0.1 cm the proof of proposition [ prop1 ] only utilizes optimality of @xmath64 and not its minimality .
therefore in the statement of the proposition , we could have written instead : @xmath27 is a controlled solution of the stokes system ( [ stokes ] ) by the control function @xmath63 if and only if @xmath64 is a stationary point for the functional @xmath88 over @xmath60 .
this is relevant from the perspective of the numerical simulation for it guarantees that the numerical procedure based on a descent strategy can not get stuck in local minima .
0.1 cm for any @xmath89 , the cost @xmath48 can be formulate as follows @xmath90 this justifies the least - squares terminology .
the use of least - squares type approaches to solve linear and nonlinear problem is not new and we refer to @xcite for many applications .
the use of least - squares type approaches in the controllability context comes from @xcite .
0.1 cm [ divepsilon ] the quasi - incompressibility case is obtained in the same way .
it suffices to add @xmath91 ( for any @xmath92 ) to the divergence term in the functional @xmath48 .
this is also in practice a classical trick to fix the constant of the pressure @xmath21 ( see section [ sec4 ] ) .
0.1 cm [ compact_time_support ] the approach allows to consider compact support control jointly in time and space .
it suffices to replace the function @xmath93 in ( [ stokes ] ) by any compact support function in time and space , say @xmath94 , where @xmath95 denotes a non - empty subset of @xmath82 .
since theorem [ th-1 ] holds for any controllability time @xmath0 and any subset @xmath18 of @xmath10 , the controllability of ( [ stokes ] ) remains true as soon as @xmath95 contains any non - empty cylindrical domain of the form @xmath96 .
0.1 cm [ directproblem ] _ a fortiori _ , the approach is well - adapted to address the direct problem ( which consists , @xmath87 being fixed , in solving the boundary value problem ( [ stokes ] ) : it suffices to remove from @xmath49 and @xmath97 the condition ( [ null_condition ] ) , and fix the forcing term @xmath87 ( see @xcite using a similar least - squares type point of view ) . in that case , the second line of ( [ corrector_unsteady_control ] ) is replaced by @xmath98 , which implies with the first line , that @xmath57 and @xmath99 both vanish in @xmath82 .
it is worth to notice that this approach allows to treat at the same time the null controllability constraint and the incompressibility one . in this sense
, the pair @xmath100 can be regarded as a control function for the set of constraints @xmath101 these two conditions are compatibles : there is no competition between them . in the uncontrolled situation , from the uniqueness , the pressure @xmath21 is unique as soon as the source term ( here @xmath102 ) is fixed .
on the other hand , in our controllability context , the pair @xmath103 is not unique : the pressure @xmath21 depends on @xmath104 and _ vice versa_. therefore , the optimization with respect to both variables at the same time makes sense . from this point of view
, we may reformulate the problem as a general controllability problem for the heat equation : @xmath105 @xmath106 being a control function such that ( [ eq : contraint_gen ] ) holds .
the control function @xmath106 acts on the whole domain , but on the other hand , should take the specific form @xmath107 . again , this perspective is different with the classical one , which consists in finding a control @xmath108 , such that @xmath109 in @xmath10 where @xmath27 solves ( [ stokes ] ) .
this can done by duality , penalization technique , etc .
conversely , one may also consider iteratively first the null controllability constraint , that is , for any @xmath21 fixed in @xmath110 , find a control @xmath111 such that ( [ null_condition ] ) holds , and then find the pressure @xmath21 such that @xmath112 holds in @xmath82 . using again a least - squares type approach ( for the heat equation , as developed in @xcite ) , the first step reduces to solve , for any @xmath113 fixed , the problem @xmath114 where @xmath115 solves ( [ corrector_unsteady ] ) and @xmath116 is given by @xmath117 the second step consists in updating the pressure according to a descent direction for the function @xmath118 defined by @xmath119 .
we get that the first variation of @xmath120 at @xmath21 in the direction @xmath121 is given by @xmath122 where @xmath123 solves @xmath124 again , this direct problem may be solved within the variational approach developed in this work ( see remark [ directproblem ] ) .
proposition [ prop1 ] reduces the approximation of a null control for ( [ stokes ] ) to a minimization of the functional @xmath48 over @xmath49 . as a preliminary step ,
since @xmath49 is not a vectorial space , we remark that any element of @xmath49 can be written as the sum of one element of @xmath49 , say @xmath125 , plus any element of the vectorial space @xmath97 .
thus , we consider for any @xmath126 the following problem : @xmath127 problems ( [ extremal_pb ] ) and ( [ extremal_pb_bis ] ) are equivalent .
any solution of problem ( [ extremal_pb_bis ] ) is a solution of the initial problem ( [ extremal_pb ] ) .
conversely , any solution of problem ( [ extremal_pb ] ) can be decomposed as the sum @xmath128 , for some @xmath129 in @xmath97 .
we endow the vectorial space @xmath97 with its natural norm @xmath130 such that : @xmath131 recalling that @xmath132 where @xmath133 solves @xmath134 in @xmath10 .
we denote @xmath135 the corresponding scalar product .
@xmath136 is an hilbert space .
the relation ( [ eprime ] ) allows to define a minimizing sequence in @xmath97 for @xmath137 .
it turns out that minimizing sequences for @xmath138 which belong to a precise subset of @xmath97 remain bounded uniformly .
this very valuable property is not _ a priori _ guaranteed from the definition of @xmath138 .
the boundedness of @xmath138 implies only the boundedness of the corrector @xmath57 for the @xmath139-norm and the boundedness of the divergence @xmath83 of the velocity field for the @xmath140-norm .
actually , this property is due to the fact that the functional @xmath138 is invariant in the subset of @xmath97 which satisfies the state equations of ( [ stokes ] ) . in order to construct a minimizing sequence bounded in @xmath97 for @xmath138 ,
we introduce the linear continuous operator @xmath141 which maps a triplet @xmath142 into the corresponding vector @xmath143 , with the corrector @xmath57 as defined by ( [ corrector_unsteady ] ) .
then we define the space @xmath144 composed of the elements @xmath64 satisfying the stokes system and such that @xmath20 vanishes on the boundary @xmath145 . note that @xmath146 is not the trivial space : it suffices to consider the difference of two distinct null controlled solutions of ( [ stokes ] ) . finally , we note @xmath147 the orthogonal complement of @xmath146 in @xmath97 and @xmath148 the ( orthogonal ) projection on @xmath149 .
we then define the following minimizing sequence @xmath150 as follows : @xmath151 where @xmath152 is defined as the unique solution of the formulation @xmath153 @xmath154 denotes a positive descent step .
in particular , ( [ fvdes ] ) implies that @xmath155 and @xmath156 ( actually in @xmath157 ) .
one main issue of our variational approach is to establish the convergence of the minimizing sequence defined by ( [ minimizing_sequence ] ) .
we have the following result .
0.1 cm [ convergence ] for any @xmath158 and any @xmath159 , the sequence @xmath160 converges strongly to a solution of the extremal problem ( [ extremal_pb ] ) .
0.1 cm this proposition is the consequence of the following abstract result which can be adapted to many different situations where this variational perspective can be of help .
0.1 cm [ basic ] suppose @xmath161 is a linear , continuous operator between hilbert spaces , and @xmath162 , a closed subspace , @xmath163 .
put @xmath164 1 .
@xmath165 is quadratic , non - negative , and strictly convex , where @xmath166 is the orthogonal complement of @xmath146 in @xmath167 .
if we regard @xmath48 as a functional defined on @xmath167 , @xmath168 , and identify @xmath167 with its dual , then the derivative @xmath169 always belongs to @xmath166 . in particular , a typical steepest descent procedure for @xmath168 will always stay in the manifold @xmath170 .
3 . if , in addition , @xmath171 , then the steepest descent scheme will always produce sequences converging ( strongly in @xmath172 ) to a unique ( in @xmath170 ) minimizer @xmath173 with zero error
. proof of lemma [ basic]- suppose there are @xmath174 , @xmath175 , such that @xmath176 due to the strict convexity of the norm in a hilbert space , we deduce that this equality can only occur if @xmath177 .
so therefore @xmath178 , and @xmath179 .
for the second part , note that for arbitrary @xmath180 , @xmath181 , and so @xmath182 therefore the derivative @xmath183 , the steepest descent direction for @xmath48 at @xmath184 , has to be orthogonal to all such @xmath180 .
finally , assume @xmath185 .
it is clear that this minimizer is unique in @xmath170 ( recall the strict convexity in ( i ) ) .
this , in particular , implies that for arbitrary @xmath186 , @xmath187 because this inner product is the derivative of the section @xmath188 at @xmath189 , and this section must be a positive parabola with the minimum point at @xmath190 .
if we consider the gradient flow @xmath191 then , because of ( [ desigualdad ] ) , @xmath192 this implies that sequences produced through a steepest descent method will be minimizing for @xmath48 , uniformly bounded in @xmath172 ( because @xmath193 is a non - increasing function of @xmath194 ) , and due to the strict convexity of @xmath48 restricted to @xmath170 , they will have to converge towards the unique minimizer @xmath173 . 0.1 cm despite the strong convergence in this statement
, it may not be true that the error is coercive , even restricted to @xmath170 , so that strong convergence could be very slow . because of this same reason , it may be impossible to establish rates of convergence for these minimizing sequences .
0.1 cm the element @xmath195 determines the non - homogeneous data set of each problem : source term , boundary conditions , initial and/or final condition , etc .
the subspace @xmath167 is the subset of the ambient hilbert space @xmath172 for which the data set vanishes .
@xmath141 is the operator defining the corrector , so that @xmath196 is the subspace of all solutions of the underlying equation or system .
the subspace @xmath146 is the subspace of all solutions of the problem with vanishing data set .
in some situations @xmath146 will be trivial , but in some others will not be so .
the important property is ( iii ) in the statement guaranteeing that we indeed have strong convergence in @xmath172 of iterates .
the main requirement for this to hold is to know , _ a priori _ , that the error attains its minimum value zero somewhere , which in the situation treated here is guaranteed by theorem [ th-1 ]
. 0.1 cm proof of proposition [ convergence]- the result is obtained by applying lemma [ basic ] as follows .
if we put @xmath197 @xmath172 is taken to be @xmath198 .
@xmath167 is taken to be @xmath97 as given in ( [ spacea0 ] ) and @xmath199 .
the operator @xmath141 maps a triplet @xmath200 into @xmath201 as explained earlier .
the construction of the minimizing sequence only requires the resolution of standard well - posed elliptic problems over @xmath82 , well - adapted to general situations ( time dependent support , mesh adaptation , etc ) . on the other hand , it is important to highlight that the @xmath202 control function @xmath87 obtained from the minimizing procedure does not _ a priori _ minimize any specific norm ( for instance the @xmath203-norm ) . without the projection on @xmath204 in ( [ minimizing_sequence ] ) , the sequence @xmath205 remains a minimizing sequence for @xmath206 : actually , the values of the cost @xmath138 along the sequence @xmath205 are equal with or without the projection .
this is due to the fact that the component of the descent direction @xmath207 on @xmath208 does not affect the value of the cost : on the other hand , without the projection , the minimizing sequence may not be bounded uniformly in @xmath97 , in particular the control function @xmath87 may not be bounded in @xmath202 .
the subset @xmath166 is not explicit , so that in practice the projection @xmath209 may be defined by @xmath210 , where @xmath211 solves the extremal problem : @xmath212 recalling that @xmath146 is by definition the set of triplets @xmath64 satisfying @xmath213 , @xmath99 in @xmath82 such that @xmath20 vanishes on @xmath145 , this extremal problem is nothing else than a controllability problem for the stokes system , similar to the one considered in this work .
therefore , we shall bypass this projection and shall introduce instead a stopping criteria for the descent method measuring how far from @xmath149 the descent direction is .
we refer to @xcite for numerical experiments .
the least - squares approach allows to address non - linear problem .
we consider here simply the steady navier - stokes and address the direct problem : find @xmath27 solution of @xmath214 we recall the following result ( see @xcite ) : in order to solve this boundary value problem , we use a least - squares type approach .
we consider the space @xmath218 and then we define the functional @xmath55 by @xmath219 where the corrector @xmath57 is the unique solution in @xmath220 of the ( elliptic ) boundary value problem @xmath221 from theorem [ th - temam ] , the infimum is equal to zero and is reached by an element solution of ( [ navierstokes ] ) .
conversely , we would like to state that the only critical point for @xmath48 correspond to solution of ( [ navierstokes ] ) .
we then check that we can take @xmath235 , i.e. @xmath236 uniquely given by the solution of ( [ corrector_unsteadyqua ] ) remains bounded with respect to @xmath237 . by definition , the corrector @xmath57 solves the variational formulation : @xmath238 taking @xmath239 , we get that @xmath240 so that , in view of the poincar inequality , @xmath241 leading to the result
. then , taking @xmath235 in ( [ eprime ] ) , we get @xmath242 ( * ? ? ?
* corollary 1.8 ) let @xmath167 be a hilbert space .
suppose @xmath265 is an error functional for which there is a unique @xmath266 with @xmath267 .
if @xmath268 , then @xmath269 strongly in @xmath167 , and @xmath46 complies with the palais - smale condition .
[ strongconvergence ] if @xmath270 is small enough , then any minimizing sequence @xmath271 for @xmath48 converges strongly toward @xmath27 , unique solution of the boundary value problem ( [ navierstokes ] ) .
multiplying the state equation by @xmath57 and using ( [ lemma1 ] ) , we get @xmath233 so that , at the optimality , the corrector funcion @xmath57 solves the following linear boundary problem : @xmath275 proposition [ functionalerror ] implies that if @xmath276 , then @xmath277 .
therefore , the formulation ( [ navierstokes_v ] ) implies that @xmath81 and @xmath99 . this may be seen as a unique continuation property for navier - stokes equation .
as a very interesting consequence , proposition [ strongconvergence ] reduces the approximation of the vanishing elements of @xmath48 to the construction of one arbitrary minimizing sequence , say @xmath271 , for @xmath48 .
e. fernndez - cara and a. mnch , _ numerical null controllability of semi - linear 1d heat equations : fixed points , least squares and newton methods , _ mathematical control and related fields , vol .
2 , no . 3 ( 2012 ) . 217 - 246 .
r. glowinski , j.l .
lions and j. he , _ exact and approximate controllability for distributed parameter systems : a numerical approach _ encyclopedia of mathematics and its applications , 117 .
cambridge university press , cambridge , 2008 .
imanuvilov , j .-
puel , m. yamamoto , _ carleman estimates for parabolic equations with nonhomogeneous boundary conditions _ , chin .
30b(4 ) , 2009 , 333 - 378 . o.yu .
imanuvilov , _
remarks on exact controllability for the navier - stokes equations _ , esaim control optim .
6 ( 2001 ) , 39 - 72 .
i. lasiecka and r.triggiani , _ control theory for partial differential equations : continuous and approximation theories . i. abstract parabolic systems . _
encyclopedia of mathematics and its applications , 74 .
cambridge university press , cambridge , 2000 .
|
we discuss the approximation of distributed null controls for partial differential equations .
the main purpose is to determine an approximation of controls that drives the solution from a prescribed initial state at the initial time to the zero target at a prescribed final time . as a non trivial example
, we mainly focus on the stokes system for which the existence of square - integrable controls have been obtained in _ [ fursikov & imanuvilov , controllability of evolution equations , 1996 ] ) _ via carleman type estimates .
we introduce a least - squares formulation of the controllability problem , and we show that it allows the construction of strong convergent sequences of functions toward null controls for the stokes system .
the approach consists first in introducing a class of functions satisfying _ a priori _ the boundary conditions in space and time - in particular the null controllability condition at time @xmath0 - , and then finding among this class one element satisfying the system .
this second step is done by minimizing a quadratic functional , among the admissible corrector functions of the stokes system .
we also discuss briefly the direct problem for the steady navier - stokes system .
the method does not make use of any duality arguments and therefore avoid the ill - posedness of dual methods , when parabolic type equation are considered .
|
since the discovery of @xmath4 lyman break galaxies ( lbgs @xcite ) , many efforts are being made to select the whole population of star - forming galaxies at the highest possible redshifts ( see @xcite for a review ) , and to constrain the star formation history of the universe ( e.g. @xcite ) .
the detection of @xmath5 galaxies allows to study the modes of early galaxy formation and the interplay between the first galaxies and the intergalactic medium ( igm ) .
the redshift range @xmath6 is a very intriguing time during cosmic evolution , when hydrogen reionization is believed to be basically complete and the igm starts to be polluted with metals @xcite . because a significant part of the bolometric luminosity of primeval star - forming objects escapes as ly@xmath0 emission , especially if dust is not yet ubiquitous , it seems profitable to search for the redshifted ly@xmath0 line of high redshift galaxies .
the advent of ten meter class telescopes made this approach successful and efficient , and both narrow - band imaging and optical _ serendipitous _ spectroscopy found ly@xmath0 emitters at @xmath7 @xcite . thanks to the recently increased sensitivity of ccds at @xmath8 m , the discovery of @xmath9 galaxies has also become possible . using narrow band filters sensitive to wavelengths in the range @xmath10 , i.e. ly@xmath0 emission at @xmath11 and subsequent spectroscopy ,
two groups have detected the first three galaxies known at @xmath5 .
the first of these was found due to the strong lensing amplification ( 4.5 ) by the cluster abell 370 @xcite .
the other two galaxies were found in a survey made with subaru suprime - cam instrument which provides a huge field of view ( 814 arcmin@xmath2 @xcite ) .
we have used a new approach to search for ly@xmath0 emitters at @xmath12 based on slitless spectroscopy in combination with a narrow band ( 2.2% ) filter and very deep imaging in a field with very low galactic extinction ( @xmath13=0.002 ) .
the @xmath14 filter employed has a central wavelength of 9135 and fwhm of 200 and is therefore sensitive to ly@xmath0 at @xmath15 .
the imaging was carried out through the @xmath14 filter ( to identify the sources of emission lines detected in the spectroscopy ) and the bessel @xmath16 filter ( to determine the @xmath17 colour or spectral break over the ly@xmath0 line ) .
the observations were carried out with the fors2 instrument at the vlt during eight nights in october , november 2002 and february 2003 .
total exposure times were 7.6 , 6.33 and 4.0 hours for the slitless spectroscopy , @xmath14 and @xmath16 band direct imaging , respectively .
observing conditions were excellent resulting in 0.7 arcsec seeing on the resulting combined images .
the field was also imaged earlier in bessel @xmath18 and gunn @xmath19 band .
three authors ( jk , ac , ssa ) have independently searched for emission lines on the spectroscopic frame , resulting in a combined list of 161 emission lines .
each emission line can originate from a counterpart on the direct image within a region of about 50@xmath20 .
the emission lines are related to 345 possible counterparts in the @xmath14 image .
four of these have @xmath21 , one of which is an [ oiii ] emitter at @xmath22 .
this leaves three counterparts which may be ly@xmath0 emitters at @xmath23 .
furthermore , among the @xmath24 objects detected by sextractor on the @xmath14 image , there are ten other objects with these colours , which are considered lbg candidates at @xmath5 .
the number of possible ly@xmath0 emitters at @xmath3 detected in our field is consistent with the number densities of @xmath25 galaxies reported by other authors @xcite .
in particular , the recent subaru survey by @xcite resulted in the detection of 73 ly@xmath0 candidates at @xmath3 ( with @xmath26 ) in an 814 arcmin@xmath2 field .
based on this result , we expect to find in our available 43 arcmin@xmath2 field three candidate ly@xmath0 emitters at @xmath3 .
slitmask spectroscopic observations are needed to confirm the identity of the candidate ly@xmath0 emitters and the lbgs .
these observations will provide higher signal to noise and higher spectral resolution to measure the characteristic asymmetric line profile of ly@xmath0 at high redshift .
the larger wavelength coverage will allow the detection of emission lines from the presumed lbgs outside the @xmath14 filter .
time has been allocated to carry out these observations with fors2 .
steidel c. et al .
1996 , apj , 462 , l17 stern d. & spinrad h. 1999 , pasp , 111 , 1475 madau p.
1996 , mnras , 283 , 1388 gnedin n.y .
2000 , apj , 535 , 530 ciardi b. et al .
2003 , mnras , 344 , l7 hu e.m .
2000 , in `` the hy - redshift universe : galaxy formation and evolution at high redshift '' , eds .
bunker and w.j.m .
van breugel , a.s.p .
series , astro - ph/9911477 malhotra s. et al .
2001 , apj , 561 , 766 hu e.m .
2002 , apj , 568 , l75 kodaira k. et al .
2003 , pasj , 55 , l17
|
we are carrying out a search for ly@xmath0 emitting galaxies at @xmath1 employing slitless spectroscopy at the vlt . in our 43 arcmin@xmath2 field
we find three isolated single emission lines with seds consistent with line emitting galaxies at @xmath3 .
|
recent years , an increasing number of geosynchronous satellites are being planned to cope with the need of providing regional navigation services or augmentation to global systems like the gps . from various announcements made by the indian space research organization ( isro )
, one can expect at least 9 geosynchronous satellites to be commissioned by isro within the next few years with dual frequency synchronized payloads @xcite@xcite . among these
, the first satellite , gsat-8 , has been launched recently and is likely to commence regular broadcast of waas messages from september 2011 , as part of the gagan project @xcite .
gagan aims to enhance the precision achievable by gps - based systems for assisting air - craft landing .
a second geosynchronous satellite with a gagan payload has been announced for launch during early 2012 .
both these broadcast waas messages on the l1 ( 1575 mhz ) and l5 ( 1176 mhz ) carriers .
these satellites will soon be followed by a series of geosynchronous satellites with dual frequency ( l and s - band ) navigation payloads as part of the indian regional navigation satellite system ( irnss)@xcite . all these , and perhaps some satellites for digital audio broadcasting , have a unique advantage of continuous visibility from the indian subcontinent .
( e.g. , nine or more l - band signals from geosynchronous satellites are visible all the time ) .
this brings an interesting combination of benefits to radio astronomy and satellite orbit estimate requirements . here
, passive interferometry is used between signals received at different locations within the space covered by a synthesis radio telescope like the giant meterwave radio telescope ( gmrt ) .
we refer the reader to @xcite for an example of a satellite observation with such a co - located facility .
a simple illustration of the geometry of interferometry is given in fig .
[ fig : satint - geometry ] . in this figure ,
offset of the state vector with respect to a reference position is represented by @xmath0 , where , @xmath1 is a reference position * * , * * and**@xmath2 * * is the true position of the satellite .
similarly , satellite velocity is then given by @xmath3 .
a cross - correlation of the signals received at two stations in two different frequencies can be used to measure instantaneous values of the components of @xmath4 and @xmath5 along the baseline vector joining the two receiving stations .
initial value for the reference position is generally available ( or obtainable as two - line elements from the internet ) and can be updated using the interferometric measurements using a suitable orbit propagation software . in view of the wide variety of inexpensive antennas and receivers available off - the - shelf for receiving satellite signals , it is possible to construct a simple system for satellite interferometry ( and co - locating with a radio astronomy facility ) to obtain valuable data for radio astronomical imaging , ionospheric studies and satellite orbit estimation . a preliminary attempt for realizing such a co - located system with the gmrt was attempted by us in 2005 . the subsystems developed for this project are being augmented to establish a simple facility at the raman research institute . a conceptual description of the system constituting this facility has been given in fig .
[ fig : array - element - for ] .
this facility consists of a common l - band subsystem interfaced to off - the - shelf rf units corresponding to the satellite c , l or ku - band .
it may be noted that the primary navigation signals are in the l - band , and provide the most valuable information on the satellite range / doppler or ionospheric delay along the line of sight .
while this is the most useful band for measuring ionospheric contribution to interferometric phases for observing a celestial radio source , the satellite orbit estimation problem itself requires supplementary data using interferometric observations which are sensitive to satellite state vector components along directions perpendicular to the line of sight . in this connection
, we note that the information related to components orthogonal to the line of sight is a natural outcome of radio interferometry .
furthermore , since radio interferometry provides a measure of the arrival time differences of signals reaching different array elements , it does not require any knowledge of the nature of signals broadcast by the satellite .
hence , any band - limited signal transmitted by the satellite can be used for this purpose .
a cross - correlation of the signals - ( after compensating for delay differences resulting from an assumed reference ) , provides an estimate of the arrival time differences in excess of those that can be traced to the reference . in view of the inherent signal strength coming from satellites ,
the required bandwidths and integration times are well within simple processing capabilities of a normal workstation
. however , the angular resolution which can be achieved by this method ( which improves with decreasing wavelength ) depends on the array extent measured in units of wavelength .
we propose to exploit this fact , and the decreasing importance of ionosphere at higher frequencies , to use the available high frequency signals from the satellite for interferometry .
interestingly , all the geosynchronous satellites operated by isro ( including those with gagan or irnss payloads ) will have their telemetry signals operating in the c - band . on the other hand
, the recently launched satellite ( gsat-8 ) with a gagan payload has a rich set of communication transponders in the ku - band , offering a further advantage for angular resolution .
there are off - the - shelf ku - band low - noise - block converters ( lnbc ) which can take an external reference , and these can be used to provide coherent translation to l - band for signals from different elements of the array , by distributing a common local oscillator reference . based on these considerations
, we have included c and/or ku - band rf subsystems to co - exist with a simple l - band antenna ( and lna ) for our array element , as illustrated in fig .
[ fig : array - element - for ] .
as an illustration , we present in this paper some results obtained in an experiment in satellite interferometry by observing the l - band signals from a worldspace satellite ( which had a coverage in the indian subcontinent during the observations ) using commercial yagi antennas located at two different gmrt sites . in this experiment , we exploited the spare capacity of the gmrt fibre - optic network to obtain a long baseline l - band interferometer .
the specific sites chosen for this experiment were those in the original plan of the gmrt ( and hence with proper termination of fibre - optic cables ) but where no gmrt antenna were installed due to a small reduction of the gmrt antennas arising from funding constraints .
these two locations were used to position a pair of low - cost antennas ( meant for receiving digital audio signals from worldspace ) effectively separated by about 11.2 kilometres .
the l -band signals received by these low - cost antennas were brought to a central location using the gmrt fibre - optic network .
they were then down - converted to 70 mhz if , using a common local oscillator , digitized and recorded using a pci data acquisition card .
the recorded data were then processed by a simple software to cross - correlate signals from the two antennas after suitable delay compensation .
the slow relative motion of the satellite with respect to the earth results in fringes visible in the estimated cross - correlation , as shown in fig .
[ fig : plot - showing - the ] .
the figure shows two sets of fringes from a one - hour stretch of data , with a one - hour gap between the two sets .
these are typical interferometric fringes , in which a linear variation of path length difference is translated into a sinusoidal variation of the cross - correlation . by counting cycles and measuring the residual phase offset , one can infer that the time taken for a path length variation of 25 cycles is about 7220 seconds , implying that the mean time for path length to vary by one wavelength ( 20.17 cm ) was 7220/25 = 288.8 seconds .
the signal - to - noise ratio was adequate for achieving an accuracy of a fraction of degree in phase estimation , with a one - second integration .
thus , one can conveniently detect a tiny relative displacement ( by a fraction of a millimetre ) of the satellite along a direction parallel to the line joining the two antennas .
further experiments using the l - band navigation ( waas ) signals and the ku - band signals from gsat-8 are planned soon after the satellite payloads are officially commissioned for regular services after the in - orbit - tests .
the results will be presented elsewhere .
plot showing the fringes in two independent 1 hour stretches , separated by an hour.,width=340 ]
an important conclusion drawn from the above experiments was that the achievable accuracies for satellite interferometry provide valuable data for satellite orbit estimation even with much slower phase variations , as would result from a short ( in - campus ) baseline . with this in mind
, the facility being established by us is planned to be realized in a relatively short baseline , within the campus of the raman research institute .
most of the results reported in this paper were obtained from a project sponsored by isro under the respond programme with generous support from insat master control facility ( mcf ) .
we thank the director , ncra for giving us permission to use the gmrt infrastructure for our experiments in satellite interferometry .
the entire effort has been inter - institutional and could not have been realized without the whole - hearted support from many colleagues from the raman research institute as well as from the gmrt observatory .
|
in recent years , a large number of geosynchronous satellites are being planned to provide augmentation services for enhancing the precision to global positioning systems , e.g. , gps , in applications such as aircraft landing . in this paper , we present a scheme for co - locating passive satellite observational facilities with a radio astronomy facility to open a new possibility of providing valuable data for radio astronomical imaging , ionospheric studies and satellite orbit estimation .
interferometry , navigation satellites , radio astronomy
|
the fact that gauge coupling unification is a `` near - miss '' within the standard model ( sm ) provides an indication in favor of the idea of unification @xcite .
likewise , the existence of neutrino masses , required to account for neutrino oscillation data @xcite , also provides another motivation towards unified or gut ( grand unified theory)-like extensions of the sm .
however , the most characteristic feature of gut - type unification , namely matter instability , has so far defied experimental confirmation @xcite . on the other hand , neither the generation of neutrino masses nor the tilting in the evolution of the gauge couplings require unification in the conventional sense .
for instance , it is well known that the gauge couplings merge in the minimal supersymmetric extension the sm , provided that supersymmetric states lie around the tev scale @xcite .
so far , though , there has been no trace of such states in the lhc data @xcite .
here we consider an alternative approach in which new physics at the tev scale realizes an extended electroweak gauge structure with perturbatively conserved baryon - number . for definiteness
we consider the @xcite ( 3 - 3 - 1 ) framework , which implies that the number of generations equals the number of colors , in order to cancel anomalies .
this scheme has attracted attention recently also in connection with b physics @xcite , or flavor symmetries @xcite . in this letter
we present a model in which the gauge couplings can naturally unify at some accessible energy , and where small calculable neutrino masses are induced by new gauge bosons exchange , in the absence of supersymmetry .
neutrino masses arise at the tev scale @xcite instead of the conventional high - scale seesaw mechanism @xcite .
we first recall that , by adding three gauge singlet fermions @xmath0 , the light neutrinos acquire mass only at one - loop order @xcite .
unfortunately , however , unification does not occur , as can be seen in fig .
[ fig : unificationplots0 ] .
this is mostly due to fact that the new gauge bosons make @xmath1 weaker at high energies , while the new colored particles strengthen @xmath2 .
hence we contemplate the possibility of unifying the gauge couplings in such a scheme by promoting the three fermion singlets to three octets of the enlarged electroweak symmetry .
the new variant not only opens the possibility of reconciling neutrino mass generation with gauge coupling unification but also provides a novel radiative seesaw mechanism .
the gauge group is broken down to the standard model at some scale @xmath3 characterizing the new gauge boson masses .
this scale is found to lie in the @xmath4 range , with a plethora of new states expected to be directly accessible to lhc searches .
+ scale is set to @xmath5 tev . ]
we consider a simple variant of the model introduced in @xcite , where the fermion singlet is now promoted to an octet representation of @xmath6 . the model is based on the same gauge symmetry , extended with a global @xmath7 , which is necessary in order to consistently define lepton number , and an auxiliary parity symmetry whose purpose will be made clear below .
.[tab : field_content]field content of the model . [ cols="^,^,^,^,^,^,^,^,^,^,^,^ " , ] the model contains three generations of lepton @xmath6 anti - triplets , two generations of quark triplets and one of anti - triplets ( quarks and charged leptons are accompanied with their right - handed @xmath6 singlet partners ) , three generations of fermion octets , and finally three scalar boson anti - triplets .
we summarize the particle content of the model in table [ tab : field_content ] . the allowed lepton interactions compatible with the quantum number assignments given in table [ tab : field_content ] are the following : @xmath8 the components of @xmath9 and the @xmath10 are written as : @xmath11 the scalars take vacuum expectation values ( _ vevs _ ) in the directions @xmath12 and @xmath13 . as for the octets ,
one can write @xmath14 such that @xmath15 is transformed into @xmath16 under an @xmath17 gauge transformation , where @xmath18 is the transformation matrix of the triplet representation .
+ under , each @xmath15 breaks into the representations @xmath19 , @xmath20 , @xmath21 , @xmath22 , so there are four new charged leptons ( @xmath23 , @xmath24 , @xmath25 , @xmath26 ) and four new neutral fermion states ( @xmath27 , @xmath28 , @xmath29 , @xmath25 ) in each generation .
+ note that the @xmath7 charge assignment of @xmath30 and @xmath31 differs from the one of the singlet of ref .
@xcite so as to allow for a ( vector - like ) octet mass term , @xmath32 , in eq .
[ eq : lag ] . on the other hand
, the @xmath33 symmetry forbids a @xmath34 coupling , which leads to the existence of one massless neutrino state in each generation , at the tree level .
tev mass . ]
note also that the electric charge and lepton number assignments of the particles of the model follow from is a gauge generator , there is no physical goldstone boson associated with spontaneous lepton number violation @xcite . ]
@xmath35 where @xmath36 and @xmath37 are the diagonal generators of @xmath6 .
the one - loop renormalization group equation of the @xmath38 is given by @xcite : @xmath39 where @xmath40 is the logarithm of the energy scale , and the @xmath41 coefficients are functions of the casimir of the gauge group , @xmath42 , and of the dynkin index of the scalar and ( weyl ) fermion representations , @xmath43 and @xmath44 , respectively : @xmath45 values of the gauge coupling constants near unification , assuming that the 3 - 3 - 1 scale is 1 tev .
the thickness of the @xmath46 line reflects the @xmath47 uncertainty in the measurement of the strong coupling constant at the @xmath48 scale @xcite , while the octet mass affects the running of @xmath49 .
one sees how unification prefers @xmath50 to lie roughly between 1 and 6 . ] for the sm , the @xmath41 are @xmath51 , while in the phase they read @xmath52 , for @xmath53 active fermion octets @xmath30 .
it should be noted that while we do not speculate here about the possible embedding of into some bigger group ( for example @xmath54 ) , it can be shown on very general grounds that the @xmath55 charge normalization should be @xmath56 .
given the relation between @xmath57 and the sm hypercharge indicated by eq .
[ eq : ql ] , it follows that @xmath58 at the 3 - 3 - 1 breaking scale .
figure [ fig : unificationplots ] illustrates the running of the gauge coupling constants in our model , with the 3 - 3 - 1 scale fixed at 1 tev and the three octets @xmath30 integrated out at 3 tev .
the exotic scalar states are also integrated out at the 1 tev scale , although the running of the gauge couplings is not very sensitive to this value .
given that the @xmath59 coefficients are not very different from @xmath60 with the three octets , unification is sensitive mostly to the ratio @xmath50 , and not to the 3 - 3 - 1 scale _ per se_. the effect is shown in fig .
[ fig : unificationplot_zoomed ] ; allowing for threshold and 2-loop effects , one can see that unification constrains @xmath61 to lie between 1 and 6 .
lepton mass matrices arise from the lagrangian in eq . [
eq : lag ] after breaks to @xmath62 , @xmath63 in the basis where @xmath64 and @xmath65 , the charged leptons mass matrix reads @xmath66 where the entries are given by @xmath67 note that the _ vev _ @xmath68 ( together with @xmath69 ) sets the breaking scale .
in contrast , the _ vevs _ @xmath70 and @xmath71 must lie at the electroweak scale since they belong to @xmath72 doublets after the breaking of @xmath6 .
+ from these expressions it follows that amongst the charged leptons , there is only a pair which is light in each generation , namely : @xmath73 where @xmath74 and @xmath75 . here
the parameter @xmath76 is constrained by unification to lie in the range @xmath77 .
these two 2-component states form the standard dirac charged lepton which now has a squared mass given by @xmath78 , an expression that differs from the sm one .
notice that the presence of states which do not come from the @xmath79 electroweak representation is @xmath80-suppressed .
the two remaining pairs of charged leptons are heavy , with octet - scale masses .
+ turning now to the neutral fermions , their mass matrix reads : + + in the eigenbasis @xmath81 . in the one - family approximation , @xmath82 has a null eigenvector @xmath83 with @xmath84 given as @xmath85 , where @xmath86 is some normalization factor.note that the observed neutrinos are mainly a mixture of @xmath87 and @xmath29 which are both in the @xmath79 representation of the group .
the admixture of the remaining neutrino states are suppressed by at least a factor @xmath80 , which can be vanishingly small .
we have verified that in the multi - generation case @xmath82 has a null eigenvector associated to each of the three generations of leptons . as a result neutrinos are massless in the tree level approximation . this property forms the basis of the radiative mechanism discussed below .
+ diagrams contributing to light neutrino mass . ] at the one - loop level , the exchange of gauge bosons will give rise to a dimension - nine operator which , after symmetry breaking , yields a small neutrino mass through diagrams such as those displayed in fig . [ fig : massinserionexample ] . in order to understand the result of a detailed exact calculation here we simply focus on a typical contribution illustrated in fig .
[ fig : massinserionexample ] , which is found to be of the form @xmath88 where @xmath89 is some loop function with dimensions of @xmath90 , depending on the internal masses .
this approximation captures the key features of the exact result such as ( @xmath91 ) the gauge nature of the underlying radiative dimension - nine seesaw mechanism , requiring two bosonic and three fermionic mass insertions in the internal lines ; as well as ( @xmath92 ) the symmetry structure , both gauge as well as lepton number , as can be seen explicitly .
one also sees that no mass is generated in the limit where @xmath93 is set to zero . ] . with a _
alignment which minimizes this factor , neutrinos get a sub - ev mass even for large @xmath94 values . should @xmath95 be significantly smaller than unity
, one must suppress the mixing between @xmath96 and @xmath97 , since in this case one of the neutrino mass eigenstates is mainly @xmath96 , with approximate mass @xmath98 .
this is readily achieved by setting @xmath99 .
in this letter we have proposed an electroweak gauge extension of the sm in which neutrinos are massless at tree - level .
even though the neutral fermion mass matrix has a seesaw structure , the messengers only provide mass at the loop level , thanks to the symmetry protection .
gauge mediated radiative corrections generate small calculable neutrino masses .
the physics responsible for providing small neutrino masses is also responsible for gauge coupling unification , which can be achieved at a characteristic scale of order tev in the absence of supersymmetry and of gut - like interactions .
a plethora of new states such as new gauge bosons and fermions makes the model directly testable at the lhc , with a non - trivial interplay between the quark sector and the lepton sector .
the presence of such new features is presently under investigation .
+ we thank martin hirsch for valuable discussions .
this work was supported by spanish grants fpa2011 - 22975 and multidark csd2009 - 00064 ( _ mineco _ ) , and prometeoii/2014/084 ( _ generalitat valenciana _ ) .
fgc acknowledges support from _ conacyt _ under grant 208055 .
rf was supported by _
fundao para a cincia e a tecnologia _ through grants cern / fp/123580/2011 and expl / fis - nuc/0460/2013 .
21ifxundefined [ 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 ( ) , http://stacks.iop.org/1674-1137/38/i=9/a=090001 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevd.22.738 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1007/jhep08(2014)039 [ * * , ( ) ] , link:\doibase 10.1007/jhep02(2014)112 [ * * , ( ) ] , link:\doibase 10.1007/jhep02(2013)023 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.81.053004 [ * * , ( ) ] link:\doibase 10.1007/jhep04(2014)133 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.90.073001 [ * * , ( ) ] , link:\doibase 10.1155/2014/831598 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.22.2227 [ * * , ( ) ] link:\doibase 10.1103/physrevd.90.013005 [ * * , ( ) ] link:\doibase 10.1142/s0217751x07036142 [ * * , ( ) ] , \doibase http://dx.doi.org/10.1016/0370-2693(81)90011-3 [ * * , ( ) ] link:\doibase 10.1103/physrevd.25.774 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.30.1343 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.30.1346 [ * * , ( ) ]
|
the physics responsible for gauge coupling unification may also induce small neutrino masses . we propose a novel gauge mediated radiative seesaw mechanism for calculable neutrino masses .
these arise from quantum corrections mediated by new ( 3 - 3 - 1 ) gauge bosons and the physics driving gauge coupling unification .
gauge couplings unify for a 3 - 3 - 1 scale in the tev range , making the model directly testable at the lhc .
|
most stars are born spinning rapidly . observations of young clusters reveal a wide range of rotation rates , with some stars spinning more than 100 times faster than the sun ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this scatter is generally attributed to the vortical nature of the turbulent interstellar clouds from which these stars form . though much of the initial angular momentum is lost during the gravitational collapse of a protostellar core , the dramatic decrease in the moment of inertia during
the collapse is enough to impart a substantial rotational velocity in a large fraction of young stars . as the stars age
, torques from magnetized stellar winds cause them to spin down and this reduces the scatter of the general population @xcite .
the rapid rotation of young stars can have a dramatic influence on their internal structure and dynamics . in the most extreme cases ,
the centrifugal force can alter the stars shape , causing it to be significantly oblate .
models of rotating stars suggest that the effects of rapid rotation and oblateness can alter the stars s convective and radiative heat transport , reduce their luminosities , induce shearing instabilities and global circulations , and modify their rotational and compositional evolution @xcite .
much of this work has focused on high - mass stars because of their short lifetimes , which makes them intrinsically more likely for fast rotation to play an important evolutionary role from birth to death , when they explode as supernovae and enrich the galaxy with heavy elements .
high - mass stars are also luminous and large , which makes them relatively easy to observe . in the last fifteen years , researchers have been able to directly detect the oblate shapes of several rapidly - rotating , massive stars by means of optical and infrared interferometry @xcite .
more direct detections are likely to be made in the future as observations continue to improve @xcite .
one of the major theoretical predictions that has been addressed by these observations is the phenomenon of gravity darkening , which was introduced nearly a century ago in a pioneering paper by @xcite . for a massive star with a radiative outer envelope
, von zeipel argued that the effective temperature of the stellar surface should scale as @xmath1 , where @xmath2 and @xmath3 is the effective gravity , equal to the newtonian gravity plus the centrifugal force .
this in turn implies that the radiative energy flux emitted from the stellar surface scales as @xmath4 . since the centrifugal force acts to reduce @xmath3 at the equator
, gravity darkening predicts that the equatorial regions should be darker than the polar regions .
though observations have confirmed the existence of gravity darkening , they suggest that the exponent may be somewhat less than predicted , with @xmath5 0.13 - 0.25 @xcite .
lower - mass , late - type stars with convective envelopes are also expected to exhibit gravity darkening but again , with a smaller latitudinal variation than that predicted by von zeipel , with @xmath6 possibly as low as 0.08 @xcite .
however , these expectations are based on simplified models of convection , with little observational guidance .
substantial progress has been made in recent years in modeling the structure of rapidly rotating stars .
a number of models have been used to investigate the effects of oblateness on stellar structure and evolution , with implications for asteroseismology @xcite .
though these models are ostensibly @xmath7 , the thermodynamic quantities only depend on the effective gravitational potential . furthermore
, like most stellar structure models , these models do not address the internal dynamics of oblate stars ; they solve only for the hydrostatic structure using a mixing - length formulation for convection , though some do include idealized cylindrical or shellular differential rotation profiles .
@xcite took a somewhat different approach , calculating the non - spherical shape and internal structure of a rapidly rotating gaseous body using a three - dimensional , finite - element code with an unstructured grid . but
again , the fluid dynamics was not considered . other analytical and numerical work on fluid dynamics in ellipsoidal geometries
has focused on flows induced by precession ; for a recent example that employs a finite - element approach see @xcite .
the first numerical model ( to our knowledge ) capable of capturing the internal dynamics of oblate stars is the ester ( evolution stellaire en rotation ) code @xcite .
ester solves the axisymmetric , steady - state fluid equations self - consistently , taking into account nuclear energy generation , differential rotation and meridional circulation .
however , convection is still treated using either a turbulent diffusion approximation @xcite or an isentropic approximation @xcite in which the effective convective heat flux is inferred from the nuclear energy generation and radiative heat flux under the assumption of a steady state .
furthermore , published ester simulations are currently limited to high - mass stars with convective cores ( @xmath8 ) ; they have not yet considered convective envelopes .
there have been many global 3d hydrodynamic ( hd ) and magnetohydrodynamic ( mhd ) simulations of convection in the envelopes of lower - mass stars that address the rapid rotation regime ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
. however , these do not take into account the oblateness caused by the centrifugal force .
in this study , we present the first 3d hd simulations of convection in the oblate envelopes of rapidly - rotating solar - type stars .
these simulations are performed using the compressible high - order unstructured spectral difference ( chorus ) code recently introduced by ( * ? ? ?
* hereafter wlm15 ) .
we focus in particular on the influence of the oblateness on the convective structure , heat transport , mean flows , and the thermodynamic stratification .
this paper is organized as follows . in section [ sec : mp ] we discuss the chorus code and its application to model oblate spheriodal geometries .
we follow this in section [ sec : setup ] with a description of the numerical experiments that we have performed , which compare oblate simulations at varying rotation rates with their spherical counterparts .
we consider rotation rates up to @xmath9 where @xmath10 is the critical angular velocity at which the outward centrifugal force at the equator equals the inward gravitational force .
we then present our results , focusing first on the convective structure and energetics ( section [ sec : con ] ) , then on the energy transport ( section [ sec : transport ] ) and finally on the mean flows and associated thermal gradients ( section [ sec : mean ] ) .
we summarize our results , conclusions and future plans in section [ sec : summary ] .
modeling solar and stellar convection is challenging because the flow is turbulent and a vast range of spatial and temporal scales are involved . the reynolds number @xmath11 ( where @xmath12 and @xmath13 are characteristic velocity and length scales and @xmath14 is the kinematic viscosity ) is typically very high so direct numerical simulations that effectively resolve all scales in the convective envelope are far beyond today s computational affordability . in this study , we only aim to capture large - scale dynamics which are crucial in order to establish convection , differential rotation , and thermal flux balance .
to do so , the three - dimensional chorus code is employed to perform a series of global simulations .
chorus solves the equations of hydrodynamics using the high - order spectral difference method based on unstructured grids ( wlm15 ) .
it can be naturally extended to deal with complex geometries including oblate spheriodal shells . in what follows
we will generally refer to a spherical polar coordinate system @xmath15 , although the chorus code employs a cartesian coordinate system to represent a spherical or oblate spheriodal geometry using an unstructured mesh ( sec .
[ sec : der_grid ] ) .
we consider an ideal gas within an oblate spheroidal shell uniformly rotating at a constant rate @xmath16 about the @xmath17axis . for simplicity
, we assume the bulk of mass is inside the inner radius of the oblate spheroidal shell , giving the gravitational acceleration @xmath18 , where @xmath19 is the gravitational constant , @xmath20 is the interior mass and @xmath21 is the radial unit vector . in slowly rotating solar - type stars ,
the centrifugal force has typically been neglected in the momentum equation considering its sufficiently small magnitude relative to the buoyancy force ( e.g. * ? ? ?
* ; * ? ? ?
* ) . however , in rapidly rotating stars , the centrifugal force is comparable to gravity and has a significant contribution to stellar structure .
thus , in addition to the coriolis force , we include the centrifugal force in the hydrodynamic equations .
the resulting equations in the rotating reference frame are @xmath22 @xmath23 @xmath24 where @xmath25 , @xmath26 , @xmath27 , @xmath28 and @xmath29 are time , pressure , temperature , density and the velocity vector respectively .
@xmath30 is the total energy per unit volume and is defined as @xmath31 where @xmath32 is the adiabatic index ( the ratio of specific heats at fixed pressure and volume ) .
@xmath33 is the viscous stress tensor for a newtonian fluid and is given by @xmath34,\ ] ] where @xmath35 is the strain rate tensor and @xmath14 is the kinematic viscosity .
the term @xmath36 in equation ( [ eqn : energy ] ) represents viscous heating .
the diffusive flux @xmath37 is treated in the form of @xmath38 where the first component is a radiation diffusion flux with the molecular radiation diffusion coefficient @xmath39 and @xmath40 is the specific heat at constant pressure .
the second component is the entropy diffusion due to unresolved , subgrid - scale convective motions .
this is a crude but commonly used representation of subgrid - scale ( sgs ) heat transport @xcite .
@xmath41 is the specific entropy .
the kinematic viscosity @xmath14 and the thermal diffusivity @xmath42 are assumed to be constant in time and space ( independent of radius ) , with magnitudes as specified in section [ sec : setup ] .
the radiative diffusivity @xmath43 peaks at the inner boundary and decreases outward .
it is computed from the temperature and density distribution according to the expression @xmath44 . since the initial @xmath28 and @xmath27 profiles depend only on the roche potential @xmath45 ( see section [ sec : ic ] ) , @xmath43 is also initially a function of @xmath45 alone .
it does evolve in time along with @xmath28 and @xmath27 but variations along a potential surface remain relatively small so @xmath46 .
the value of @xmath47 is chosen such that the total flux passing through the lower boundary is equal to the stellar luminosity @xmath48 ( see section [ sec : bdry ] ) .
the use of the total energy equation ( [ eqn : energy ] ) is advantageous from the point of view of the conservation of energy but it can pose problems in stellar interiors where the stratification is adiabatic to within about one part in @xmath49 . to capture the small departures from adiabaticity that drive the convection requires a very accurate numerical method .
however , as discussed in section [ sec : setup ] , we have artificially increased the luminosity in our models in order to achieve more tractable values of the mach number .
this has the additional advantage that it increases the superadiabatic entropy gradient .
thus , the nondimensional amplitude of the entropy gradient , @xmath50 , is only a factor of 600 less than the nondimensional temperature gradient @xmath51 . in other words ,
the departure from adiabatic stratification is more than one part in @xmath52 .
we estimate the accuracy of our spectral difference method ( described in section [ sec : method ] ) to be at least @xmath53 for length scales on the order of the entropy gradient and we see no signs of spurious behavior . in the chorus code , the spectral difference ( sd ) method is used to discretize equations ( [ eqn : mass_conservation])-([eqn : energy ] ) in space by using their original strong form ( differential form ) @xcite .
in contrast , the traditional discontinuous galerkin ( dg ) method starts from the weak ( integrated ) form of these equations ( e.g. * ? ? ?
the sd method can be viewed as a particular form of nodal dg method @xcite . to discretize these equations in space , the computational domain
is divided into a collection of non - overlapping hexahedral elements .
these elements in physical coordinates ( @xmath54 , @xmath55 , @xmath56 ) are then transformed to a unit cube ( @xmath57 , @xmath58 , @xmath59 ) using an isoparametric mapping with quadratic relations .
consequently , the governing equations are transformed accordingly through jacobian matrices . in the unit cube , two sets of points are defined , namely the solution and flux points . in each coordinate direction of ( @xmath57 , @xmath58 , @xmath59 ) ,
@xmath60 solution points are chosen to be the chebyshev - gauss - quadrature points and @xmath61 flux points are selected to be roots of the legendre polynomial plus two end points .
then the solution and flux polynomials can be built through lagrange interpolation basis functions . in order to deal with flux discontinuities across element interfaces , an approximate riemann solver ( rusanov or lax - friedrichs scheme ) and br1 @xcite scheme
are employed for inviscid and viscous fluxes respectively . over each elemental interface ,
@xmath62 flux points are needed to construct both inviscid riemman and viscous br1 fluxes where @xmath60 is the order of the accuracy of the sd method .
the chorus code has proven to be particularly attractive for parallel computing ( wlm15 ) .
the high - order sd method is locally intensive in term of computation associated with @xmath63 solutions points per element and requires a light amount of communication across elemental interfaces with @xmath62 flux points . for temporal discretization , a fourth - order accurate , strong - stability preserving five - stage runge - kutta scheme
is employed @xcite .
for further details on the numerical method see wlm15 . for anelastic codes ,
the initial conditions are often based on a 1d , unstable , equilibrium , polytropic solution derived from the governing equations with an imposed superadiabatic entropy gradient , @xmath64 @xcite .
however , this approach does not apply here for two reasons .
one is the exclusion of the centrifugal force .
taking this into account in the initial hydrostatic balance requires a 2d setup ( both radial and latitudinal directions ) which poses a problem for anelastic codes that generally expand about a 1d reference state .
the second reason that we can not use a typical anelastic setup is the compressible nature of chorus . in an anelastic code
, the superadiabatic entropy gradient can be incorporated into the thermodynamic perturbations and decoupled from the specification of the reference state . since chorus does not linearize the equations about a reference state , this does not apply .
for these reasons , we take a different approach .
instead , we initialize chorus with a 2d , hydrostatic , adiabatic stratification ( @xmath65 ) . this has the advantage that it can be defined analytically , even for oblate geometries .
however , in contrast to some anelastic setups , this initial state is not an equilibrium solution to the equations because it does not satisfy the steady - state energy equation .
in particular , there is a non - zero convergence of the radiative flux which heats the plasma and establishes the superadiabatic entropy gradient that eventually triggers convection . in this section
we describe how this initialization is accomplished for both spherical and oblate spheroidal shells .
note that equations ( [ eqn : p_initial])-([eqn : t_initial ] ) reduce to the 1d polytropic solutions of @xcite if the centrifugal force term is omitted from the effective gravitational potential in equation ( [ eqn : potential ] ) .
as shown there , @xmath87 corresponds to the number of density scales heights across the shell .
the initial profile is completely specified by choosing the parameters @xmath20 , @xmath82 , @xmath83 , @xmath16 , @xmath32 , @xmath40 , @xmath88 , and @xmath89 . for our oblate simulations ,
we use the 2d profiles in equations ( [ eqn : p_initial])-([eqn : t_initial ] ) and for our spherical shell simulations we use the 1d profiles of @xcite which can be obtained from ( [ eqn : p_initial])-([eqn : t_initial ] ) by setting @xmath90 . since we wish to model solar - type stars , we use appropriate parameter values of @xmath91 , @xmath92 g @xmath93 , @xmath94 g , @xmath95 erg g@xmath96 k@xmath96 , and @xmath97 .
note also that the initialization procedure described here implies that the oblate geometries will have both a larger volume and a larger mass than their spherical counterparts ( see section [ sec : setup ] ) .
these polytropic , adiabatic initial conditions do not satisfy the schwarzschild criterion for convective instability , which requires a negative ( superadiabatic ) radial entropy gradient ( @xmath98 ) .
furthermore , the rayleigh number @xmath99 is subcritical .
thus , convection will not occur immediately .
however , as noted at the beginning of this section , there is a fixed radiative heat flux @xmath100 that carries the stellar luminosity in through the bottom boundary ( see also sections [ sec : hydro ] , [ sec : bdry ] and [ sec : flux ] ) .
the convergence of this heat flux heats the plasma in the shell , raising its temperature .
since the temperature at the top is fixed ( see section [ sec : bdry ] ) , this heating of the plasma increases the temperature gradient .
soon the temperature gradient becomes both superadiabatic and supercritical , and convection ensues .
the temperature and entropy gradients will eventually equilibrate when the simulation achieves flux balance , such that the combined convective and diffusive energy fluxes carry the full stellar luminosity through the entire shell ( see section [ sec : flux ] ) .
this process is illustrated in fig .
[ fig : entropy_s60 ] , which shows how the entropy gradient in a typical simulation ( s60 ; see table [ tab : parameters ] ) evolves with time .
it starts out adiabatic ( @xmath65 ) , but it quickly becomes superadiabatic ( @xmath101 ) .
the heating of the shell due to the convergence of the radiative heat flux not only increases the temperature , it also raises the adiabat .
this increases the negative entropy gradient throughout the shell , but particularly near the upper boundary . by @xmath102 6 days
, it becomes superadiabatic enough to trigger convection ( see figs . [
fig : ke ] and [ fig : ke_so ] below ) . by @xmath102 50 days ,
the simulation has achieved flux balance and the entropy gradient has equilibrated .
[ sec : bdry ] the inner and outer boundaries , @xmath103 and @xmath104 , are assumed to be impenetrable and free of viscous shear stresses @xmath105 and @xmath106 where @xmath107 is the unit vector normal to equipotential surfaces @xmath45 and pointing radially outward . at the bottom boundary a fixed heat flux
is imposed such that @xmath108 where @xmath48 is the stellar luminosity and @xmath109 is the area of the inner surface @xmath103 . for the spherical simulations , @xmath110 , but for the oblate simulations the area is greater than this and
is calculated numerically . since @xmath111 at @xmath103 ( see fig . [
fig : entropy_s60 ] ) , this imposed heat flux is almost entirely due to the radiative diffusion term @xmath112 .
the magnitude of @xmath43 is specified such that equation ( [ eq : fbot ] ) is satisfied ( see section [ sec : hydro ] ) . at the upper boundary , @xmath104
, the temperature is fixed at the value specified by the initial conditions ( section [ sec : ic ] ) .
[ sec : der_grid ] in our model , the oblate outer surface of the star is defined by the equipotential surface @xmath104 . in order to determine the outer surface shape , the polar ( @xmath113 ) radius @xmath114 is specified and kept unchanged .
then the potential at the pole can be compared to that at an arbitrary polar angle @xmath67 .
the resulting cubic equation for @xmath115 is @xmath116 by solving equation ( [ eqn : geo_equipotential2 ] ) for @xmath117 , the axisymmetric ( @xmath118-independent ) surface is obtained .
the inner equipotential surface can be computed in a similar way by replacing the outer polar radius @xmath83 with the inner polar radius @xmath82 in equation ( [ eqn : geo_equipotential2 ] ) .
the geometry of an oblate spheroidal shell is then defined by both inner and outer equipotential surfaces .
it is straightforward to generate grids for spherical shells by using commercial or publicly available grid generation software ( e.g. gmsh ) .
spherical shells can also be represented by a cubed - sphere grid @xcite .
however , it is very challenging to generate grids for oblate spheroidal shells as defined above because we need to represent the curvature on both the bottom and top surfaces very accurately in order to conserve mass , momentum , and energy . here , an alternative approach has been chosen .
we first generate a spherical shell grid with inner radius @xmath119 and outer radius @xmath120 using gmsh software @xcite .
the following procedure is taken to deform this generated spherical shell grid to the desired oblate spheroidal shell grid with the rotating rate @xmath16 . 1 . in the spherical shell grid , for an arbitrary grid node point @xmath75 ,
find the points of intersection between line @xmath121 and bottom and top surfaces .
denote these as @xmath109 and @xmath122 respectively ( see fig .
[ fig : deform_fig]a ) .
note that point @xmath123 is the center .
2 . calculate lengths of @xmath124 , @xmath125 , @xmath126 , @xmath127 and @xmath128 .
3 . define new bottom and top surfaces using equation ( [ eqn : geo_equipotential2 ] ) .
4 . keep the latitude and longitude of line @xmath129 identical to that of line @xmath121 and find the new intersection points @xmath130 and @xmath131 ( see fig .
[ fig : deform_fig]b ) .
as shown in figure [ fig : move_points ] , the position of point @xmath132 is obtained by satisfying @xmath133 the spherical shell grid and the newly generated oblate spheroidal grid are shown in figure [ fig : sphere_oblate ] .
this procedure is performed at the preprocessing stage and all desired oblate spheroidal shell grids with varying @xmath16 ( discussed in section [ sec : setup ] ) are generated from a single spherical shell grid .
the generated oblate spheroidal shell grids are smooth as shown in figure [ fig : mesh_plane ] ; the node positions on both top and bottom surfaces are exactly specified ( to machine precision ) by solving equation ( [ eqn : geo_equipotential2 ] ) .
the objective of this study is to investigate the effects of oblateness on the properties of the convection , heat transfer , and mean flows in rapidly - rotating solar - type stars . to this end , we have performed global - scale numerical simulations within spherical shells ( neglecting the centrifugal force ) as well as oblate spheroidal shells that include the centrifugal force in order to investigate the difference .
simulations within spherical shells are labeled with leading letter @xmath134 while those within oblate spheroidal shells are marked with leading letter @xmath135 as shown in tables [ tab : parameters ] and [ tab : diagnostics ] . the computational domain for
the spherical shell extends from @xmath136 to @xmath137 where @xmath138 cm is the solar radius .
the computational domains for the oblate spheroidal shells are derived from the spherical shell using the approach described in section [ sec : der_grid ] .
as described there , the inner and outer radii at the poles are the same as in the spherical cases , but the depth of the convection zone becomes wider at low latitudes due to the centrifugal distortion .
the low mach numbers of deep stellar interiors pose a significant challenge for fully compressible codes .
in particular , if explicit timestepping is used as is the case here , they set strict limits on the size of the time step needed to satisfy the courant - freidrichs - lewy ( cfl ) condition .
they also imply long thermal equilibration times .
we address this challenge as in previous simulations by artificially increasing the luminosity ( e.g. * ? ? ?
* wlm15 ) .
this has the effect of increasing the velocity amplitudes ( which scale roughly as @xmath139 ) and decreasing the thermal relaxation time . for all simulations reported here we use @xmath140 , where @xmath141 erg s@xmath96 is the solar luminosity .
thus , we estimate that the velocity amplitudes in our simulations are artificially enhanced by about a factor of 40 .
our studies here explore a range of rotation rates from 60@xmath142 to @xmath143 ( cases @xmath144-@xmath145 and @xmath146-@xmath147 ) where @xmath148 rad @xmath149 is the fiducial solar rotation rate
. however , the effective rossby number ro @xmath150 in our simulations is artificially increased by roughly a factor of forty because of the larger convective velocity amplitudes ( relative to a star rotating at the same rotation rate ) .
this problem is sometimes addressed by artificially increasing the rotation rate along with the luminosity to achieve a desired rossby number ( e.g. * ? ? ?
however , we do not have that luxury here . since we include the centrifugal force ,
there is a critical rotation rate @xmath151 where the outward centrifugal acceleration of rotation equals the inward gravitational acceleration @xcite . if the star were to rotate faster than @xmath10 it would break up .
our range of rotation rates extends up to 85% of @xmath10 ( see table [ tab : parameters ] ) . in other words ,
the oblateness of these stars is comparable to a real star rotating at these rotation rates but the rossby number is about a factor of forty higher .
we could in principle remedy this situation by adjusting @xmath20 and @xmath19 or we could simply accept that , though the convection in our simulations is strongly rotationally constrained ( ro @xmath152 ; see table [ tab : diagnostics ] ) , it is less so than in a real star with the same rotation rate . for this initial study
we follow the latter approach and take it into account in our interpretation .
all of the simulations discussed here employ the third - order accuracy spectral difference method on 393,216 hexahedral elements , resulting in a total of 10,616,832 degrees of freedom ( @xmath153 ) .
this corresponds to 96 solution points in radius and 192 solution points in latitude .
the number of solution points in longitude is twice this at the equator , 384 , but diminished toward higher latitudes to achieve a nearly uniform horizontal resolution .
the values of @xmath14 and @xmath42 are selected to be small , but large enough to ensure that the flows are well resolved .
since we wish to focus on the effects of the oblateness , @xmath14 and @xmath42 are held fixed at a value of @xmath154 @xmath155 s@xmath96 , implying a prandtl number @xmath156 .
however , in order to make contact with the previous global simulations of @xcite , we also run a series of spherical simulations in which @xmath14 and @xmath42 are reduced as the rotation rate is increased in order to maintain the supercriticality of the convection .
these are listed in tables [ tab : parameters ] and [ tab : diagnostics ] as cases @xmath157 , @xmath158 , and @xmath159 . in these simulations , @xmath160 , where the normalization is chosen such that @xmath154 @xmath155 s@xmath96 when @xmath161 .
thus , this @xmath162 series of simulations also has @xmath163 and follows a parameter path that includes @xmath145 .
all of the simulations considered here are relatively laminar .
this is appropriate for the innovative nature of this study ; these set the baseline for future simulations that can explore the parameter sensitivities more thoroughly .
reynolds numbers are on the order of 10 - 20 , as shown in table [ tab : parameters ] . recall that the convective velocity is artificially enhanced so the reynolds numbers are higher than might otherwise be expected for this value of @xmath14 .
the supercriticality of these simulations is not easy to establish due to the lack of a linear theory for convection in oblate , compressible spheroids and due to the nature of the initial conditions , which do not express an unstable equilibrium ( sec .
[ sec : ic ] ) .
however , we have determined the critical rayleigh number ra@xmath164 for the least supercritical simulation , namely case o120 . here
convection is inhibited by both the rapid rotation and the oblateness .
we determine ra@xmath164 by progressively increasing @xmath14 and @xmath42 until convection is suppressed .
the results indicate a relatively small value of ra / ra@xmath165 .
all other simulations are expected to have a higher supercriticality .
for example , in boussinesq systems , the critical rayleigh number for the onset of convection in rotating spherical shells scales as @xmath166 @xcite .
this would imply a supercriticality on the order of 6 for case o60 .
other quantities listed in table [ tab : diagnostics ] include the volume - averaged kinetic energy density relative to the rotating reference frame , ke , as well as the contributions to ke from the convection ( cke ) , differential rotation ( drke ) , and meridional circulation ( mcke ) .
these quantities are defined as follows : @xmath167 @xmath168 @xmath169 and @xmath170 where @xmath171 denotes the volume of the computational domain and angular brackets denote averages over longitude .
the thermal diffusion time @xmath172 is roughly 15 days , the convective turnover time scale is roughly 1.2 days , and the rotation period ranges from 0.23 - 0.47 days .
all simulations have been run for at least 48 days so this spans at least three diffusion times , at least 40 turnover times , and at least 100 rotation periods .
all appear to be equilibrated ( see figs .
[ fig : ke ] and [ fig : ke_so ] below ) .
one more comment is in order before proceeding to the main results .
when interpreting the ke data in table [ tab : diagnostics ] and below , it is important to realize that the oblate shells have more mass and volume than their spherical counterparts .
this is a consequence of our initialization approach discussed in section [ sec : ic ] . in particular , the density on the inner equipotential surface , @xmath173 is the same for all simulations .
however , the oblate geometries have a greater volume , which gives rise to a larger mass .
the mass of the convective envelope @xmath174 for each simulation is listed in table [ tab : parameters ] relative to the central mass @xmath20 .
the most extreme case , o120 , has about 37% more mass in the convection zone than s120 .
this is somewhat less than the volume difference , which is about 51% greater in case o120 than in s120 ( table [ tab : parameters ] ) .
thus , although the volume - averaged ke density listed in table [ tab : diagnostics ] is slightly smaller in case o120 compared to s120 , the total kinetic energy integrated over the entire volume ( in ergs ) is larger by about 36% , comparable to the difference in mass . in other words ,
the mean kinetic energy per unit mass ( @xmath175 multiplied by @xmath171 and divided by @xmath174 ) is almost the same in cases o120 and s120 ( within 1% ) .
this is a measure of the density - weighted velocity amplitude .
the other oblate cases have slightly higher kinetic energy per unit mass than their spherial counterparts , ranging from 3% higher in o60 to 6% higher in o70 , o100 , and o110 to 11% and 15% higher in cases o90 and o80 .
the downward trend in this ratio for @xmath176 may be due to a saturation of drke / ke in the fastest - rotating oblate stars ( see section [ sec : energetics ] ) .
note also that @xmath177 in all cases ( table [ tab : parameters ] ) , justifying our assumption in section [ sec : ic ] that the bulk of the mass lies inside the inner radius .
lccccc @xmath147 & @xmath178 & @xmath179 & 1.51 & @xmath180 & @xmath181 + @xmath182 & @xmath183 & @xmath184 & 1.38 & @xmath185 & @xmath186 + @xmath187 & @xmath188 & @xmath189 & 1.29 & @xmath190 & @xmath191 + @xmath192 & @xmath193 & @xmath194 & 1.21 & @xmath195 & @xmath196 + @xmath197 & @xmath198 & @xmath199 & 1.16 & @xmath200 & @xmath201 + @xmath202 & @xmath203 & @xmath204 & 1.12 & @xmath205 & @xmath206 + @xmath146 & @xmath207 & @xmath208 & 1.08 & @xmath209 & @xmath210 + @xmath145 & @xmath178 & @xmath179 & 1 & @xmath211 & @xmath212 + @xmath213 & @xmath183 & @xmath184 & 1 & @xmath211 & @xmath212 + @xmath214 & @xmath215 & @xmath216 & 1 & @xmath211 & @xmath212 + @xmath217 & @xmath193 & @xmath194 & 1 & @xmath211 & @xmath212 + @xmath218 & @xmath198 & @xmath199 & 1 & @xmath211 & @xmath212 + @xmath219 & @xmath203 & @xmath204 & 1 & @xmath211 & @xmath212 + @xmath144 & @xmath207 & @xmath208 & 1 & @xmath211 & @xmath212 + @xmath157 & @xmath220 & @xmath216 & 1 & @xmath211 & @xmath212 + @xmath158 & @xmath221 & @xmath199 & 1 & @xmath211 & @xmath212 + @xmath159 & @xmath222 & @xmath208 & 1 & @xmath211 & @xmath212 + [ tab : parameters ] lcccccc @xmath147 & @xmath223 & @xmath224 & @xmath225 & @xmath226 & @xmath227 & @xmath228 + @xmath182 & @xmath229 & @xmath230 & @xmath231 & @xmath232 & @xmath233 & @xmath234 + @xmath187 & @xmath235 & @xmath230 & @xmath236 & @xmath237 & @xmath238 & @xmath239 + @xmath192 & @xmath240 & @xmath241 & @xmath242 & @xmath243 & @xmath244 & @xmath245 + @xmath197 & @xmath246 & @xmath247 & @xmath248 & @xmath249 & @xmath250 & @xmath251 + @xmath202 & @xmath215 & @xmath252 & @xmath253 & @xmath254 & @xmath255 & @xmath256 + @xmath146 & @xmath257 & @xmath258 & @xmath259 & @xmath260 & @xmath261 & @xmath262 + @xmath145 & @xmath263 & @xmath264 & @xmath265 & @xmath232 & @xmath266 & @xmath267 + @xmath213 & @xmath260 & @xmath268 & @xmath269 & @xmath237 & @xmath270 & @xmath271 + @xmath214 & @xmath272 & @xmath273 & @xmath274 & @xmath275 & @xmath276 & @xmath277 + @xmath217 & @xmath278 & @xmath279 & @xmath242 & @xmath263 & @xmath280 & @xmath281 + @xmath218 & @xmath282 & @xmath283 & @xmath235 & @xmath284 & @xmath285 & @xmath286 + @xmath219 & @xmath287 & @xmath288 & @xmath289 & @xmath290 & @xmath291 & @xmath292 + @xmath144 & @xmath293 & @xmath294 & @xmath295 & @xmath296 & @xmath297 & @xmath298 + @xmath157 & @xmath299 & @xmath300 & @xmath301 & @xmath302 & @xmath303 & @xmath304 + @xmath158 & @xmath305 & @xmath306 & @xmath307 & @xmath308 & @xmath309 & @xmath310 + @xmath159 & @xmath305 & @xmath311 & @xmath312 & @xmath313 & @xmath314 & @xmath315 + [ tab : diagnostics ]
the evolution of the kinetic energy for the @xmath135 and @xmath134 series of simulations is illustrated in figure [ fig : ke ] .
all of these simulations start from the polytropic , adiabatic stratification described in section [ sec : ic ] .
therefore , the initial entropy gradient is zero for all cases . with the heating due to the radiative heat flux at the bottom and the constant temperature at the top
, the entropy gradient becomes superadiabatic .
once the negative entropy gradient is built up and the rayleigh number @xmath99 exceeds a critical value , the flow locks on to the fastest - growing eigenmode which then grows exponentially .
however , even before this exponential growth phase , there is a slow initial algebraic growth in the ke in response to the radiative heating .
we have verified that this growth does not occur when the luminosity is set to zero so it is not due to numerical errors , such as those associated with the conservation of angular momentum or hydrostatic balance , which can pose problems on unstructured grids .
we find our spectral difference method and grid generation procedure described in section [ sec : mp ] to be very effective at minimizing these numerical errors . to demonstrate this , consider the three cartesian components of the angular momentum vector , @xmath316 , @xmath317 , and @xmath318 , integrated over the volume of the shell .
if the numerics were perfect , these three values would remain zero relative to the rotating frame throughout the simulation .
our test cases with zero luminosity indicate that the errors associated with angular momentum conservation and hydrostatic balance are on the order of one part in @xmath319 . in order to prevent the gradual accumulation of these errors over tens of thousands of iterations
, we apply an angular momentum correction procedure every 5000 iterations as described in wlm15 . for our most oblate simulation , case o120 , near the end of the simulation interval ( @xmath320 77 days ) , we find that @xmath321 , @xmath322 and @xmath323 , where @xmath324 is the initial angular momentum of the shell .
the structure of the flow in this initial establishement period suggests that it is indeed weak convective motions that are induced by the heating . the duration of this initial establishment period before the exponential growth phase varies with the rotation rate and oblateness . as the rotation rate
is increased , this duration increases from 6 days for case @xmath144 to 12 days for case @xmath145 ( figure [ fig : ke]a ) .
this can be attributed to the stabilizing influence of rotation , which requires a greater entropy gradient to trigger convection .
a similar effect is also observed in the oblate spheroidal shell simulations shown in figure [ fig : ke]b . in order to assess the effect of oblateness on the growth and saturation of the convection ,
four cases are compared in figure [ fig : ke_so ] . the initial establishment period in case @xmath146 is only a little longer than that in case @xmath144 , which is expected because the oblateness in case @xmath146 is minimal as shown in table [ tab : parameters ] .
in contrast , the difference between cases @xmath145 and @xmath147 is pronounced ( about @xmath325 days ) , implying that the significant oblateness in case @xmath147 has a substantial influence on the establishment of a supercritical entropy gradient . this can be attributed to the greater depth of the convection zone at the equator for the oblate cases , which dilutes the flux convergence and weakens thermal gradients .
it is clear from figure [ fig : ke ] and table [ tab : diagnostics ] that the more rapidly - rotating cases saturate at a lower level of ke than the more slowly - rotating cases . this is a consequence of the stabilizing effects of rotation and rotational shear ( differential rotation ) and it occurs in both the spherical and oblate cases .
however , the stabilizing influence on ke is somewhat larger in the oblate cases .
for example , the ke in o120 is about 36% of that in o60 whereas the ratio between s120 and s60 is 41% .
however , recall that ke is the mean kinetic energy density . if the difference in volume is taken into account , then the total kinetic energy in case o120 is about 51% of the total kinetic energy in o60 .
we also note that the decrease in total ke with increasing @xmath326 is a result that is sensitive to our chosen path through parameter space , which is to keep @xmath14 and @xmath42 constant .
recall from section [ sec : setup ] that we decrease @xmath14 and @xmath42 as we increase @xmath326 in the _ a _ series of simulations ( s60a , s80a , s100a , and s120 ) in order to maintain a relatively constant supercriticality . in these simulations ,
the total ke decreases only slightly from s60a to s100a and is in fact largest for s120 , due largely to a strong differential rotation ( table [ tab : diagnostics ] ) . for both the spherical and the oblate cases , more of the ke
is contained in the differential rotation as @xmath326 is increased ( table [ tab : diagnostics ] ) .
this is a common feature of global convection simulations and is attributed to the enhancement of the convective reynolds stress ( e.g. * ? ? ?
in the spherical cases , the ratio of drke / ke increases steadily over the entire range in rotation rates , from 60 to 120 @xmath142 .
however , interestingly , the drke / ke ratio seems to saturate in the oblate cases at about 0.61 for @xmath327 . in this range of rotation rates ,
the drke / ke ratio is less than in the corresponding spherical cases , indicating that pronounced oblateness tends to suppress differential rotation .
this result may or may not be related to the observed saturation of the differential rotation @xmath0 in rapidly - rotating stars , which will be discussed in section [ sec : dr1 ] . for more moderate rotation rates , @xmath328 ,
the oblate cases tend to have a slightly higher ratio of drke / ke compared to their spherical counterparts .
in contrast to drke / ke , the mcke / ke ratio decreases as @xmath326 is increased .
this again is commonly seen in spherical convection simulations and is attributed to the nature of the reynolds stress and the tendency for flows to align with the rotation axis as expressed by the taylor - proudman theorem @xcite . in short ,
the meridional circulation is established in response to the convective angular momentum flux in a process known as gyroscopic pumping @xcite . as the rotational influence
is increased , this flux becomes perpendicular to the @xmath56 axis and is less effective at inducing a meridional circulation .
though both the spherical and oblate cases exhibit a decrease in the mcke / ke ratio as @xmath326 is increased , the effect is somewhat more pronounced in the oblate cases .
furthermore , the mcke / ke ratio is generally less in the oblate cases than in the spherical cases with the same rotation rate . in other words
, the oblateness tends to suppress the mc , at least in terms of its kinetic energy .
this is somewhat of a surprise ; one might have expected the oblateness to induce strong baroclinic circulations but we see no evidence of this .
we will return to these issues in section [ sec : dr1 ] .
in figure [ fig : con_pattern ] , the structure of the convection is illustrated .
shown is the velocity normal to equipotential surfaces @xmath329 , in mollweide projection for a horizontal surface in the upper convection zone .
all simulations are dominated by a series of columnar convective rolls approximately aligned with the rotation axis but sheared slightly in the prograde direction at low latitudes by the differential rotation .
these are the well - known banana cells characteristics of convection in rotating shells for laminar parameter regimes @xcite .
there is little apparent difference between simulations in spherical shells and those in oblate spheroidal shells . in particular ,
the azimuthal wavenumber of the banana cell pattern is similar , on the order of @xmath330 for both s120 and o120 .
this qualitative similarity in appearance is confirmed by the quantitative power spectra , which are compared in figure [ fig : spectrum ] .
each curve represents the power spectrum obtained by applying a spherical harmonic transformation to the normal velocity fields shown in figure [ fig : con_pattern ] and then plotting the resulting power as a function of the spherical harmonic degree , which is , in effect , the horizontal wavenumber .
the differences between the spherical and oblate cases are insignificant . .
however , closer scrutiny of figure [ fig : con_pattern ] does reveal some notable differences , particularly in the polar regions .
it is well known that convection in rotating spherical shells can be divided into two distinct dynamical regions , delineated by the so - called tangent cylinder @xcite .
the tangent cylinder is a 2d cylindrical surface with a central axis aligned with the rotation axis and a cylindrical radius that is equal to the radius of the base of the convection zone at the equator . outside the tangent cylinder ,
the convection is generally dominated by banana cells or related vortex sheets .
these are the equatorial modes .
meanwhile , the polar convection modes inside the tangent cylinder tend to be more horizontally isotropic , particularly at moderate rossby numbers and high rayleigh numbers .
though all simulations exhibit a clear difference between polar and equatorial modes , the difference is much more pronounced in the fastest - rotating oblate cases .
this can be seen most clearly by comparing cases s120 and o120 in figure [ fig : con_pattern]@xmath162 and @xmath331 . in case s120 , the polar modes are highly anisotropic and linked with the equatorial banana modes . by contrast , in case o120 , the polar modes clearly decouple from the equatorial modes and exhibit a more isotropic cellular structure ( see also figs . [
fig : vn_so ] and [ fig : vn_fat ] below ) .
the same effect can also be seen in cases s90 and o90 ( figure [ fig : con_pattern]@xmath332 , @xmath13 ) .
the difference in the structure of the polar convection modes in cases s120 and o120 is further illustrated in figure [ fig : vn_so ] . here
it can be seen that the polar convection in case o120 is not only more isotropic and more decoupled from lower latitudes , but it is also more vigorous . we will return to this issue in section [ sec : flux ] .
we close this section with an orthographic view of case o120 from a point closer to the equatorial plane , highlighting its oblateness .
differential rotation established by convection is a common feature in stars .
the latitudinal angular velocity contrast on the surface , @xmath0 , is an important quantity both for interpreting stellar observations and for dynamo theories .
various observational techniques have been employed to measure @xmath0 , and to investigate the relationship between @xmath0 and the stellar angular velocity @xmath16 .
these observational techniques range from photometric variability studies @xcite to doppler imaging techniques @xcite and fourier transform methods @xcite .
global numerical simulations and mean - field models of convection in spherical shells have also been used to study that relationship @xcite .
specifically , @xmath0 is often defined as the difference in angular velocity between the equator and @xmath333 latitude , namely , @xmath334 the differential rotation profiles for all of the simulations considered in this paper are solar - like in the sense that the equator rotates more rapidly than the polar regions ( @xmath335 ) .
detailed rotation profiles will be discussed in section [ sec : mean ] . here
we focus on the dependence of @xmath0 ( computed using eq .
[ eq : omega ] below ) on @xmath326 , which is shown in figure [ fig : delta_omega ] . for the spherical simulations @xmath144-@xmath145 , @xmath0 is insensitive to @xmath326 , showing no clear trend .
this is somewhat of a surprise , since previous studies of convection in rotating spherical shells exhibit an increase of @xmath0 with @xmath326 ( e.g. * ? ? ?
. however , recall from section [ sec : setup ] that in this series of simulations @xmath14 and @xmath42 were kept constant as @xmath16 was increased .
by contrast , @xcite decreased @xmath14 and @xmath42 ( @xmath336 @xmath337 ) in order to maintain a relatively constant supercriticality .
this is the motivation behind our _ a _ series of simulations , represented in figure [ fig : delta_omega ] by the blue boxes .
these use the same scaling for @xmath14 and @xmath42 as in @xcite , and we find similar results
. in particular , @xmath0 monotonically increases with @xmath326 approximately as a power law ; @xmath338 with @xmath339 .
this is qualitatively consistent with the results of brown et al .
( see also * ? ? ?
* ) , though our exponent is somewhat steeper than their value of @xmath340 , valid for rossby numbers below about 0.4 .
these trends are generally consistent with observations , which suggest @xmath341 for solar - type stars @xcite
. however , this observed scaling relation only holds out to about @xmath342 , after which @xmath0 may remain constant or even decrease with @xmath326 , though observational evidence for this is still only preliminary @xcite .
this saturation and perhaps reversal of the @xmath0 vs @xmath326 scaling relation appears to be associated with a similar saturation of the magnetic field strength , @xmath122 , which increases with @xmath326 until about the same point , @xmath343 , beyond which @xmath122 becomes approximately independent of @xmath326 ( e.g. * ? ? ?
this suggests that the saturation of the @xmath0 vs @xmath326 relationship for rapidly - rotating solar like stars is likely due to strong lorentz - force feedbacks associated with vigorous dynamo action . however , our simulations suggest that oblateness may also contribute to the saturation of @xmath0 for very fast rotation rates .
we already noted in section [ sec : energetics ] that the drke / ke ratio saturates at about 0.61 for @xmath344 ( @xmath345 ) . this result
is also reflected in the @xmath0 plot of figure [ fig : delta_omega ] , which shows a peak value at @xmath346 ( @xmath347 ) , only for the oblate cases . at more rapid rotation rates
, the oblateness seems to supress the differential rotation .
this is likely due to the decoupling between polar and equatorial modes noted in section [ sec : patterns ] .
if banana cells reach beyond the tangent cylinder and into the polar regions as in case s120 , the associated reynolds stress can efficiently spin down the poles , enhancing the differential rotation .
however , if the polar and equatorial modes decouple due to oblateness as in case o120 , this can inhibit the angular momentum transport and diminish @xmath0 .
we will explore the mean flows in more detail in section [ sec : mean ] . however , before we proceed , one more comment is warranted in order to relate our simulations to previous work .
our numerical approach described in section [ sec : mp ] neglects the effect of differential rotation on oblateness .
is this justified ?
@xcite showed that differential rotation could significantly distort the shape of rapidly rotating stars compared to corresponding cases with uniform rotation .
however , in their model the reference rotation rate @xmath16 was allowed to be larger than the critical rotation rate @xmath10 and an anti - solar differential rotation profile was introduced to achieve stable solutions .
this extreme situation is likely to be rare in real stars .
furthermore , @xcite considered a full , 2d stellar structure model that included differential rotation down to the region of energy generation in the stellar core . given the sensitivity of energy generation to the thermal structure , even a slight oblateness of the core could have a substantial influence on the stellar luminosity .
if the differential rotation is solar - like in the sense that the equator spins faster than the pole and if it is confined to the convective envelope , then we expect the differential rotation to have a negligible influence on the oblateness .
this is demonstrated in appendix [ appen : a ] .
from a star s perspective , perhaps the most important function of convection is energy transport ; it carries the energy liberated by fusion in the core to the surface , where that energy is then radiated into space .
since this paper is the first to consider 3d simulations of convection in oblate stars , we believe that it is useful to consider the issue of outward energy transport with some care . a careful derivation is given in appendix [ appen : b ] but
the result is analogous to previous models of convection in spherical shells .
four principle components of the energy flux are involved in transporting energy in the outward direction , namely the enthalpy flux @xmath348 , the kinetic energy flux @xmath349 , the radiative flux @xmath350 and the diffusive entropy flux @xmath351 . in a statistically steady state
, these four fluxes together must account for the full luminosity imposed at the bottom boundary such that @xmath352 explicit expresstions for @xmath353 , @xmath354 , @xmath355 , and @xmath356 are given in appendix [ appen : b ] . for a spherical shell @xmath357 but
this is no longer true for an oblate shell where we compute @xmath358 numerically .
figure [ fig : flux_balance-60 ] shows the flux balance through equipotential surfaces for cases @xmath144 and @xmath146 .
it is apparent that the energy flux distribution in case @xmath144 is very similar to that in case @xmath146 , which is expected because of the small oblateness . by design , the radiative flux @xmath350 carries all of the imposed flux at the bottom where the normal velocity vanishes , as do @xmath354 and @xmath353 . the radiative flux @xmath350 also dominates the heat transport in the lower convective envelope while the entropy flux @xmath351 carries energy through the top boundary and the upper convective envelope .
the strong correlations between normal velocities and temperature fluctuations yield the enthalpy flux @xmath348 , which gradually increases towards the top from the bottom until it peaks near @xmath359 .
it then drops down to zero rapidly as it approaches the impenetrable top boundary .
as seen in figure [ fig : flux_balance-120 ] , the difference in the energy flux distribution between cases @xmath145 and @xmath147 is more significant .
this is especially true for the enthalpy flux @xmath348 and the entropy flux @xmath351 . in case @xmath145
, @xmath348 reaches a maximum value of 0.32 around @xmath360 while the peak @xmath353 value in case @xmath147 is only 0.25 , achieved around @xmath361 .
this suggests that the oblateness is slightly inhibiting the net convective energy transport through the shell .
the normalized sum of the four principle flux components in all of our simulations is close to one throughout the shell , signifying that they have achieved an equilibrated state .
it is well known that rapid rotation also tends to inhibit convection @xcite .
this can be seen in our simulations through its effect on the mean entropy stratification .
we find that the convection is less able to establish an adiabatic profile throughout the convection zone in the rapidly rotating simulations .
instead , much of the convection zone remains slightly superadiabatic ( @xmath362 ) .
the radial profile of the normal entropy gradient @xmath363 for our simulations is shown in figure [ fig : dsdr ] , highlighting the dependence on the rotation rate .
for the spherical simulations ( fig .
[ fig : dsdr]a ) , the normal entropy gradient exhibits a greater departure from the adiabatic profile as the rotation rate is increased from @xmath364 to @xmath365 .
this signifies a decrease in the efficiency of the convective energy transport for two reasons .
first , the convection is less able to mix entropy to establish an adiabatic stratification .
second , the reduced enthalpy flux ( @xmath353 ) in the upper convection zone requires a steeper entropy gradient ( larger @xmath356 ) in order to satisfy the flux balance equation ( [ eq : fbalance ] ) .
a similar trend is observed in the upper convection zone for the oblate simulations ( fig .
[ fig : dsdr]b ) . here
again , the magnitude of the superadiabatic entropy gradient increases with increasing @xmath326 .
however , this trend reverses near the outer boundary as highlighted in the inset .
this can be attributed to the larger surface area of the oblate spheroids . in order for each simulation to achieve equilibrium ,
the total flux through the outer boundary must equal the total flux imposed through the inner boundary , which is in turn equal to the luminosity , @xmath48 . in our simulations ,
the flux through the outer boundary is carried by the entropy flux @xmath356 , which is proportional to the normal entropy gradient , as expressed in equation ( [ eqn : fu ] ) .
thus in order to achieve equilibrium , @xmath356 must be equal to @xmath366 at @xmath104 , where @xmath367 is the area of the outer surface .
thus , for a given @xmath48 , the @xmath356 is reduced as @xmath367 is increased . in other words , the total luminosity is spread over a larger area in oblate stars so the flux through any fixed area on the surface is reduced . in order to further investigate the effect of oblateness on the entropy stratification , we compare four cases in figure [ fig : dsdr_so ] . as expected
, there is little difference between cases @xmath144 and @xmath146 since the oblateness is small ( see table [ tab : parameters ] ) .
however , the normal entropy gradient in case @xmath147 is significantly smaller in magnitude than in case @xmath145 .
several factors contribute to this result .
one is the larger surface area , which dilutes @xmath356 in the upper convection zone as noted in the previous paragraph .
another is the somewhat less efficient convective transport , as noted above in the discussion of figure [ fig : flux_balance-120 ] .
yet another is the larger depth of the convection zone , which tends to weaken thermal gradients in general .
for example , consider the normal entropy flux at the equator where @xmath368 is parallel to the radial direction .
there @xmath369 $ ] , where @xmath370 and @xmath371 are the pressure and density scale heights
. the weaker effective gravity due to the centrifugal force tends to increase both @xmath370 and @xmath371 , which decreases @xmath372 .
viewed another way , a given entropy difference @xmath373 is stretched out over a wider radial distance at low latitudes in an oblate star because of the relatively large spacing between the potential surfaces . as mentioned in section [ sec : intro ] , one of the major theoretical results associated with rapidly - rotating , oblate stars is the concept of gravity darkening .
* this was originally predicted by @xcite for the particular case of massive stars with radiative envelopes . in this case the surface flux is dominated by the radiative flux @xmath374 . if @xmath27 is constant on equipotential surfaces , @xmath375 ( see [ sec : ic ] for the case of adiabatic stratification ) , then @xmath376 , where @xmath3 is the effective gravity ( see section [ sec : intro ] ) . since @xmath3 is reduced at the equator
, the von zeipel model predicts that the equatorial regions should possess less energy flux than the polar regions and should therefore appear darker .
one problem with this result is that the radiative envelopes of oblate stars are unlikely to be barotropic ( @xmath377 ) , as emphasized by @xcite .
still , the qualitative prediction of gravity darkening has indeed been verified observationally for massive stars , though the quantitative contrast between equator and pole is less than predicted @xcite .
* however , the von zeipel model does not apply to late - type stars with convective envelopes . in these stars ,
the surface flux is not dominated by @xmath355 so the line of reasoning that gives rise to gravity darkening should be regarded as questionable .
several attempts have been made to extend the von zeipel model to late - type stars but they all involve assumptions about the nature of the convective heat transport . @xcite
used mixing - length models of convection to predict a result that resembles gravity darkening but with a weaker latitudinal dependence , such that the emergent flux is @xmath378 .
however these mixing - length models are local in nature and do not take into account coherent structures such as banana cells that are known to dominate convective heat transport in global simulations of convection in rotating spherical ( or spheriodal ) geometries . @xcite and
* @xcite * also presented a model that in principle should apply to convective envelopes as well as radiative envelopes .
however , this model assumes that the convective heat flux is directed normal to @xmath45 surfaces , which may or may not be true .
global convection simulations often exhibit a significant latitudinal heat flux that can have a substantial influence on the flux that eventually emerges through the outer boundary ( e.g. * ? ? ?
* ) . in this paper , we are the first to address the issue of gravity darkening from the perspective of global , 3d simulations of convection in oblate stars .
as noted in section [ sec : flux ] , the convective enthalpy flux @xmath348 in our simulations peaks in the upper convection zone ( see fig . [
fig : flux_balance-120 ] ) . both warm upflows and cool downflows serve to transport flux out of the star . however , this enthalpy flux is not uniform in latitude , as demonstrated in figure [ fig : fe_so ] which shows the zonally - averaged @xmath353 in cases @xmath145 and @xmath147 .
this shows efficient heat transport through the convective envelope near the equatorial region in both cases .
this is attributed to the banana cells that arise because of the influence of rotation on convection .
the alignment of convective banana cells with the rotation axis allows for more efficient heat transport .
we also find that more enthalpy flux flows through the polar regions in case @xmath147 than in case @xmath145 .
this difference arises because of the thinner depth of the convection zone at high latitudes , which tends to increase the radial temperature and entropy gradient as discussed in section [ sec : flux ] .
this in turn increases the buoyant driving , making the convection more vigorous .
the greater vigor of polar convection in case @xmath147 is likely also associated with the decoupling of the polar and equatorial convection modes as discussed in section [ sec : patterns ] .
the prominent vertical striping of the polar enthalpy flux apparent in figure [ fig : fe_so]@xmath331 is associated with strong mean flows as discussed in section [ sec : mean ] .
the enthalpy flux is within the star but only the heat flux coming out at the surface can be observed . in our model
, the heat flux passing through the outer surface of the star is carried by the diffusive entropy flux , @xmath356 .
thus , this is the quantity to consider in order to assess the latitudinal dependence of the emergent flux , within the context of gravity darkening .
figure [ fig : fu_so ] shows the comparison of latitudinal entropy flux @xmath351 at the outer surface simulated in cases @xmath145 and @xmath147 . for comparison ,
we have also plotted the profiles predicted by @xcite and @xcite in figure [ fig : fu_so ] .
there are two noteworthy results to be gleaned from figure [ fig : fu_so ] .
first , a comparison between cases @xmath145 and @xmath147 clearly shows an enhanced polar heat flux in the latter case . the oblateness will tend to make the poles brighter , as in the standard gravity darkening paradigm @xcite .
however , in stark contrast to the standard paradigm , case @xmath147 exhibits a secondary peak in the emergent heat flux at the equator .
this is the second noteworthy result and , unlike the first , is somewhat unexpected .
though the emergent flux is carried by @xmath356 , both of these results can be ultimately attributed to the nature of the deeper enthalpy flux @xmath353 illustrated in figure [ fig : fe_so ] .
since @xmath353 must vanish at the surface due to the impenetrable boundary condition , there must be a convergence of the outward enthalpy flux just below the surface , which preferentially heats the fluid in the polar and equatorial regions .
since the temperature is fixed at the outer surface , this heating increases the normal temperature and entropy gradients .
in other words , the enhanced enthalpy flux @xmath353 in the polar and equatorial regions is converted into an emergent diffusive entropy flux @xmath356 .
heat transport by global convective motions under the influence of rotation and stratification is not taken into account in the analysis of @xcite and @xcite . only global , 3d convection simulations
can explicitly capture these effects .
ours are the first such simulations and they are calling into question common assumptions . for example , since the imposed heat flux on the bottom boundary is uniform , a non - uniform heat flux across any equipotential surface @xmath45 necessarily implies a nonzero heat flux along @xmath45 surfaces somewhere deeper in the convection zone , provided that the simulation is equilibrated .
this violates the assumptions made by both @xcite and @xcite .
our results imply that rapidly - rotating , oblate , lower - mass stars with convective envelopes might not have darker equators after all .
however , there is a caveat to this conclusion that has to do with the parameter regimes that we have considered in our simulations .
it is well known that banana cells are the linearly preferred instability modes for convection in rapidly - rotating spherical shells @xcite .
thus , as the rayleigh number ra is increased beyond the critical value for convection to set in , the equatorial modes are excited first , causing the convective enthalpy flux to peak at the equator . as ra is increased further , the polar modes are excited .
thus , the energy flux profiles for cases @xmath145 and @xmath147 shown in figures [ fig : fe_so ] and [ fig : fu_so ] , featuring prominent minima at mid latitudes , are commonly seen in global spherical convection simulatons at moderate rayleigh numbers ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . at the much higher rayleigh numbers characteristic of real stars
, these mid - latitude minima may fill in at least partially , making the emergent heat flux more uniform .
thus , we may be overestimating the magnitude of the latitudinal variation . still , the enhanced brightness at the poles and equator are real physical effects , grounded in the physics of rotating convection .
these are the regions where convective structures can align with the rotation axis , enhancing the efficiency of the convective energy transport .
the simulations of @xcite in particular demonstrate another interesting effect ; the suppression of the equatorial heat transport by strong differential rotation .
this occurs at high rayleigh numbers and moderate rotational influence ( ro @xmath379 ) , where the relative differential rotation @xmath380 is largest .
the resulting differential rotation peaks strongly at the poles .
if this were to occur in oblate stars , it could enhance the gravity darkening , making the equator relatively dim , as in the classical picture of von zeipel , though for different reasons .
however , stars that are rotating rapidly enough to be oblate are likely to possess low rossby numbers ( ro @xmath152 ) , where the yadav et al simulations exhibit an enhanced heat flux at the equator .
furthermore , oblateness and magnetism may prevent the rotational shear from becoming strong enough to suppress the equatorial heat transport ; dynamo simulations by yadav et al . with the same parameters exhibit a mid - latitude minimum in the heat flux as described in the previous paragraph .
more generally , rapidly - rotating stars are known to exhibit vigorous dynamo action which could influence the results reported here .
in addition to altering the convective heat transport , magnetism can influence the latitudinal dependence of the heat flux through polar starspots and faculae .
another relevant phenomenon that we have neglected is small - scale surface convection .
the steep superadiabatic entropy and density stratification in the surface layers of late - type stars drives small - scale convective motions that can not be resolved in global models .
familiar examples include solar granulation and supergranulation
. these will tend to homogenize the outgoing heat flux , possibly mitigating any latitudinal variations that may be established deeper in the convection zone .
further research is needed to establish how all of these effects contribute to the emergent flux in oblate , late - type stars .
as mentioned previously , the results presented here should be regarded as a baseline for further study and certainly not the final word .
some aspects of the differential rotation were discussed in section [ sec : dr1 ] , within the context of stellar observations , previous global convection simulations and previous models of oblate stars . here
we look at the mean flows ( differential rotation and meridional circulation ) in more detail , considering the influence of oblateness on their spatial structure .
we define mean flows as averages of the velocity field @xmath29 or the mass flux @xmath381 over longitude .
longitudinal averages are denoted by angular brackets , e.g. @xmath382 .
the differential rotation ( in nhz ) is expressed in terms of the angular velocity as follows : @xmath383 we define thermal variations @xmath384 and @xmath385 by averaging over longitude and time and then subtracting the spherically - symmetric component in order to highlight variations relative to the mean stratification .
the mean flows for cases @xmath147 and cases @xmath145 are shown in figure [ fig : mc_120 ] . as mentioned in section [ sec : dr1 ] , the simulations considered here all have a solar - like differential rotation profile , with relatively fast rotation at the equator that decreases toward higher latitudes ( fig .
[ fig : mc_120]@xmath13,@xmath386 ) .
this is a common feature of convection simulations at low rossby number and is established by the convective angular momentum transport associated with banana cells ( e.g. * ? ? ?
* ; * ? ? ?
these rotation profiles are generally in thermal wind balance , maintained in part by baroclinic torques that offset the tendency for @xmath387 contours to align with the rotation axis in accordance with the taylor - proudman theorem .
it is this thermal wind effect that is responsible for the poleward gradients that dominate the temperature and entropy profiles in figure [ fig : mc_120 ] , frames @xmath331 , @xmath332 , @xmath388 , and @xmath389 . however , superposed on this monotonic decrease of @xmath387 from equator to pole , case @xmath147 exhibits a striking series of alternating zonal jets in both polar regions ( fig .
[ fig : mc_120]@xmath13 ) .
these are associated with a similar series of strong meridional circulation cells that can be seen prominently in the latitudinal mass flux plot of figure [ fig : mc_120]@xmath162 .
though the zonal jets are confined to the upper convection zone , the meridional cells extend throughout the convection zone , with a significant poleward tilt from bottom to top .
similar features are also present in cases @xmath187 and @xmath182 ( not shown ) . the faint thermal signature in fig .
[ fig : mc_120]@xmath331,@xmath332 suggests that these are axisymmetric convection cells , with the zonal jets arising from the coriolis - induced deflection of meridional flows toward and away from the rotation axis near the outer surface .
this thermal signature is also seen in the enthalpy flux of figure [ fig : fe_so]@xmath331 and in the emergent flux of figure [ fig : fu_so ] ( as the high - latitude wiggles ) .
the presence of these convective cells near the polar regions in case @xmath147 is likely due in part to the larger entropy gradient since the polar convection zone depth is relatively small compared to the equatorial depth ( see section [ sec : flux ] ) . in any case
, they appear to be induced by the oblateness . at lower latitudes , outside the tangent cylinder , the meridional flow in both cases @xmath145 and @xmath147 exhibits a multiple - cell structure characterized by alternating regions of northward and southward mass flux ( fig . [
fig : mc_120]@xmath162,@xmath390 ) .
this too is a common feature in global convection simulations and is attributed to the nature of the convective reynolds stress induced by banana cells @xcite . to study the temporal evolution of the high - latitude zonal jets in case @xmath147 , we have plotted in figure [ fig : df_o120 ] a latitude - time map of @xmath387 in the vicinity of the north pole ( latitudes @xmath391 ) .
recall that the rotation period in this case is about 0.23 days ( a factor of 120 less than the sun ) , so the time period covered in the plot spans over 100 rotation periods ( and over 20 convective turnover times ) .
some of the zonal jets persist for this entire interval , though they meander in latitude .
each jet is about 4 - 5@xmath392 wide .
we have preformed the first global 3d hydrodynamic simulations of convection in the oblate envelopes of rapidly - rotating solar - type stars using the recently introduced compressible high - order unstructured spectral difference ( chorus ) code ( wlm15 ) .
the chorus code employs unstructured grids , giving it the capability to deal with oblate spheroidal shell geometries .
this unique tool allows us to investigate the influence of the oblateness on the convection and mean flows , including heat transport , differential rotation , and the thermal structure of the convective envelope .
we have found that the influence of rotation and oblateness leads to an enhancement of the emerging heat flux in the polar and equatorial regions .
this is in stark contrast to the conventional concept of gravity darkening and has important observational implications ; oblate , low - mass stars with convective envelopes may not exhibit gravity darkening in the same way that higher - mass stars do . in particular , the minimum suface brightness might occur at mid - latitudes rather than at the equator .
the enhancement of the polar heat flux due to oblateness arises for essentially the same reason as in classical gravity darkening arguments ; the stronger effective gravity relative to lower latitudes enhances thermal gradients ( see section [ sec : flux ] ) .
this makes the convection more vigorous .
however , the secondary peak in the emergent heat flux at the equator arises from physical effects that are not taken into account in the classical models of @xcite , @xcite , and @xcite .
oblate stars are rapidly - rotating by nature , implying strong rotational constraints on the convective motions ( low rossby numbers ) . in these parameter regimes ,
the convection in the equatorial region ( outside the tangent cylinder ) is dominated by banana cells , columnar convective modes aligned with the rotation axis .
these banana cells can transport heat more efficiently than at mid - latitudes . in short , the convective heat transport peaks at the poles and at the equator because this is where the convective structures can most easily organize themselves to mitigate the strong rotational constraints : vertical plumes at high latitudes and banana cells at low latitudes . furthermore , though banana cells are essentially laminar flow features , this fundamental argument also extends to more turbulent parameter regimes ( higher rayleigh and reynolds numbers ) .
for example , in the high - resolution boussinesq simulations of @xcite , the convective columns break up into vortex sheets as the diffusion is decreased .
however , these vortex sheets are still aligned with the rotation axis outside the tangent cylinder and would still be expected to yield an enhancement of the equatorial heat flux .
thus , we expect that the general latitudinal dependence of the emergent heat flux in oblate solar - type stars to be similar to that found here , with maxima at the poles and equator . however ,
as discussed in section [ sec : darkening ] , we may be overestimating the magnitude of the variation , which here is on the order of 40% ( see figure [ fig : fu_so ] ) .
the latitudinal dependence of the emergent heat flux is a symptom of the more general tendency for convection in rapidly - rotating spherical or spheroidal shells to separate into distinct polar and equatorial modes , inside and outside the tangent cylinder .
we find that the oblateness enhances this dichotomy , causing the polar regions to largely decouple from the equatorial regions ( figs .
[ fig : con_pattern ] , [ fig : vn_so ] ) . however , other aspects of the convective structure such as the power spectrum and the width of banana cells , are insensitive to the oblateness .
another notable result from our simulations is the presence of banded zonal jets at high latitudes ( section [ sec : mean ] ) .
remarkably , these only occur in the oblate simulations , so they are clearly induced by the oblateness .
they appear to be due to narrow , axisymmetric convective modes that are approximately aligned with the rotation axis but significantly tilted ( fig .
[ fig : mc_120 ] ) .
each jet is about 4 - 5@xmath392 wide and many persist for over 100 rotation periods , though they meander in latitude ( figure [ fig : df_o120 ] ) .
it is unclear whether or not these jets would persist in the presence of magnetic fields .
all stars are ionized and , given the extreme parameter regimes , all stellar convection zones are expected to undergo vigorous dynamo action .
dynamo - generated magnetic fields might suppress the polar jets directly by means of magnetic tension or indirectly by inhibiting the intrinsic self - organization processes of rapidly - rotating hydrodynamic turbulence ( e.g. the inverse cascade of kinetic energy and the anisotropic suppression of this cascade that gives rise to the rhines scale ) .
however , these jets do not appear to arise from the inverse cascade of smaller - scale vorticies . whether or not they occur in the presence of magnetism
is a topic for future work . though chorus does not currently have the capability to solve the full set of mhd equations , we intend to implement this capability in the future .
the amplitude of the differential rotation is not very sensitive to either the rotation rate or the oblateness .
if we decrease @xmath14 and @xmath42 as we increase the rotation rate in order to mainin a roughly constant supercriticality , we find that the differential rotation scales as @xmath393 .
this result is for spherical ( non - oblate ) simulations and is qualitatively consistent with previous global convection simulations and stellar observations ( section [ sec : dr1 ] ) . alternatively ,
if @xmath14 and @xmath42 are held constant , we find no clear scaling of @xmath0 versus @xmath326 ( fig . [ fig : delta_omega ] ) .
however , the fraction of the total kinetic energy contained in the differential rotation , drke / ke , does increase monotonically in these same spherical simulations as @xmath326 is increased from 0.526 in case @xmath144 to 0.681 to case @xmath145 ( table [ tab : diagnostics ] ) .
it is interesting that the oblateness tends to suppress this increase of drke / ke with @xmath326 .
for the oblate series of simulations , @xmath146 through @xmath147 , the drke / ke ratio saturates at a maximum value of 0.61 for @xmath344 ( table [ tab : diagnostics ] ) . this trend is also visible in the surface differential rotation @xmath0 , which decreases slightly beyond about @xmath394 ( fig .
[ fig : delta_omega ] ) .
the fractional differential rotation @xmath395 decreases with increasing rotation for both the spherical and the oblate stars .
furthermore , we argue that the differential rotation is likely to be too weak to have a significant effect on the oblateness ( appendix [ appen : a ] ) .
we acknowledge that our simulations certainly have their limitations , the most important being the relatively laminar parameter regimes and the absence of magnetic fields .
however , as the first simulations of their kind , we regard these as a baseline for future work . we have identified interesting features that warrant further study , such as a departure from the classical gravity darkening paradigm for late - type stars and the presence of polar jets induced by oblateness . and , we have demonstrated some of the potential of the new chorus code , which promises to be a unique and powerful tool for the ongoing investigation of stellar and planetary convection .
we thank matthias rempel and the anonymous referee for helpful comments on the manuscript .
junfeng wang is funded by a newkirk graduate fellowship from the national center for atmospheric center ( ncar ) .
chunlei liang is grateful for the faculty start - up fund provided by the george washington university and a recent career award ( award number 1554005 ) from the national science foundation .
ncar is operated by the university of corporation for atmospheric research under sponsorship of the national science foundation .
a fully self - consistent simulation would properly capture the nonlinear interaction between the convection , the differential rotation , and the oblateness , allowing the star to deform in response to the differential rotation that the convection establishes .
however , this poses significant computational challenges and we show in this appendix that even if we were to overcome these challenges , it would be unlikely to change our results . in this appendix
we investigate the effect of differential rotation on the oblateness of rotating stars in a simple model . in this model , the desired oblate spheroidal shell
is obtained by deforming a spherical shell with the uniformly rotating rate @xmath16 while the mass and angular momentum in the @xmath56 direction is kept unchanged .
we neglect convection and meridional circulation in our analysis ; the only motion is the differential rotation .
furthermore , to make the problem more analytically tractable , we assume that the differential rotation follows a conservative rotation law , @xmath396 where @xmath397 is the cylindrical radius , so the centrifugal force can be represented as the gradient of a potential . the rotation profile in the oblate spheroidal shell is assumed to be a linear function of @xmath398 ; @xmath399 where @xmath400 is the outer radius of the original spherical shell , @xmath401 is the fractional differential rotation and @xmath402 is a coefficient that is adjusted to conserve angular momentum as described below .
thus , the effective gravitational potential is @xmath403 we then compute the structure of the oblate spheroidal shell as described in section [ sec : mp ] , using this potential instead of equation ( [ eqn : potential ] ) . during the deformation from the spherical shell to the oblate spheroidal shell with the desired linear rotation profile ,
the mass within the domain is kept unchanged , @xmath404 where @xmath405 is inner radius for the spherical shell and @xmath406 and @xmath407 are the inner and outer radius for the oblate spheroidal shell .
meanwhile , the angular momentum in the @xmath56 direction is also conserved during the transformation , @xmath408 note that the density profile is as described in section [ sec : ic ] and the integrals ( [ eqn : mass_def_con ] ) and ( [ eqn : ang_def_con ] ) are computed numerically , on a grid as described in section [ sec : der_grid ] . by solving equations ( [ eqn : mass_def_con ] ) and ( [ eqn : ang_def_con ] ) , @xmath402 is obtained if @xmath401 is specified .
we vary the amplitude of the differential rotation by varying @xmath401 and , for each value of @xmath401 , we compute the oblateness . the results of this analysis are listed in table [ tab : oblateness ] for @xmath409 rad s@xmath96 . in all cases @xmath410 because the shell is required to have the same mass and angular momentum as a spherical shell with the same rotation rate .
for example , for no differential rotation ( @xmath411 ) , the larger moment of inertia associated with the oblate shell requires @xmath412 rad s@xmath96 = 105 @xmath413 to achieve the same angular momentum as a spherical shell rotating at a rate of @xmath414 . for @xmath415
, @xmath387 varies from @xmath416 at the rotation axis ( @xmath417 ) to @xmath418 at the equator ( @xmath419 ) .
averaging the minimum and maximum values of @xmath387 then gives us a fiducial rotation rate of @xmath420 .
the fractional differential rotation can then be defined as @xmath421 . for @xmath422
this gives a fractional differential rotation of 0.5 .
thus , according to table [ tab : oblateness ] , varying the fractional differential rotation from 0 ( @xmath411 ) to 0.5 ( @xmath422 ) only changes the oblateness by about 0.22% .
this can be compared with the fractional differential rotation @xmath395 achieved in our simulations , as shown in figure [ fig : delta_omega ] . as shown there
, all cases have a @xmath0 of less than @xmath423 rad s@xmath96 .
this corresponds to fractional differential rotation of about 0.10 for @xmath424 and a fractional differential rotation of about 0.05 for @xmath414 .
thus , the amplitude of the differential rotation achieved in our simulations is well within the range spanned in table [ tab : oblateness ] .
though the differential rotation profiles in our simulations are not precisely linear as expressed in equation ( [ eqn : ro_ln ] ) , these results suggest that the influence of the differential rotation on the oblateness should be negligible .
lcccccc @xmath402 & @xmath425 & @xmath426 & @xmath427 & @xmath428 & @xmath429 & @xmath430 + @xmath123 & @xmath431% & @xmath432% & @xmath433% & @xmath434% & @xmath435% & @xmath436% [ tab : oblateness ]
in this appendix we derive the equation of flux balance in an oblate star ( [ eq : fbalance ] ) and give explicit expressions for @xmath353 , @xmath354 , @xmath355 , and @xmath356 .
we begin by integrating equation ( [ eqn : energy ] ) over a volume @xmath437 that is bounded on the bottom by the lower boundary of the computational domain , defined by @xmath103 , and on the top by an arbirtary geopotential within the computational domain , @xmath438 .
this yields @xmath439 { \mbox{\boldmath $ \cdot$}}d\mathbf{a } + \int_{v_s } \rho \mathbf{u } { \mbox{\boldmath $ \cdot$}}{\nabla}\phi dv ~~~.\ ] ] in deriving equation ( [ eq : evs ] ) we have used stokes theorem to express the volume integral of the flux divergences as a surface integral and we have expressed the effective gravity as the gradient of the roche potential , eq .
( [ eqn : potential ] ) .
the area @xmath109 in the surface integral includes both the inner boundary at @xmath103 , which we call surface @xmath440 , and the upper boundary of @xmath441 at @xmath438 , which we call surface @xmath442 .
we proceed by assuming that the flow is in a statistically steady state .
this implies that the time derivative on the left - hand - side of eq .
( [ eq : evs ] ) can be neglected .
together with equation ( [ eqn : mass_conservation ] ) , it also implies that the mass flux is approximately divergence - free , @xmath443 .
the divergence - free nature of the mass flux also follows from the low mach number and nearly adiabatic , hydrostatic background stratification , which forms the basis for the anelastic approximation ( e.g. * ? ? ?
this allows us to write the last term in eq .
( [ eq : evs ] ) as @xmath444 the final equality follows from the impenetrable lower boundary condition , @xmath71 on surface @xmath440 , and the divergence - free nature of the mass flux .
note in particular that @xmath445 thus , the buoyant and centrifugal acceleration do not change the total energy contained within the volume @xmath441 .
note also that , since @xmath71 on surface @xmath440 , the only contribution to the integrated flux on that surface is from the diffusive fluxes @xmath37 .
equation ( [ eq : fbot ] ) then implies @xmath446 { \mbox{\boldmath $ \cdot$}}d\mathbf{a } = l ~~~.\ ] ] thus , in what follows we need only concern ourselves with the energy flux across surface @xmath442 .
now consider the integrated enthalpy and kinetic energy flux , which can be written as @xmath447 \mathbf{u } { \mbox{\boldmath $ \cdot$}}d{\mbox{\boldmath $ { \cal s}$}}= \int_{\cal s } \left[\frac{\rho}{2 } u^2 + c_p ~ \rho t \right ] ~ \mathbf{u } { \mbox{\boldmath $ \cdot$}}d{\mbox{\boldmath $ { \cal s}$}}= ( f_k + f_e ) a_{\cal s } ~~~,\ ] ] where @xmath448 and @xmath449 the overbars on @xmath450 and @xmath451 denote averages over the surface @xmath442 and @xmath358 denotes the total area of the surface @xmath442 . in this derivation
we have used the expressions @xmath452 , @xmath453 , and the ideal gas law @xmath454 , where @xmath455 and @xmath456 are the specific heat at constant pressure and volume respectively .
we have also made use of the expression @xmath443 which , as noted above , implies that there is no net mass flux through the surface @xmath442 .
so , @xmath457 . in equations ( [ eq : fe ] ) and ( [ eq : fk ] ) we have also assumed that @xmath458 so that the results can be easily compared with anelastic simulations ( see wlm15 ) .
the thermal flux terms arising from radiative and entropy diffusion follow straightforwardly from the expressions for @xmath37 given in section [ sec : hydro ] : @xmath459 and @xmath460 we neglect the viscous flux @xmath461 which is generally small in global stellar convection simulations ( e.g. * ? ? ?
* ; * ? ? ?
* ) .
saar , s. h. 2009 , in solar - stellar dynamos as revealed by helio- and asteroseismology : gong 2008/soho 21 , ed .
m. dikpati , t. arentoft , i. g. hernandez , c. lindsey , & f. hill , asp conference series , vol . 416 ( san francisco , ca : asp ) , 375384
|
we present the first global 3d simulations of thermal convection in the oblate envelopes of rapidly - rotating solar - type stars .
this has been achieved by exploiting the capabilities of the new compressible high - order unstructured spectral difference ( chorus ) code .
we consider rotation rates up to 85% of the critical ( breakup ) rotation rate , which yields an equatorial radius that is up to 17% larger than the polar radius .
this substantial oblateness enhances the disparity between polar and equatorial modes of convection .
we find that the convection redistributes the heat flux emitted from the outer surface , leading to an enhancement of the heat flux in the polar and equatorial regions .
this finding implies that lower - mass stars with convective envelopes may not have darker equators as predicted by classical gravity darkening arguments .
the vigorous high - latitude convection also establishes elongated axisymmetric circulation cells and zonal jets in the polar regions .
though the overall amplitude of the surface differential rotation , @xmath0 , is insensitive to the oblateness , the oblateness does limit the fractional kinetic energy contained in the differential rotation to no more than 61% .
furthermore , we argue that this level of differential rotation is not enough to have a significant impact on the oblateness of the star .
|
observations are a central source of knowledge on almost any entity in astrophysics . over several centuries of intense research
, several principal observational techniques have been developed that are now routinely used to study stars and stellar populations in the milky way and other galaxies .
we have information from astrometry , photometry , spectroscopy , but also interferometry , and astroseismology , that give complementary information on the physical parameters of stars ( detailed chemical composition , gravities , temperatures , masses and ages ) and their kinematics ( radial velocities , distances , and orbital characteristics ) .
however , in contrast to e.g. cosmology , where sophisticated bayesian schemes are well established ( e.g. * ? ? ?
* ; * ? ? ?
* ) , stellar parameter determinations are still widely based on best - fit estimates and simple averages between different methods . the advent of large stellar spectroscopic and photometric surveys like segue / sdss @xcite , rave @xcite , apogee @xcite , gcs @xcite , and the gaia - eso survey @xcite , as well as astroseismic surveys like kepler @xcite , makes it necessary to develop fully automated methods for data analysis and determination of stellar parameters .
standard spectroscopic inversion methods are commonly assumed to be accurate , however , they usually involve subjective and hardly reproducible elements , like line fitting and normalisation , or decisions on spectral diagnostic features .
manual analysis of stars is limited to sample sizes of @xmath0 stars , unsuitable for large surveys .
existing automated methods usually suffer from weakly constrained systematics as well as idealised error estimates .
so far , most attempts to overcome these problems have concentrated on simple weighted averaging between different methods ( e.g. * ? ? ?
* ; * ? ? ?
the large stellar surveys change stellar astronomy into a precision science , where we can not limit ourselves to pointing out structures in diagrams , but where knowledge of the error distributions is key to make meaningful model comparisons , e.g. of galactic evolution and stellar structure .
the approach we need must be flexible , objective , applicable to very large datasets , and provide an optimal combination of the different bodies of observational data . the only mathematical apparatus known to permit a systematic combination of different quantities are bayesian frameworks .
the first steps in this direction were made by @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , and @xcite .
the scope and applicability of these studies is limited : they either addressed the problem of fitting a spectrum only @xcite , partly focussed on the problem of finding the maximum likelihood solution , or rely on simplifications of the observational likelihoods ( e.g. * ? ? ?
* ; * ? ? ?
* ) . in part
, this problem appears rooted in the introduction of `` observables '' , like effective temperature , ( see e.g. fig . 4 in * ? ? ?
* ) , which have no well - defined place in a bayesian approach and which are in fact just parameters constrained by another observation . in contrast , a bayesian scheme can only fulfil its claim of unbiased information , if a fair account of the observations is given and the full dimensionality of the constraints in parameter space is preserved . in the following
we will present a new method for the determination of stellar parameters that provides an optimal exploitation of different observational information .
the method offers a homogeneous full - scale quantitative recovery of the full probability distributions in parameter space , which are given by the available observations , i.e. photometry , astrometry , spectroscopy , and well - established knowledge from stellar evolution theory and galaxy structure .
the method is objective , computationally efficient , can be readily applied to data from all existing surveys and is robust to missing bits of data , e.g. damaged pixels in a spectrum or low - quality photometry . by embedding spectroscopic analysis directly in scheme ,
the bayesian method allows for consideration of all pieces of relevant information at once , thus avoiding unnecessary information loss . in this first paper of the series ,
our main goal is to to determine effective temperature , surface gravity , metallicity , mass , age and distances of individual stars .
thus , we limit the input data to spectroscopy , photometry , stellar evolution models and facultative parallax measurements .
however , the method can be readily generalised to any number of parameters , such as kinematics or stellar rotation , and include other input information , e.g. astero - seismology and interferometric angular diameters .
furthermore , it is straightforward to analyse star formation history of a whole stellar population , e.g. a young cluster or an old galaxy , using its integrated colours and spectra .
thus the bayesian method has a very broad scope to applications both in the context of galactic and extra - galactic research .
the paper is structured as follows . in sections 2 and 3 , we present the details of the algorithm and its implementation , illustrated on two examples . in section 4
we apply the method to a sample of stars with very high - resolution observations and for a sub - sample of calibration stars from the sdss / segue catalogue .
section 5 compares to the use of a simplified spectroscopic pdf .
discussion of the algorithm and results and conclusions are found in the last two sections .
so far , the majority of observational studies of stars , be it photometric or spectroscopic , have focussed on providing best - fit estimates of stellar parameters
. however , accurate comparisons to theoretical models of e.g , galaxy evolution , require the full probability distribution of the derived parameters given the available observations .
this demands a bayesian formalism . in this context
we need to express the probability of a set of parameters @xmath1 given a set of observations @xmath2 by the probability that this observation could take place given the set of parameters . by definition the conditional probability @xmath3 , that @xmath4 given @xmath5 , derives from the combined probability @xmath6 as : @xmath7 .
we can hence write down : @xmath8 where the _ posterior _ probability @xmath9 is the conditional probability of the parameter set @xmath4 given @xmath5 .
@xmath10 , which we call _ observed likelihood _ , is the probability of making the set of observations @xmath5 given the set of parameters @xmath4 and @xmath11 is the prior probability we ascribe to that set of parameters .
@xmath12 is the probability that the set of observations was made , which we set to @xmath13 @xcite .
this simplifies our problem to @xmath14 where @xmath9 is the posterior probability distribution function ( pdf ) on the chosen parameter space . in our work , observations are conditionally independent given the parameters , i.e. if all parameters are perfectly known , the observations do not provide additional information about each other . hence we can disentangle the observations by : @xmath15 the parameter set @xmath4 contains all parameters relevant to the problem under investigation and important to the description of a star .
this may include surface and interior structure parameters ( effective temperature , surface gravity , mean density , etc ) as well as any other pieces of information like chemical composition , age , distance , position in the sky , etc .
since we are dealing with a single object , all these parameters are related in some way .
however , we can break their dependencies into main groups , using the fact that each type of observations constraints only a sub - set of these parameters , whereas it bears no information on others . in this work
, we define the core parameter space @xmath16 of metallicity ( expressed by iron abundance ) , effective temperature and surface gravity .
the parameters in @xmath17 impact all our observations and models .
other parameters are constrained by only a subset of observations : e.g. detailed abundances are of importance for spectroscopic observations , while stellar magnitudes in different colour bands @xmath18 span the space of the photometric parameters .
age @xmath19 , initial mass @xmath20 and present mass @xmath21 fall into the domain of stellar models .
distance @xmath22 and parallax @xmath23 are determined either from direct astrometric observations or via the distance modulus when comparing stellar models with photometry .
thus the full parameter space can be disentangled into individual contributions : @xmath24 where @xmath17 is the core parameter space the other r@xmath25 are the parameters of importance to different types of observations or prior expectations ( see sec . 3.2 to 3.6 ) .
in contrast to parameters , which span the n - dimensional space of the posterior probability distribution , the nature of observations is irrelevant .
observations can be anything , from the numbers of electrons on a ccd to a needle on a scale .
each observation puts a constraint on our parameter space , which is its corresponding observed likelihood @xmath26 as a function on parameter space . instead of just writing down an observational likelihood ,
the common approach in astronomy is to `` simplify '' this by the introduction of `` observables '' .
while this term is not well - defined in a bayesian context , `` observables '' commonly denote best - fit values for some parameters ( like @xmath27 ) , which appear to be relatively well - constrained by ( single ) observations . some studies , like ( e.g. * ?
* ) , go even further to introduce the errors on those `` observables '' as further variables in their formalism . from an aesthetic point of view , this results in a rather clumsy and complicated bulk of variables to achieve a simple goal : describing the real observed likelihood .
it has two practical consequences .
first , `` observables '' lead to an oversimplification of the observational likelihood , usually with the unjustified ( and damaging , see section [ sec : gaussian ] ) approximation as a product of separate gaussians in each parameter termed `` observable '' .
second , their introduction artificially introduces a `` better '' class of parameters , raising the wrong suggestion that their values are fixed .
this is not true . for
any parameter the bayesian formalism will in general give an estimate different from the best - fit value . while selection functions are in most cases essential for understanding observations with theoretical models
, this does not apply to the discussed bayesian schemes . yet
, some studies introduce a selection function in their equations ( see a longer discussion in sec . [
sec : sf ] of the appendix ) .
we refrain from using such a selection function , because only selection criteria based on the parameter space would affect the bayesian scheme , while a survey selection must be based on random choice or previous observations .
to facilitate reading the equations we quickly summarize the main notations : we denote the set of observations by @xmath5 , the set of parameters by @xmath4 , the parameter space by @xmath28 and all probability distributions by @xmath29 . to cope with the different sources of information , we introduce indices : `` ph '' for photometry , `` sp '' for spectroscopy , `` astr '' for astrometry ( parallaxes ) , in addition we use `` mod '' for knowledge from stellar models and `` pr '' for priors . hence the observational likelihood in full parameter space from spectroscopic observations reads @xmath30 . to facilitate the reading we contract the notation for the conditional probabilities by decorating @xmath29 with a prime : e.g. @xmath31 .
commonly used variables are age @xmath19 , stellar mass @xmath21 , solar mass @xmath32 , initial mass @xmath33 , logarithmic iron abundance @xmath34 , general metallicity @xmath35 , parallax @xmath23 , distance @xmath22 , and distance modulus @xmath36 .
in this pilot study , we restrict ourselves to the most important basic case : the calculation of stellar parameters , when we have spectroscopic , and/or photometric observations .
we will show how to expand this to include parallax measurements . after validating the method on the high - resolution spectroscopic data of nearby stars ,
we apply it to a sample of low - resolution spectra from segue / sdss @xcite calibration sample . with the conditional independence ( equation [ eq : condimp ] ) we simplify the calculation of our posterior : @xmath37 where @xmath38 denote the photometric and spectroscopic and astrometric observations , @xmath39 the probability derived from stellar models and @xmath40 the prior probability distribution function . in the second line we abbreviate our observational likelihoods , representing their conditional nature based on the observations in short by a prime .
note that any combination of observational constraints can be dropped from these equations , as well as new observations ( e.g. interferometry ) can be added by multiplication .
the pdfs from equ .
[ eq : ar ] have two interesting qualities : * some pdfs describe sharp structures in the n - dimensional parameter space , thus lowering the dimensionality of the probability distribution and reducing computational costs ( by the multiplication , the combined pdf can not have higher dimensionality than its components ) . in other words ,
the space volume where their pdf is non - negligible has a lower dimensionality than the overall space .
for example , stellar models together with model atmospheres map directly from the fundamental stellar parameters ( @xmath41 to their observed space ( @xmath42 ) . *
some pdfs constrain only a subset of @xmath43 parameters , i.e. they are flat the other dimensions .
though they can be generalised to the n - dimensional space of the aggregate pdf , most of these dimensions will be redundant , i.e. we have @xmath44
. it can be efficient to merge them in an early step with another pdf that carries more dimensions .
an especially valuable case are parameters that are nearly conditionally independent from the other parameters .
e.g. the detailed abundances for the majority of chemical elements hardly affect temperature and gravity estimates .
the meaning and structure of the single contributors to the posterior @xmath29 will be examined below .
the priors encode our previous knowledge on the distribution of the examined stellar population in parameter space .
the model knowledge @xmath45 will be treated here as a separate prior , though it could be in fact understood as an observation .
appropriate priors are essential to avoid biases in weakly constrained data ( see for an example ) .
further , to have set `` no prior '' means to have adopted a flat prior , which is not fixed under parameter transformations .
however , priors must be handled with great caution to obey cromwell s rule ( avoid excluding any outcomes a priori * ? ? ?
* ) and to avoid overconfidence biases and reproducibility problems .
how can one cope with uncertainties in a priori parameter distributions ? making a prior just `` shallower '' or adopting a completely flat prior is not useful , since it actually adopts a worse probability distribution claiming the same certainty on this information as the original prior .
however , one typically has estimates on how uncertain the distribution in a set of parameters is . if , within those uncertainties , the precise shape of a prior has important impacts on results , there are two strategies : the uncertainty of a prior can be constrained by demanding consistency with calibration data ( data of higher quality , where the prior has less impact ) and by imposing internal consistency between different subsamples ( e.g. red versus blue stars ) .
the remaining uncertainty in a prior can be covered by hyper - parameters : the shape of the prior can be parametrised and a probability distribution function estimated on these parameters .
there are strong dependencies between most parameters . as the posterior pdf has lower effective dimensionality than the parameter space @xmath28 , we can not set constraints on every single dimension in @xmath28 without risk of over - constraining the priors .
we circumvent this problem by limiting our effective priors to age @xmath19 , initial mass @xmath20 , metallicity @xmath35 , and distance @xmath22 ; all other dimensions are indirectly constrained by these priors and we adopt no additional constraints on them . throughout this work we will use the following priors : @xmath46},s ) \sim \\ & \sim & p(\tau | { \rm [ fe / h ] } ) \cdot p(\mi ) \cdot p(s , l , b ) \cdot p(\meh , \feh ) . \notag
\end{aligned}\ ] ] @xmath47 is a fixed relation between metallicity ( required for the isochrones ) and the iron abundance , which we have to introduce , since we do not measure detailed abundances in this paper . the adopted relation is given in sec .
[ sec : amprior ] of the appendix .
for @xmath48 we employ a salpeter imf @xcite , with exponent @xmath49 and independent from metallicity and age .
we account for the metallicity - dependent age distribution by adopting a shorter timescale in the star formation history of metal - poor stars .
details are given in the appendix .
due to its importance we need to discuss the spatial prior @xmath50 . in general
, every sample will cover some fixed angle on the sky ( be it so - called pencil beams like in segue or a complete sky coverage ) , so the actual volume is a cone that covers an effective area @xmath51 , where @xmath22 is the distance .
the constant @xmath52 , given by the sky coverage and selection probabilities , is irrelevant in our context .
however , the likelihood to end up in the sample is also proportional to the density of the population in the observed region , so that we obtain : @xmath53 where we integrate over the sky position @xmath54 , and @xmath55 is the spatial density of stars at position @xmath54 and distance @xmath22 . in our formulation , this must be multiplied by another factor of distance @xmath22 , since a fixed magnitude range samples a spatial depth proportional to @xmath22 ( formally this derives from a parameter transformation of the probability density from magnitudes to distance ) .
neglect of this prior is not equivalent to a flat distance prior , but one that drops steeply with the third power of the distance , and hence in the absence of parallax measurements would give a strong ( and uncertainty dependent ) bias towards lower surface gravities .
stellar models describe an effectively three - dimensional constraint in the full parameter space . the corresponding pdf can be represented in the core parameter space @xmath56 . in this work ,
we neglect other dimensions , adopting a simple relationship for alpha enhancement and neglecting initial / present stellar rotation
. the calculation of the pdf is performed by summing the weights of available stellar model points falling into the cells of a dense grid in the target space @xmath17 . throughout this work
we fold the models with a gaussian kernel with widths @xmath57 in @xmath58 .
this error represents the internal uncertainty of the stellar models in parameter space and fills gaps caused by the discrete data representation .
in addition , the effective width is augmented by the grid spacing on which the pdf is calculated .
we use a dense grid of stellar isochrones from the basti database @xcite , kindly provided to us by s. cassisi for the stellar parameter determinations in @xcite .
we interpolate the models in the initial mass @xmath20 to ensure a narrow mass spacing , but do not attempt an interpolation in age @xmath19 or metallicity @xmath35 .
when summing over the isochrones , we assign to each point @xmath59 a weight @xmath60 proportional to the parameter space volume it represents : @xmath61 where @xmath62 is the normalisation , @xmath63 is the average distance to its neighbours in initial mass , @xmath64 is the average distance in metallicity between the isochrone and its nearest neighbours , and @xmath65 is the average distance in age @xmath19 . on the boundary of the grid
we take the distance to the neighbouring point .
note that the approach is identical for stellar tracks instead of isochrones .
the model probability at each point @xmath4 in our parameter space can then be represented as a weighted sum over all relevant models points @xmath59 : @xmath66 where @xmath67 is the vector in parameter space given by the model grid . here
we represent the uncertainty of the models by an n - dimensional gaussian @xmath68 with a dispersion vector @xmath69 .
specifically , we assume @xmath70 on our core parameter space as above , and no additional uncertainties in the other dimensions .
stellar models couple photometric colours with other parameters ( see * ? ? ?
* for details on the colour calibrations in sdss filters of basti models ) .
thus we best calculate the photometric pdf simultaneously with the stellar models . denoting the stellar model magnitudes at model point @xmath59 and colour band @xmath52 by @xmath72 and the photometric observation in band @xmath52 with @xmath73
, we have : @xmath74 with distance @xmath22 and reddening @xmath75 .
lacking sufficient data on the true pdfs , we represent the observational likelihoods of photometric colours , the reddening values and the model uncertainties by a gaussian @xmath76 , which enables us to combine them into : @xmath77 with the distance modulus @xmath78 , the reddening strength @xmath75 multiplied with the reddening vector ( for sdss colours , see * ? ? ?
* ; * ? ? ?
* ) in each colour @xmath79 , and with @xmath80 as the combined variance / uncertainty of models , observations and reddening .
we assume a magnitude uncertainty on stellar models @xmath81 for the high - res sample , neglecting this term for the low - resolution sample .
partly , @xmath82 covers the same uncertainties as our error term * @xmath83 * on stellar parameters ( see section [ sec : mod ] ) .
we stress that this uncertainty can not and should not comprise systematic deviations in stellar models , as those uncertainties should be explored on a larger sample , using hyper - parameters as discussed in section [ sec : priors ] .
the other assumption is the universal reddening vector .
this may have to be relaxed when dealing with very different ism environments .
the dust peak and the slope of the reddening spectrum can be shifted , or stars may be individually reddened , e.g. by a circumstellar envelope .
we note that this method can be used to create reddening maps .
since that is beyond the scope of this work , we restrict the sample to stars with relatively low reddening , use reddening values from other sources assuming a fractional reddening error of @xmath84 .
the observational likelihood @xmath85 in eq . ( [ eq : ar ] ) incorporates all available spectral information .
this comprises spectral type , element abundances , rotation , stellar activity ( chromospheric emission in cores of strong lines , magnetic effects ) , inter- and circum - stellar reddening , convection characteristics , etc .
a spectrum offers by far the largest information content we can obtain on a star .
however , this information is limited by incomplete physical knowledge and approximations in modelling the theoretical spectra . at present , calculations of large spectral grids are only possible with 1d hydrostatic codes assuming local thermodynamical equilibrium ( lte ) , while full hydrodynamic 3d non - lte calculations are slowly becoming feasible @xcite .
further , the high dimensionality of the problem forbids computing separate grids for all possible chemical compositions .
here we use the mafags - odf @xcite grid of model atmospheres designed for late - type ( spectral type fgkm ) stars , with @xmath86 k , @xmath87 ,
the synthetic spectra are computed with the revised version of the siu code @xcite , which has been extensively used during the past two decades for high - precision stellar spectroscopy ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in comparison to the other three available spectrum analysis codes ,
moog @xcite , synthe @xcite , and turbospectrum @xcite , siu already has an implementation of nlte line formation for any element with pre - computed nlte level populations .
we can thus more easily update it with more realistic physics . for our grid of theoretical spectra
, we use a spacing of ( @xmath89 ) in ( @xmath90 ) to make linear interpolation between the points reasonable . while the grid covers @xmath91 values of micro - turbulence from @xmath13 to @xmath92 , for this pilot study we adopt a fixed micro - turbulence of @xmath93 for giants and @xmath94 for dwarfs ( @xmath95 ) ( bergemann et al . in prep ) , and an @xmath96-enhancement of @xmath97 dex for @xmath98 ( e.g. * ? ? ?
* ) . such hard cuts and changes in assumed parameters will introduce anomalies in the derived pdf , as exemplified by the micro - turbulence cut at @xmath99 in . in total , the three - dimensional ( @xmath35 , @xmath27 , @xmath100 ) grid contains @xmath101 theoretical spectra and covers the full hrd , as shown in fig .
[ fig : specstuff ] . any other model grid can be easily implemented , with no requirement on symmetry or shape , since our code includes a robust interpolation scheme .
alternatively , one could perform calculations of line formation on the fly using a grid of model atmospheres .
this latter approach is cleaner , however , it is still computationally too costly .
we sample the wavelength windows around the spectral features important for diagnostic of fgkm stars : @xmath102 ( ca i lines ) , @xmath103 ( g - band , cn sensitive ) , @xmath104 ( @xmath105 ) , @xmath106 ( mg i triplet , main gravity diagnostics ) , @xmath107 ( @xmath108 ) , @xmath109 ( ca ii triplet , also used in gaia and in rave stellar survey ) .
however , not all pixels in these intervals are used in the analysis . the high - resolution observed spectra ( see sec .
4.3.1 ) do not cover the regions below @xmath110 and above @xmath111 .
we exclude from our statistics all regions which contain spectral lines of chemical elements other than the temperature- and pressure - sensitive wings of balmer and mg i triplet lines , and the fe i and fe ii lines . precisely , the weight of all other spectral features , is set to zero .
the flat regions are used for the iterative continuum normalization and are not masked out . to avoid over - confident estimates ,
we demand that either the temperature uncertainty @xmath112 or the metallicity uncertainty @xmath113 , and otherwise flatten the pdf by multiplying the @xmath114 distribution with a fixed factor until the condition is met . before evaluating the test statistics ,
the spectra are continuum - normalised and radial - velocity corrected by cross - correlating with the template theoretical spectrum for each input combination of stellar parameters . to obtain the spectroscopic observational likelihood @xmath115 at each point in parameter space , we resample the synthetic spectrum to the wavelength scale and resolution of the observations and evaluate the goodness - of - fit - statistics @xmath114 at each pixel @xmath59 of the observed spectrum : @xmath116 where @xmath117 the template comparison spectrum , @xmath118 the observed spectrum , @xmath119 the weighted observational uncertainty .
noisy and un - informative regions are given less weight using special masks .
the final pdf is gained by summing over all pixels within a given segment , and over all segments .
the original resolution of the synthetic grid is @xmath120 .
thus , the method can be potentially applied to any observed dataset , e.g. , low - resolution and high - resolution spectra . for the analysis of the segue spectra , we post - convolved the spectral grids with instrumental resolution , @xmath121 . a typical fit to a segue spectrum is shown in . in the high - resolution mode , we use the resolution of the uves - instrument ( @xmath122 ) .
the gaia mission will derive parallax measurements for nearly all stars with spectroscopic information .
parallax measurements only affect the distance @xmath22 ( and distance modulus @xmath36 ) , so that it is straightforward to combine the observational likelihood from parallax measurements @xmath123 with the photometric and model information . in the following ,
we assume a gaussian parallax error .
cromwell s rule does not apply to mathematical truths , so negative parallaxes are excluded by setting the prior to @xmath124 .
this yields : @xmath125 where n is a normalisation , @xmath126 is the heaviside - function ( @xmath13 for @xmath127 and @xmath124 for @xmath128 ) , @xmath129 is again a gaussian distribution around the measured parallax @xmath130 ( which can be negative ) with standard deviation @xmath69 .
it is important not to clip negative values of @xmath130 : a small negative value of @xmath130 has still a different information content than a large negative value . in the case of a gaussian error distribution ,
the probability ratio between a smaller parallax and a larger parallax rises , the further the measurement is away from zero . or to use an example : the likelihood ratio between having failed by @xmath131 and by @xmath132 is larger than the likelihood ratio between having failed by @xmath131 and by @xmath133 . demonstrates how the parallax distributions get more concentrated towards zero , the more negative the measured value is . to combine @xmath134 it with the photometric and model pdf , we integrate over the possible distance moduli @xmath36 , at each stellar model point @xmath59 : @xmath135 with the jacobian @xmath136 . in this work ,
p@xmath137 is considered for the stars with high - resolution spectra only ( see sec . [
sec : highres ] ) . equipped with these results
, we can now assemble the combined pdf in equation [ eq : ar ] . in simple words , the strategy is separate all pdfs into pdfs on the core parameter space @xmath138 and the conditional pdfs on the remaining parameter space given that point in @xmath17 . depending on our needs
we can then represent those remaining parameter estimates either as simple moments ( expectation value , variance , etc . ) at each point in @xmath17 , or as full distributions .
formalising this is a bit tedious , since it involves a conditional probability derived from a conditional probability . to simplify the notation
, we use the previous abbreviation of observational dependence with a prime .
the combined calculation of photometric and model part yields : @xmath139 where @xmath140 is the vector of parameters in our core parameter space @xmath17 and @xmath141 is the vector of remaining parameters constrained by the photometric and astrometric observations , models and priors , i.e. @xmath142 .
similarly , we separate the spectroscopic information : @xmath143 where @xmath144 denotes all other parameters constrained by spectroscopic observations , like detailed abundances , or stellar rotation . in this work
we do not use this supplementary information , so that we can drop the term @xmath145 .
most of the parameters in @xmath144 will not coincide with the parameters in @xmath141 , but if they correspond , they must be written into the core parameter space .
for example , rotation and stellar activity available from high - quality spectra constrain stellar ages .
we will discuss this in a future work .
we can now calculate the final probability distribution function : @xmath146 where @xmath147 we can gain the conditional probability distribution in a lower number of parameters by marginalising , i.e. by integrating out the other dimensions in the joint conditional probability distribution function .
e.g. , to exclude the parameter x@xmath148 , we write : @xmath149 from this we can obtain the moments of the probability distribution in each variable or group of variables : @xmath150 where @xmath151 denotes the expectation value of the parameter @xmath152 , and the standard deviation @xmath153 . in short
the steps are as follows : * \1 ) combine photometric and astrometric information together with the priors and sum over all stellar model points to obtain a preliminary pdf @xmath154 in core parameter space , calculate moments or full pdfs for the remaining dimensions .
* \2 ) in regions of parameter space , where the probability is larger than a threshold value per bin . compared to a number of @xmath155 bins we hence neglect a negligible fraction of the probability mass .
] , calculate a coarse grid of spectroscopic probabilities and approximate the pdf @xmath156 by interpolation . *
\3 ) multiply @xmath154 with @xmath156 to obtain an approximate posterior pdf @xmath29 .
determine a refined grid in parameter space to better sample the spectroscopic pdf and iterate steps 2 ) + 3 ) ( fig .
[ fig : search ] ) .
the threshold value on a binned pdf was chosen as @xmath157 where @xmath158 is the number of stars and @xmath21 is the effective number of bins , because in a large sample we have to expect the presence of rare objects , which will have low preliminary probabilities . as a different condition one can formulate that the integral of the pdf over parameter space must be @xmath159 to illustrate the algorithm ,
we describe here the results for two stars .
one is @xmath160 hyi from our high resolution data sample , for which we have basic johnson photometry , high resolution spectroscopy and a hipparcos parallax .
the other star , randomly selected from the segue data sample ( plate number @xmath161 and fiber number @xmath162 ) , is a turn - off subgiant . in this case
we have sdss photometry and a low - resolution spectrum from segue .
the resulting probability distribution functions in the @xmath163-plane are shown in and in .
the top left panel shows the combined pdf from photometry , prior and stellar evolution ; the top right panel shows the spectroscopic pdf in photometrically allowed space .
these two estimates combine to the final posterior pdf in the bottom left panel .
the corresponding metallicities are shown by colour coding in the bottom right panel .
the individual probability densities from photometry - stellar evolution and spectroscopy are clearly different in shape and in location .
the hipparcos parallax combined with photometry and stellar models puts tight constraints on the surface gravity of @xmath160 hyi in .
this leads also to a tight correlation between metallicity and gravity as evident from the coloured dots in the bottom right panel .
a moderate step in the spectroscopic pdf at @xmath164 is produced by a step in micro - turbulence in our current grid of theoretical spectra , which will disappear with the improved grids in preparation .
the calculation does not cover the full allowed region of the spectroscopic pdf ( see the coarse behaviour at smaller gravities in the top right panel ) , saving computation time , since the joint pdf ( lower left panel ) is fully represented .
the final expectation values and uncertainties are @xmath165 , @xmath166 , and @xmath167 versus @xmath168 , @xmath169 , and @xmath170 in the reference sample ( described in the next section ) .
while neither the photometric part nor the spectroscopic constraints are very tight for the segue star in , the combined pdf is very well defined .
this shows the benefits of solving the problem in the full parameter space @xmath17 .
while points in the @xmath171 plane may be allowed by both derivations , the corresponding limits on the third dimension @xmath34 are in disagreement , ruling them out .
these are the regions in the bottom right panel of , where the colours are mismatched . to stress this point we show the one - dimensional probability distributions in @xmath27 in . while our parameters are nicely between the values of the segue follow - up study @xcite and segue dr9 ( see sec . [
sec : datasets ] ) , the behaviour of our pdf is more interesting : the combined pdf is not even remotely a simple combination of its two contributors .
most interestingly , the expectation value of the combined estimate is not situated between the estimates from each spectroscopy ( @xmath172 ) and photometry ( @xmath173 ) , but significantly higher ( @xmath174 ) .
this complex behaviour can only be accounted for within a full bayesian approach .
our final expectation values and uncertainties for this segue test star are @xmath175 , for comparison segue dr9 provides @xmath176 .
note that we add the reported uncertainties from the segue pipeline just for the sake of completeness .
their formally reported errors can not be considered realistic .
they are severely under - estimated ( by about an order of magnitude or more ) as shown by the comparisons in @xcite as well the discussion later in this work .
the spectral fits in our six standard bands for the best spectroscopic solution are shown in .
this discussion also shows that even a relatively uncertain information can give an improvement to more precise values that is beyond a simple one - dimensional combination .
more importantly , mismatches between different sources of information help to flag pathologies in a sample by unexpectedly small overlap of the contributing pdfs .
our approach is most needed and also most powerful , when different observations are available for a star and the information content is complementary but limited . with this in mind and to test the stability of our method , we choose both a sample featuring high - resolution spectra ( @xmath177 ) , from observations with vlt , as well as one with low - resolution spectra ( @xmath178 ) from sdss / segue .
we start this section with a description of the datasets in use .
then , we first show the performance of the approach when limiting ourselves to photometric data with and without astrometry , followed by the full approach on low- and high - resolution spectra . in the last subsection we compare derived quantities , like distances and ages , and assess the resulting distribution of stars in the temperature - gravity plane . for the high - resolution sample we obtained a comparison set of stellar parameters from @xcite .
their effective temperatures were derived from the interferometric angular diameters or calibration relations .
the gravities stem from astroseismology or hipparcos parallaxes , and their metallicities are based on the analysis of fe ii lines , which are not significantly affected by non - lte effects @xcite .
the high - resolution sample comprises @xmath179 high - resolution spectra of @xmath180 nearby stars including the sun , taken with the harps and uves instruments at vlt , and with narval at the pic du midi observatory .
this dataset was kindly provided by p. jofre .
in addition , there are hipparcos parallaxes @xcite , making the sample closely resemble future data from gaia astrometry combined with gaia - eso spectroscopy .
the sample is particularly valuable because the spectra were taken on different instruments and there are independent parameter determinations , including interferometric angular diameters and astroseismic surface gravities @xcite .
the stars cover a very wide range in metallicities , gravities and temperatures in parameter space ( see a complete description in * ? ? ?
photometry in the @xmath181 bands , was compiled from the hipparcos catalogue @xcite , from 2mass @xcite , and from @xcite .
@xmath182band photometry for hd@xmath183 stems from @xcite , improved @xmath184-photometry for @xmath185 hya from @xcite .
solar photometry was adopted from @xcite , updated with the values of @xcite .
we increased the errors in light of the general uncertainty of the sun s photometry to @xmath186 .
our low - resolution sample was selected from segue by @xcite , who did an intermediate - resolution follow - up analysis . from the original r @xmath187 with unclear consequences . ]
it consists of @xmath188 stars within the parameter range @xmath189 , @xmath190 k , and @xmath191 .
for these stars , we have low - resolution segue spectra , ( @xmath178 ) , photometry in the sdss @xmath192 bands , @xcite reddening estimates and positional data from sdss dr9 @xcite .
one star was removed from the sample , as it was flagged for strongly disagreeing observational information ( very low quality measure @xmath193 , cf .
equation [ eq : qq ] , resulting partly from a strong cosmic in the spectrum ) . throughout the text
we refer to the parameters from @xcite as `` ap08 '' and from segue dr9 as `` dr9 '' .
two important remarks should be made about these comparison sets : while they are great tools for comparisons , they are subject to uncertainties and systematics that can exceed the quoted errors .
second , the quoted errors are fundamentally different from ours .
they just report internal errors from pipelines or spectroscopic fitting routines , which are typically far smaller than realistic error estimates .
the most extreme case is the segue parameter pipeline .
this is fundamentally different from our error calculations , which attempt to calculate all uncertainties . before we explore the performance of the full algorithm against our reference samples , we test it for the simpler case , where spectroscopic information is not available .
for the vast majority of stars in the galaxy , we will have no or very limited spectral information ( e.g. 4most will cover only of order @xmath194 of the stars in the gaia catalogues ) .
however , we find that the bayesian method is capable of deriving stellar parameters also when restricted to photometric and astrometric information . in , we compare temperatures , gravities and metallicities derived from johnson photometry and parallaxes only with our high resolution reference sample ( top row ) , and from sdss photometry alone with values from segue dr9 ( bottom row ) . for the high - resolution reference sample ( top row )
the applied photometry is not competitive with what can be expected from modern photometric surveys : for most stars we are restricted to johnson @xmath195 colours at @xmath196 precision , and , due to their brightness , very uncertain @xmath197 photometry .
nevertheless , the photometry gives a good handle on effective temperatures : while our temperatures are mildly higher than the reference , the random scatter is as low as @xmath198 .
the excellent agreement for @xmath199 is a consequence of using hipparcos parallaxes .
we note that for most stars even uncertain parallaxes suffice to fix @xmath199 , as they constrain the stellar branch , i.e. whether a star is on the main - sequence or e.g. on the red giant branch .
metallicities are per se very weakly determined with johnson broad band filters , and particularly without decent @xmath200-band measurements .
the large uncertainties concentrate the values towards the centre of our grid .
this underlines the need for intermediate or narrow band photometric surveys to constrain stellar parameters .
the precise sdss photometry and the location of sdss colour bands allow for a better handle on metallicities , as well as for good temperatures . in absence of parallax measurements ,
photometry alone offers a rough classification of stars , as seen in the bottom row of with a rms scatter against the segue parameter pipeline of around @xmath201 .
while there is significant photometric information , it is not strong enough to be insensitive to the priors .
this motivates a closer look at the importance of our assumptions .
in we compare our photometric metallicities ( y - axis ) to the metallicities from @xcite ( x - axis ; for the standard segue dr9 comparison , see ) for the segue sample , plotting stars with clean photometry with larger discs and stars with bad photometry with smaller triangles .
colours encode the error estimates from the bayesian method .
evidently , there is enough information to constrain metallicities at least in the higher metallicity range to an accuracy of about @xmath202 .
contrary to common derivations like @xcite , which fail at metallicities @xmath203 ( cf .
* ) , our approach is valid throughout the entire metallicity range .
however , it is important to realize how important the age prior becomes in this case . in the lower panel
we show the same data with a fully flat age prior instead of using eq .
[ eq : ageprior ] .
this flat age prior implies a far larger uncertainty in the gravity of a star , which severely affects objects that can not be clearly identified as subgiants , or main sequence stars . via the degeneracy of @xmath204band information ,
their potentially lower gravities allow for a wider range of ( mostly lower ) metallicities , which lowers the expectation values and boosts the error estimates . despite this problem , the situation is far better than in the traditional approach : the classical metallicity calibrations like @xcite or @xcite rely on stars falling not only on a fixed age bin , but also onto a single evolutionary sequence .
this leads to a metallicity bias and overconfidence concerning the uncertainties .
in contrast , the full bayesian approach makes optimal use of all available colour information , accounts for all sources of uncertainty and allows to explore the effects of prior assumptions . shows the comparison of our parameter expectation values with the segue dr9 data release .
colour codes the quality measure @xmath205 which gives a simplified indication on how well the spectroscopic pdf agrees with the remaining information .
our temperatures are systematically colder than segue dr9 by about @xmath206 .
this is a consequence of our spectral and photometric @xmath27 scales being @xmath207 and respectively @xmath208 colder , suggesting that segue dr9 overestimates stellar temperatures .
our spectroscopic temperature scale is nearly identical with that of ( * ? ? ?
* hereafter ap08 ) , which is on average @xmath209 below segue dr9 derivations .
the strength of our full approach becomes apparent in the residual scatter of the temperature values after correcting for the systematic offset : while spectroscopic and purely photometric temperatures give a residual rms of @xmath210 and @xmath211 relative to segue dr9 , the full approach excels with @xmath212 .
the bayesian gravities are systematically higher for suspected main sequence stars ( @xmath213 in both determinations ) , reflecting the systematic gravity underestimates of segue dr9 in this range ( also confirmed by segue not matching expectations for the main sequence ) .
the purely spectroscopic gravities of our method are significantly lower than dr9 and ap08 by @xmath214 in the intermediate and lower gravity range ( using @xmath215 in ap08 ) .
this is clearly identified as a bias , since the bayesian approach reports too young ages , especially for several metal - poor stars .
though the bayesian approach can not completely eradicate a systematic bias in one of its inputs , it strongly reduces this problem by systematically increasing the surface gravities by an average of @xmath216 compared to the purely spectroscopic value .
the bayesian metallicity determinations for @xmath34 between @xmath217 and @xmath218 are robust . however , metal - rich stars have a recognizable metallicity difference between our photometric and our spectroscopic determinations , with the latter being systematically lower .
for the open cluster @xmath219 ( @xmath220 or @xmath221 * ? ? ?
* ; * ? ? ?
* ) our spectroscopy alone gives @xmath222 versus a photometric estimate of @xmath221 .
as in the case of the gravities , the bayesian method partly mitigates this problem : photometric metallicities in this range push the combined estimates towards higher values ; however , due to the intrinsic uncertainty of @xmath202 , the corrections are minor .
this also shows the importance of fair error assessment : overconfident , i.e. too small , error estimates from spectroscopy prevent a stronger correction of the value by the photometric information , which has intrinsic uncertainties of @xmath223 in this range .
tests show stability of our results down to a signal to noise ratio of @xmath224 and checks on the continuum placement yielded no conclusive evidence .
it is very likely that a finer resolution of the grid of synthetic spectra , its extension to a larger wavelength coverage of the segue spectral range . ] , and allowing for @xmath96-enhancement will solve the problem .
this work is in progress and will be presented in a future paper dealing specifically with the analysis of segue spectra . the most important result is , that even with systematic biases present in the inputs , the bayesian method itself remains robust , i.e. other parameters are not strongly affected , and the solutions are pushed towards a significantly less biased result .
figure [ fig : res2 ] compares the reference parameters ( x - axis ) to the expectation values from our full bayesian analysis ( y - axis ) .
again colours encode the value of the quality measure @xmath225 from equation [ eq : qq ] .
we also give statistics in table [ tab : highrescomp ] .
currently , our spectral grids do not cover stars with @xmath226 , @xmath227 and assume a slow stellar rotation of @xmath94 , which is typical for most g and k stars @xcite .
hence , for the spectroscopic comparison sample we have to exclude stars with @xmath228 and drop the fast rotating stars @xmath229 bootis and @xmath36 leonis , which have @xmath230 and @xmath231 .
we also remove @xmath185 hya due to contradictory results from different astroseismic derivations @xcite .
this leaves @xmath232 stars with @xmath233 spectra .
the two metallicity outliers at high metallicity in are @xmath234 and @xmath160 gem .
both have very high macro - turbulence values ( @xmath231 * ? ? ?
* ) , which contradict the current assumptions of our spectral pipeline .
l|cc|cc + parameter & @xmath235 & @xmath69 & @xmath236 & @xmath237 + + @xmath238 & @xmath239 & @xmath240 & @xmath241 & @xmath242 + @xmath243 & @xmath244 & @xmath245 & @xmath246 & @xmath247 + @xmath248 & @xmath249 & @xmath250 & @xmath251 & @xmath252 + from table [ tab : highrescomp ] , it is apparent that if we confine the sample to the subset with astroseismic determinations , the random mean scatter in all quantities is reduced , by more than half .
this implies that only the astroseismic subset can match or exceed our precision , while the bayesian method is clearly superior to the traditional analysis on comparable data .
some of the reference gravities were derived from hipparcos parallaxes , which should make them similar to our results . in this case ,
our fully bayesian determinations in gravity appear more reliable than the less sophisticated reference because they also take into account physical information from colours and spectra .
interferometric @xmath27 , although they are usually taken to be mildly model - dependent , still require an estimate of limb darkening and bolometric fluxes .
the former are determined with 1d lte model atmospheres , while @xcite showed that 3d hydrodynamical models predict different centre - to - limb variation , which may cause systematic biases in angular diameter estimates .
bolometric fluxes are estimated by interpolating between observed photometric magnitudes with the help of theoretical spectra , giving rise to another systematic uncertainty .
it is instructive to compare the full method results to the spectroscopic results . in the bottom row of we show expectation values and parameter uncertainty from purely spectroscopic information ( green error bars ) and when using spectroscopy plus the model prior ( coloured points with blue error bars ) .
spectroscopic surface gravities alone are generally too low by @xmath253 with a residual scatter of about @xmath254 compared to the full solution ( see * ? ? ?
* for discussion of similar spectroscopic underestimates ) .
using spectra in combination with stellar evolution models in the bayesian framework , but excluding the parallax and photometric information , improves the residual scatter to @xmath255 and fully removes the systematic offset .
hence , while spectroscopic information alone can not compete with astrometric information , it gives sufficient information on surface gravity to allow for decent values derived by the bayesian framework . in table
[ tab : allhighres ] we provide the stellar parameters and ages from the full bayesian method .
when more than one spectrum is present for a star , we provide the weighted average of the expectation values and errors ( we have to assume that the errors between the single determinations are highly dependent ) for single spectra .
where no spectral information is available , we fill in the results from the combination of photometry and parallax measurements . inspection of sample distributions in parameter space , like in the temperature - gravity planes shown in , provides clues about the reliability of each parameter determination . in this figure we show the hr diagrams in the @xmath256-plane with expectation values from the bayesian method ( top row ) , versus the reference parameters ( bottom ) , for both the segue sample ( left hand side ) and the high - resolution reference sample ( right ) . to facilitate the interpretation , we plot isochrones at @xmath257 and @xmath258 gyr at three different metallicities @xmath259 , matching the colour scale of the stars .
the key differences between our results and those determined by conventional methods are obvious . despite the mildly biased spectroscopic gravity estimates ,
our results show a clearly superior performance in this plot .
the bayesian results cover the main sequence , while segue dr9 does not attain main sequence values .
even more striking is the appearance of unphysical stars : both segue dr9 and the reference sample from heiter et al .
have stars in highly unphysical positions , with the error estimates not even close to the offset from the nearest evolutionary sequence .
e.g. both segue dr9 and the high - resolution reference sample place three stars around @xmath260 far right of the turn - off or respectively right of the main - sequence .
the plot suggests that the gravity offsets between the high - resolution reference values and the bayesian method track back to a neglected metallicity effect in the reference sample . in principle
the bayesian method could yield stars in between the sequences , since we here give expectation values .
a hint of this tendency can be seen , but by construction our errors will correspond to the offset , because the actual likelihood at the unphysical points is near zero . the resulting age - metallicity relation for the high - resolution sample
is displayed in . to make the plot easier to read
, we merged the entries for different spectra as in table [ tab : allhighres ] .
the picture very much resembles the results of @xcite .
the younger expectation values for one of the very metal - poor stars corresponds to a larger error estimate , forcing the expectation value away from the hard boundary given by the age of the universe .
further there is no striking trend in metallicity at younger ages .
the importance of a reliable assessment of all stellar parameters in one single approach is demonstrated in .
here we plot the same stars from segue as in the top left panel of , but now colour coded with the estimated fractional distance error .
it is apparent that even some very high gravity estimates are no guarantee for a good main sequence classification , vice versa stars with lower gravity can have high distance confidence .
as expected , these stars are usually cleanly identified subgiants , giants , or even better , red - clump stars .
while the distance and its uncertainty are in principle enough to support estimates of mean motion and velocity dispersions in a population , we point out that an investigation of velocity distributions themselves requires accurate estimates of the exact shape of the probability distribution in distance space , which the bayesian method can deliver .
in this section we attempt to compare our implementation of spectroscopic information to an approximation by one - dimensional gaussian uncertainties . while this is certainly not the only point by which our algorithm differs from other studies in the literature , the simplification of the spectroscopic information is common to the works we are aware of .
we perform this experiment on the segue sample . to degrade our spectroscopic results
, we calculate the uncertainties @xmath69 and mean / expectation values ( @xmath261 , @xmath262 , @xmath263 ) of each quantity separately and then change the spectroscopic pdf to a product of one - dimensional gaussians in each parameter : @xmath264 where @xmath265 and @xmath158 is the normalisation .
as we see from the spectroscopic pdfs in the top right panels of and the spectroscopic information can by no means be described as a product of gaussian errors in @xmath27 , @xmath199 , and @xmath34 separately : the pdf is not even remotely aligned with the coordinate axes and , for most stars we examined , shows a highly irregular shape .
the top panel in shows the relative shifts between our full approach and the more conventional approximation in @xmath27 and @xmath199 for each star , normalised by the errors derived in our normal method , i.e. @xmath266 intuitively one might expect very small changes , because apart from approximating the spectroscopic pdf , we left all information untouched .
the contrary is true , since the shape of the spectroscopic pdf gets distorted and now intersects the other constraints in parameter space at different locations ( this problem is aggravated with higher dimensionality of parameter space and a more irregular pdf ) .
consequently the expectation values of the parameters scatter by more than @xmath267 .
the failure of the `` classic '' approach can be seen in , where we plot the photometric , spectroscopic and photometric probability distributions for our full approach on the left versus the degraded approach on the right hand side .
looking at the invoked difference in the spectroscopic pdf , which , more importantly does not carry any metallicity dependence , helps to understand the stark differences in the resulting parameters .
the bottom panel in shows the errors in surface gravities from each approach .
if the classic approximation produced robust error estimates , the values should scatter tightly around the @xmath268 line , instead there is only very weak correlation .
the behaviour is a bit more benign in temperatures than in gravities and metallicities .
while the diverse systematics indicate that our results are not perfect anyway , the big deviations both in the estimated values and their quoted uncertainties show that the traditional approach to the spectroscopic pdf does not provide a suitable approximation .
thus , use of the full information is mandatory .
the method presented in this work is essential for accurate determination of astrophysical parameters of stars .
though the demonstrated scheme is essential to obtain accurate and objective error determinationsthe computational cost is affordable .
the algorithm is parallelized and without efforts to make it more efficient took about @xmath232 cpu - minutes per star way to extract information from the current and upcoming galactic surveys , several shortcomings need and will be addressed .
* we are working on extending the grids of stellar spectrum models , i.e. wider wavelength coverage ( uv to ir ) and finer grid resolution , inclusion of @xmath96-enhancement and rotation as extra dimensions in the grids . * especially on the low - resolution side , the continuum finding algorithms need to be improved . *
parameters , like micro - turbulence , which in fact parametrise the deficits of the current 1d - models in physical realism , must be better constrained or best be made obsolete by the use of more physical models . in the short and intermediate range
we will find smoother corrections on a denser grid that allow for more precise evaluation . in the far future , this problem should be solved by better physics , i.e. 3d - nlte calculations for stars , which are at present still too costly . * it is also interesting to include age- and mass- sensitive diagnostics ( such as , ca uv lines ) , that would in principle allow us to choose spectroscopic models which are more appropriate in a given domain of the hrd . at present
, the analysis of oba stars relies on nlte model atmospheres , whereas lte models are standard for fgk stars .
* the stellar evolution models still apply rather simplistic ( frequently grey ) atmosphere models .
consequent systematics can be explored via residuals from this bayesian method e.g. in magnitude space , as well as aberrations of physical parameters .
a long - term goal would be to gear the stellar evolution codes with the same atmosphere models used for the spectroscopic modelling to avoid biases by partly contradictory models . *
the photometric information in our scheme is affected by reddening .
colour distortions and mismatches between the photometric and other information can be directly used to determine reddening , in addition spectral information can be extracted at high resolution e.g. from interstellar na d lines .
since we simultaneously derive probability distribution functions for stellar distances , the method can be adapted for reddening reconstructions ( like the ones by * ? ? ?
in this paper we present the first generalised bayesian approach for stellar parameter determination .
the essence of the bayesian method is a combination of several probability distribution functions in the multi - dimensional parameter space , which can be expanded arbitrarily depending on a ) the available observational information for a star , and b ) the desired physical quantities .
the presented framework simultaneously evaluates the spectroscopic informations ( gained from comparisons to theoretical spectra ) and all other sources of information .
this allows to calculate the full probability distributions in parameter space and helps to cut computational costs by pre - constraining the parameter space that has to be searched with the spectroscopic method . in this work
we showed how to combine low or high - resolution spectroscopy , photometry , parallax measurements and reddening estimates to estimate central physical parameters @xmath269 of a star , as well as its mass , age , distance , or detailed chemical composition .
the exploitation of theoretical constraints like stellar models , as well as strong mutual dependence or independence of different parameters reduce the complexity and effective dimensionality of the problem and make the computation possible .
the scheme can be easily expanded to other sources of information , in particular to astroseismic e.g. from corot or kepler .
the presented method has unique advantages compared to other available approaches : * it makes an optimal and unbiased use of all observational data and theoretical information for a star , thus providing the parameter estimates that satisfy all observational constraints ; * the method is robust with respect to missing data , such as low quality or missing spectral or photometric information . * the method is vital to gain a grip on derived quantities .
e.g. to determine the distance of a star , it is not sufficient to know its best - fit values for surface gravity , temperature , metallicity and their errors ; a fair assessment is only possible if we know the full combined pdf in all parameters .
we showed that indeed the bayesian estimates in particular for uncertainties differ from simple expectations .
* data from different surveys can be analysed with exactly the same scheme : stellar models are available in most photometric systems and the synthetic spectra grids can be folded with any instrument response function .
this avoids systematic offsets caused by applying different analysis methods to different surveys and the bayesian method can serve as a benchmark for cross - calibration between surveys .
we compared our approach to the results of a traditional bayesian analysis on the segue sample .
we use the same photometric input , priors and even spectroscopic analysis , but approximate the spectroscopic pdf by a gaussian distribution , as usually done in the literature .
we find substantial shifts in all parameters , frequently by several standard deviations .
this demonstrates that neglect of the full pdfs leads to wrong parameter estimates and unreliable estimates of their errors .
use of our or an equivalent method , which is able to map out the true shape of the full spectroscopic ( or any other ) pdf , is hence mandatory for any analysis of stellar parameters .
the method requires unbiased assessments from all its sources of information . however , we know that systematic biases ( e.g. theoretical atmosphere flaws , stellar evolution uncertainties like convection , nuclear reaction rates , etc . ) currently affect these sources .
this vulnerability can bias the entire derived parameter set . to test the performance of our method we compared both to reference samples for low - resolution and for high - resolution spectra . in all cases where we encounter problems ,
e.g. lower spectroscopic gravities , the bayesian method remains robust and pushes all values towards the benchmark .
comparisons with each astroseismically and traditionally derived parameters shows that the bayesian method provides excellent results on the astroseismic sample and clearly superior performance compared to the traditionally derived reference .
we provide parameter estimations for these stars in table [ tab : allhighres ] . similarly the photometric information is affected by reddening .
however , this impact can be directly used to determine reddening especially in a larger sample . by the simultaneous determination of distance distributions
, the method offers an excellent basis for reddening measurements similar to @xcite . up to the last decade
, sample sizes of galactic surveys determined the scope of model comparisons : at sample sizes of @xmath0 stars , poisson noise was usually of the same importance as systematic uncertainties and knowledge of the detailed error distributions . in the future we can advance from a more qualitative understanding of best - fit parameters for our galaxies to full quantitative analysis .
the implies , however , that progress in evaluating the upcoming and present large stellar surveys for the milky way critically depends on our ability to cope both with the systematic biases and more importantly derive precise and accurate error distributions , and hence on the development and success of methods like the presented .
we thank the referee for a very helpful report and for suggesting the comparison to the classical bayesian scheme .
it is a pleasure to thank david weinberg and sergey koposov for fruitful discussions and advice and james binney for helpful comments to the text .
we thank u. heiter , p. jofre , and s. cuaresma for providing the observed high - resolution data , and t. gehren , and f. grupp for providing stellar atmosphere models used in this work .
r.s . acknowledges financial support by nasa through hubble fellowship grant hf-@xmath270 awarded by the space telescope science institute , which is operated by the association of universities for research in astronomy , inc . ,
for nasa , under contract nas 5 - 26555 .
this work was partly supported by the european union fp7 programme through erc grant number @xmath271 .
we thank for the great hospitality of the aspen physics center , where parts of this paper were written .
et al . , 2012 , apjs , 203 , 21 allende prieto , c. , sivarani , t. , beers , t. c. , et al .
2008 , , 136 , 2070 an d. et al .
, 2008 , apjs , 179 , 326 an d. et al . , 2013 ,
apj , 763 , 65 rnadttir a.s . , feltzing s. , lundstrm i. , 2010 , a&a , 521 , 40 aumer m. , binney j. , 2009 , mnras , 397 , 1286 bailer - jones c. , 2011 , mnras , 411 , 435 bergemann m. , gehren t. , 2008 , a&a , 492 , 823 bergemann m. , lind k. , collet r. , magic z. , asplund m. , 2012 , mnras , 427 , 27 bevington p.r . ,
robertson d.k . , 1992 ,
_ data reduction and error analysis for the physical sciences _ , mcgraw - hill , new york binney j. , merrifield m. , 1998 , _ galactic astronomy _ ,
pup , princeton , nj binney j. et al . , 2013 , mnras , tmp.2584b , arxiv : 1309.4270 blanco - cuaresma s. , soubiran c. , jofr p. , heiter u. , 2014 , arxiv : 1403.3090 burnett b. , binney j. , 2010 , mnras , 407 , 339 casagrande l. , schnrich r. , asplund m. , cassisi s. , ramrez i. , melndez j. , bensby t. , feltzing s. , 2011 , a&a , 530 , 138 chaplin w.j . et al . , 2011 , apj , 713 , 169 chiavassa a , collet r. , casagrande l. , asplund m. , 2010 , a&a , 524 , 93 chieffi a. , straniero o. , salaris m. , in _ the formation and evolution of star clusters _ , ed
. k. janes , aspcs , 13 , 219 drell p.s . , loredo t.j . , wasserman i. , 2000 , apj , 530 , 593 fuhrmann , k. , 2004 , an , 325 , 3 gehren t. , liang y.c .
, shi j.r . , zhang h.w .
zhao g. , 2004 , a&a , 413 , 1045 gilmore g. et al . , 2012 , msngr , 147 , 25 girardi l. , grebel e.k . , odenkirchen m. , chiosi c. , 2004 , a&a , 422 , 205 gratton r. , 2000 , aspc , 198 , 225 grupp f. , 2004 , a&a , 420 , 289 grupp f. , 2004 , a&a , 426 , 309 hekker s. , melndez j. , 2007 , a&a , 475 , 1003 ivezi et al . , 2008 , apj , 684 , 287 jofre p. et al . , 2013 , arxiv:1309.1099 johnson h.l .
, mitchell r.i . , iriarte b. , wisniewski w.z . , 1966 , colpl , 4 , 99 jrgensen b. , lindegren l. , 2005 , a&a , 436 , 127 just a. , jahrreiss h. , 2007 , arxiv:0706.3850 kitaura f.s . , enlin t.a . , 2008 , mnras , 389 , 497 koen c. , kilkenny d. , van wyk f. , marang f. , 2010 , mnras , 403 , 1949 korn a.j . , shi j. , gehren t. , 2003 , a&a , 407 , 691 kurucz r.l . , 2005 , memorie della societ astronomica italiana supplement , 8 , 14 laney c.d . , joner m.d
, pietrzyski g. , 2012 , mnras , 419 , 1637 lee y.s .
et al . , 2008 ,
aj , 136 , 2022 lee y.s .
et al . , 2008 ,
aj , 136 , 2050 lindley d.v . , 1982 , academic press , london , _ the bayesian approach to statistics , in : some recent advances in statistics _ , eds . j. tiago de oliviera and b. epstein liu c. , bailer - jones c.a.l . , sordo r. , vallenari a. , borrachero r. , luri x. , sartoretti p. , 2012 ,
mnras , 426 , 2463 madau p. , pozzetti l. , dickinson m. , 1998 , apj , 498 , 106 magic z. , serenelli a. , weiss a. , chaboyer b. , 2010 , apj , 718 , 1378 magic z. collet r. asplund m. , trampedach r. , jayek w. , chiavassa a. , stein r. , nordlund a. , 2013 , a&a , 557 , 26 majewski , s. r. , skrutskie , m. f. , schiavon , r. p. , et al .
2007 , bulletin of the american astronomical society , 39 , # 132.08 marconi m. , cignoni m. , di criscienzo m. , ripepi v. , castelli f. , musella i. , ruoppo a. , 2006 , mnras , 371 , 1503 mcmillan p. , 2011
, mnras nordstrm b. et al . , 2004 , a&a , 418 , 989 nehag a. , korn a. , gustafsson b. , stempels e. , vandenberg d.a . , 2011 ,
a&a , 528 , a85 perryman m. , 1997 , a&a , 323 , 49 pietrinferni a. , cassisi s. , salaris m. , castelli f. , 2004 , apj , 612 , 168 pietrinferni a. , cassisi s. , salaris m. , castelli f. , 2006 , apj , 642 , 797 pietrinferni a. , cassisi s. , salaris m. , percival s. , ferguson j.w . , 2009 , apj , 697 , 275 plez b. , 2012 , ascl:1205.004 pont f. , eyer l. , 2004 , mnras , 351 , 487 ramrez i. et al . , 2012 ,
apj , 752 , 5 reetz j. , 1999 , ap&ss , 265 , 171 rix h .- w . , bovy j. , 2013 , a&arv , 21 , 61 ruchti g. , bergemann m. , serenelli a. , casagrande l. , lind k. , 2013 , mnras , 429 , 126 salaris m. , weiss a. , 1998 , a&a , 335 , 943 salpeter e. , 1955 , apj , 121 , 161 schlafly e. , green g. , finkbeiner d.p . , 2013 , aas , 22114506 schlegel d.j .
, finkbeiner d.p . , davis m. , 1998 , apj , 500 , 525 schnrich r. , 2012 , mnras , 427 , 274 schnrich r. , binney j. , 2009 , mnras , 396 , 203 serenelli a. , bergemann m. , ruchti g. , casagrande l. , 2013 , mnras , 429 , 3645 shi j.r . , gehren t. , zeng j.l . , mashonkina l. , zhao g. , 2014 , apj , 782 , 80 shkedy z. , decin l. , molenberghs g. , aerts c. , 2007 , mnras , 377 , 120 skrutskie et al . , 2006 , aj , 131 , 1163 sneden c. , 1973 , apj , 184 , 839 steinmetz m. et al . , 2006 , aj , 132 , 1645 stello d. , kjeldsen h. , bedding t.r .
, buzasi d. , 2006 , a&a , 448 , 709 van leeuwen f. , 2007 , a&a , 474 , 653 yanny b. et al . , 2009 ,
aj , 137 , 4377
previous approaches ( e.g. , @xcite ) introduced a selection function . with our choice of symbols
, this would read : @xmath272 where @xmath117 denotes the selection function .
@xcite then split the selection function into two parts : the one that depends on the parameters @xmath4 and the other one , that does not and is thus of no importance .
however , there appears to be no reason to introduce the other term : selections of a sample are nearly almost made on observations and not on stellar parameters that are not known a priori .
the one example of such a selection function acting on parameter space we could find in the literature , is actually based on a misunderstanding by @xcite : knowing the available parallax measurement and its error for a star , they try to mimic a typical kinematic quality cut in a sample by zeroing all probability that produces too low parallaxes in proportion to the measured parallax error .
however , it is not clear why one should not use the full parallax information here : applying the selection function implies that one has the knowledge necessary to compute the full likelihood , the selection function instead gives an undesirable one - sided constraint against far - away stars , and when pretending not to have the parallax information for testing purposes , the selection function will arbitrarily cut away the tail of effective distance overestimates , leading to wrong confidence and biased error estimates .
for the metallicity - iron abundance prior we assume a fixed alpha enhancement .
it is known that also alpha enhanced stellar models are very well approximated by scaled solar abundance models ( cf .
* ; * ? ? ?
we use this fact by setting the relation : @xmath273 for the sake of simplicity we give each population the same upper limit of @xmath276 and allow for a constant density in age down to @xmath277 .
cosmological studies as well as observations in the milky way disc @xcite measure a significant decline of star formation rates with time even for galactic disc stars . observations and these theoretical models also derive a significantly older age for more metal - poor populations , which motivates the decreasing time constant towards lower metallicities .
the high altitude of the sdss / segue sample additionally favours older ages ( cf . * ? ? ?
* ) , but in order not to conflict with cromwell s rule on the other hand , we lean towards a relatively moderate decline with time .
segue measures mostly stars in the high disc , so we describe the spatial distribution for our stars by a primitive thick disc plus halo model , i.e. : @xmath278 where @xmath28 is the cylindrical galactocentric radial coordinate , @xmath75 the galactocentric distance , @xmath279 the altitude above the plane , @xmath280 the assumed scale height of the galactic disc , @xmath281 the scale length of the galactic disc , @xmath282 the assumed galactocentric distance of the sun from @xcite .
lccccccccccl * name * & * hip * & * spectra * & * @xmath34 * & * @xmath283 * & * @xmath27 * & * @xmath284 * & * @xmath199 * & * @xmath285 * & * @xmath19 * & * @xmath286 * & * remark * + + hd 107238 & 60172 & 0 & -0.14 & 0.26 & 4473 & 81 & 2.04 & 0.20 & 6.3 & 3.3 & phot .
+ hd 122563 & 68594 & 4 & -2.650 & 0.076 & 4809 & 47 & 1.54 & 0.13 & 9.7 & 2.9 & comb .
, bad photometric t + hd 140283 & 76976 & 4 & -2.73 & 0.11 & 5608 & 40 & 3.539 & 0.056 & 13.49 & 0.47 & comb . +
hd 173819 & 92202 & 0 & -0.42 & 0.64 & 4240 & 95 & 1.05 & 0.25 & 2.9 & 4.0 & phot + hd 190056 & 98842 & 0 & -0.11 & 0.29 & 4449 & 94 & 2.07 & 0.17 & 5.8 & 4.0 & phot + hd 220009 & 115227 & 0 & -0.43 & 0.28 & 4369 & 85 & 1.67 & 0.19 & 6.2 & 3.9 & phot + hd 22879 & 17147 & 3 & -0.592 & 0.024 * & 6006 & 19 * & 4.316 & 0.044 * & 12.1 & 1.2 * & comb.@xmath287 + hd 84937 & 48152 & 1 & -2.11 & 0.14 & 6242 & 70 & 3.931 & 0.082 & 13.57 & 0.41 & comb.@xmath288 + ksi hya & 56343 & 1 & -0.458 & 0.032 * & 4933 & 35 & 2.476 & 0.080 & 3.9 & 1.6 & comb .
, metallicity fit questionable + procyon & 37279 & 4 & -0.161 & 0.078 & 6515 & 79 & 3.993 & 0.073 & 2.44 & 0.53 & comb . + alpha cen a &
71683 & 2 & 0.275 & 0.063 & 5939 & 79 & 4.380 & 0.066 & 3.7 & 2.5 & comb . + alpha cen b & 71681 & 1 & 0.175 & 0.072 & 5364 & 58 & 4.482 & 0.041 & 6.6 & 4.4 & comb .
+ psi phe & 8837 & 0 & 0.14 & 0.36 & 3586 & 31 * & 0.65 & 0.22 & 4.9 & 4.5 & phot.@xmath289 + sun & 0 & 4 & -0.013 & 0.046 & 5842 & 49 & 4.464 & 0.063 & 4.3 & 2.9 & comb .
+ 18 sco & 79672 & 2 & -0.050 & 0.059 & 5849 & 54 & 4.492 & 0.064 & 4.3 & 3.2 & comb .
+ 61 cyg a & 104214 & 0 & -0.45 & 0.43 & 4563 & 83 & 4.717 & 0.060 & 6.8 & 4.0 & phot .
+ alpha tau & 21421 & 0 & 0.09 & 0.22 & 3889 & 57 & 1.21 & 0.14 & 5.9 & 3.8 & phot .
+ arcturus & 69673 & 0 & -0.27 & 0.31 & 4399 & 91 & 1.82 & 0.19 & 4.2 & 1.9 & phot
. + alpha cet & 14135 & 0 & -0.53 & 0.26 & 3723 & 41 & 0.50 & 0.16 & 5.5 & 3.7 & phot .
+ tau cet & 8102 & 1 & -0.520 & 0.047 & 5515 & 32 & 4.612 & 0.053 & 7.8 & 3.9 & comb .
+ beta ara & 85258 & 0 & -0.07 & 0.36 & 4118 & 83 & 1.02 & 0.20 & 3.3 & 1.3 & phot
. + mu ara & 86796 & 1 & 0.379 & 0.073 & 5950 & 97 & 4.334 & 0.077 & 3.9 & 2.3 & comb .
+ pollux & 37826 & 1 & -0.376 & 0.043 & 4846 & 53 & 2.66 & 0.12 & 4.3 & 2.1 & comb . ,
high macroturbulence + eps for & 14086 & 2 & -0.479 & 0.049 & 5218 & 68 & 3.614 & 0.072 & 7.2 & 2.0 & comb .
+ eps vir & 63608 & 2 & -0.457 & 0.050 & 5024 & 58 & 2.62 & 0.12 & 1.11 & 0.75 & comb . ,
high macroturbulence + beta vir & 57757 & 1 & -0.037 & 0.080 & 6225 & 93 & 4.163 & 0.077 & 3.8 & 1.1 & comb .
+ eta boo & 67927 & 0 & 0.12 & 0.37 & 6332 & 162 & 3.868 & 0.093 & 2.8 & 1.7 & phot .
, fast rotator + delta eri & 17378 & 2 & 0.047 & 0.050 & 5139 & 59 & 3.791 & 0.071 & 6.99 & 0.88 & comb.@xmath290 + eps eri & 16537 & 3 & -0.202 & 0.050 & 5184 & 27 & 4.562 & 0.049 & 6.5 & 3.8 & comb .
+ gam sge & 98337 & 0 & -0.13 & 0.25 & 3942 & 83 & 1.15 & 0.19 & 5.6 & 3.9 & phot .
+ gmb 1830 & 57939 & 1 & -1.56 & 0.11 & 5304 & 36 & 4.649 & 0.060 & 7.8 & 4.1 & comb .
+ mu cas & 5336 & 1 & -0.598 & 0.010 * & 5584 & 39 & 4.601 & 0.053 & 5.6 & 3.6 & comb.@xmath287 + mu leo & 48455 & 0 & 0.24 & 0.22 & 4607 & 74 & 2.43 & 0.11 & 4.7 & 3.7 & phot .
, low @xmath27 and rotating + beta hyi & 2021 & 3 & -0.189 & 0.081 & 5848 & 79 & 3.997 & 0.074 & 6.89 & 0,58 & comb .
|
we present a unified framework to derive fundamental stellar parameters by combining all available observational and theoretical information for a star .
the algorithm relies on the method of bayesian inference , which for the first time directly integrates the spectroscopic analysis pipeline based on the global spectrum synthesis and allows for comprehensive and objective error calculations given the priors .
arbitrary input datasets can be included into our analysis and other stellar quantities , in addition to stellar age , effective temperature , surface gravity , and metallicity , can be computed on demand .
we lay out the mathematical framework of the method and apply it to several observational datasets , including high- and low - resolution spectra ( uves , narval , harps , sdss / segue ) .
we find that simpler approximations for the spectroscopic pdf , which are inherent to past bayesian approaches , lead to deviations of several standard deviations and unreliable errors on the same data . by its flexibility and the simultaneous analysis of multiple independent measurements for a star
, it will be ideal to analyse and cross - calibrate the large ongoing and forthcoming surveys , like gaia - eso , sdss , gaia and lsst .
[ firstpage ] stars : fundamental parameters stars : distances techniques : photometric techniques : spectroscopic methods : statistical methods : data analysis
|
the confinement of color in quantum chromodynamics is explained if color electric fields form flux tubes .
these flux tubes appear readily in the bag model @xcite , but this is a geometric picture rather than a dynamical one .
hooft @xcite and mandelstam @xcite proposed that , just as magnetic flux tubes form in a superconductor @xcite , a condensation of magnetic charge in the qcd vacuum would lead to the formation of electric flux tubes and confinement .
this is the dual superconductor picture of confinement .
the analytical study of the confining flux tube depends largely on a classical , abelian model .
one can connect this model to qcd with t hooft s idea of abelian dominance @xcite .
starting in qcd , one fixes an abelian gauge and asserts that the dominant degrees of freedom are abelian gauge fields ( from a cartan subalgebra ) and abelian magnetic monopoles , with weak coupling to the non - abelian gauge fields which assume the guise of self - coupled charged fields .
the assumption that the effective interaction of the monopoles causes their condensation leads immediately to the dual superconductor picture .
the validity of this scenario is a subject of research and debate in the lattice gauge theory community @xcite . in superconductivity ,
one studies the static structure of magnetic flux tubes via a landau - ginzburg theory @xcite .
the simplest landau - ginzburg hamiltonian is that of the abelian higgs model in three dimensions .
classical solutions of the abelian higgs theory have been considered as models of qcd s electric flux tube as well @xcite .
suganuma , sasaki , and toki and their collaborators @xcite , using the formalism proposed by suzuki @xcite , have fixed the effective coupling constants of the abelian higgs model by comparing its flux tube with phenomenology .
there have also been attempts to do so by comparison with the flux tube that emerges in lattice gauge theory @xcite .
success in this program will establish a landau - ginzburg effective hamiltonian of qcd . in this paper
we present a study not of the statics of the confining flux tube , but of its dynamics .
the flux tube of qcd is not a static object .
creation of flux tubes and their subsequent decay through pair creation offer a detailed model of particle creation in @xmath0 , @xmath1 , and @xmath2 collisions @xcite . in the case of nucleus nucleus collisions , the flux tube has an appreciable transverse extent , so that it can be called a `` color capacitor '' if the color field is coherent across it , or a `` color rope '' @xcite if it is not .
@xmath3 pairs and gluons are created through the schwinger pair creation process @xcite , and they screen the field in the flux tube while carrying away the energy in the form of hadrons .
if the density of created particles is large enough , the quarks and gluons form a quark
gluon plasma before the final hadronization takes place .
field - theoretic analysis of pair creation and back - reaction in the flux tube @xcite has shown the buildup of particle density and subsequent plasma oscillations .
the methods were applied first to a field region of infinite spatial extent , rather than to a finite flux tube .
eisenberg @xcite carried out a calculation in a cylindrical flux tube of fixed radius , as suggested by the bag model . in our study here , the flux tube is a dynamical field configuration , not a static geometric object .
following @xcite and @xcite , we set up the electric flux tube in classical electrodynamics coupled to an abelian higgs field via the dual gauge field .
thus the higgs field represents a condensate of magnetic monopoles ; the dual meissner effect confines _ electric _ flux to flux tubes . starting from a static flux tube configuration , we release a density of electric charges in the system and allow them to accelerate and screen the electric field .
the weakening of the electric field allows the flux tube to collapse , but the inertia of the charges carries them into plasma oscillations that build up the field strength again and force open the tube .
the coupling of the plasma oscillations to the higgs field making up the flux tube is the main new feature in our work .
in section [ sec : eqns ] we describe the dual superconductor theory .
the model contains an abelian gauge field governed by maxwell s equations , with coupling to electric charges and currents on the one hand and to a magnetic higgs field on the other .
the electric currents are treated hydrodynamically ; the higgs field is a classical field that confines electric flux to the tube .
the coupling to the higgs field is accomplished by means of a dual vector potential , as introduced by zwanziger @xcite and used subsequently in @xcite for the static problem and in @xcite for phenomenological modeling .
we impose cylindrical symmetry and we assume @xmath4-independence along the flux tube in the region far from its ends .
nevertheless , we are forced to pay some attention to what happens near the ends of the tube in order to define potentials unambiguously .
we present numerical results in section [ sec : results ] .
we set the parameters of the theory in both the type i and type ii regimes , and show how plasma oscillations arise .
we consider plasma frequencies both above and below the cutoff for radiation into the higgs vacuum .
the plasma oscillations are accompanied by changes in the radius of the flux tube as the electric pressure from within decreases through screening , only to increase again as the currents overshoot .
we believe that this paper is the first to address the dynamics of an electric flux tube within this model .
it is amusing to note that this physical situation has no counterpart in superconductivity .
while our magnetic higgs field appears in the usual landau - ginzburg theory as an electrically charged condensate , our electric currents can not appear there because there are no magnetic monopoles in nature .
an abrikosov flux tube has to run all the way to the boundary of the sample , where it joins onto the external magnetic field .
our electric flux tube , on the other hand , has a finite ( but large ) length , and the charges at its ends can be screened by the electric currents that flow in it .
we review in the appendix the well - known phenomenon of flux quantization and why it has no effect on a flux tube of finite length .
our classical model for the flux tube is electrodynamics coupled to a scalar magnetic monopole field .
the equations of motion are maxwell s equations with both electric and magnetic sources , plus the klein - gordon equation for the monopoles . the latter is coupled not to the usual vector potential @xmath5 but to the dual potential @xcite @xmath6 , and contains a self - interaction that puts it into the higgs phase .
we make a cylindrically symmetric _ ansatz _ for the fields , and furthermore assume @xmath4-independence far from the sources at the ends of the tube .
we put the electrically charged sources on the @xmath4 axis at @xmath7 , and we assume that all the space - charge effects are localized there .
the central region is defined by @xmath8 , where @xmath9 is chosen to exclude the end regions ( see fig .
[ fig : geometry ] ) .
maxwell s equations with electric and magnetic currents are @xmath10 where @xmath11 .
( [ mgauss4 ] ) replaces the bianchi identity of ordinary electrodynamics
. we can solve it with a vector potential , but there is a new term , @xmath12 defining an arbitrary vector @xmath13 , one finds that ( [ mgauss4 ] ) is solved if @xmath14 where @xmath15 represents an integration from infinity with suitable boundary conditions .
the field equation for @xmath16 then follows from ( [ egauss4 ] ) . solving maxwell s equations in the opposite order
, the same field strength can be represented alternatively by a _
dual _ vector potential , according to @xmath17 with @xmath18 in order to satisfy ( [ egauss4 ] ) . now
the field equation for @xmath19 follows from ( [ mgauss4 ] ) .
we will return to the vector potentials below . in three - dimensional notation ,
@xmath20 @xmath21 @xmath22 @xmath23 in the aforementioned central region @xmath8 , we make the _ ansatz _
@xmath24 @xmath25 @xmath26 @xmath27 ampre s law ( [ eq : ampere ] ) has only a @xmath28 component , @xmath29 and faraday s law ( [ eq : faraday ] ) has only a @xmath30 component , @xmath31 the two gauss laws show that the charge densities are zero , @xmath32 eqs .
( [ eq : az ] ) and ( [ eq : fth ] ) are the equations of motion for @xmath33 and @xmath34 . near the ends of the flux tube , @xmath35
, we must allow for new components @xmath36 and @xmath37 , as well as for @xmath38 , and all fields will be @xmath4-dependent . in order to calculate the time evolution of the matter fields ,
we need the vector potentials . in fact , we need only the dual potential @xmath19 @xcite , which enters the field equation for the monopole field . from ( [ dual1 ] ) and ( [ dual2 ] ) , @xmath39 where we have chosen @xmath40 .
we choose the gauge @xmath41 .
inverting ( [ eq : eb ] ) gives @xmath42 our _ ansatz _ for @xmath43 then gives @xmath44 , with @xmath45 this integral only gets contributions from the regions around the sources , since @xmath46 far from the charges .
this makes @xmath47 independent of @xmath4 in the central region @xmath8 .
we can relate the integral in ( [ be ] ) to the charge distribution near the sources as follows .
we take a cylindrical surface of radius @xmath48 with ends at @xmath49 and @xmath50 .
gauss law gives for this surface @xmath51 where @xmath52 is the charge inside the cylinder ( composed of the original source plus the space charge around it ) .
noting that @xmath53 is the @xmath4-independent @xmath54 , we obtain the relation @xmath55 and its inverse @xmath56 given @xmath33 and some model for the charge @xmath52 ( see below ) , we can calculate @xmath47 .
the magnetic monopoles are represented by a classical klein - gordon equation with a higgs potential , @xmath57 where @xmath58 we make the _ ansatz _ @xmath59 , giving @xmath60 \rho e^{i\chi}+\lambda\left(\rho^2-v^2\right)\rho e^{i\chi}=0 \>.\ ] ] we assume @xmath61 ( for a flux tube with @xmath62 units of electric flux ) , so @xmath63 \rho+\lambda\left(\rho^2-v^2\right)\rho = 0 \>. \label{eq : kg}\ ] ] this is the equation of motion for @xmath64 .
it feeds back into maxwell s equations via the magnetic current , @xmath65 which contains only a @xmath30 component , @xmath66 as shown in the appendix , @xmath62 represents electric flux coming in from infinity , so that we will set @xmath67 for simplicity .
the flux generated by @xmath68 at the ends of the flux tube is _ not _ quantized .
we represent the electrically charged matter by classical two - fluid magnetohydrodynamics .
the positively and negatively charged fluids both have particle density @xmath69 ( so that @xmath70 ) and their velocities are @xmath71 .
the fluid motion is determined by euler s equations @xmath72
= \pm e \be \pm e \bv^{\pm } \times \bh - \frac{1}{n_e}{\nabla p}\ ] ] and the continuity equation @xmath73 the electric current is given by @xmath74 we assume in the central region that all quantities depend only on @xmath48 and @xmath75 , so that @xmath76 .
recalling that @xmath77 and @xmath78 , we have @xmath79 and @xmath80 not surprisingly , these equations allow the _ ansatz _
@xmath81 we find @xmath82 these equations feed back into maxwell s equations via the current , which is in the @xmath28 direction and has the magnitude @xmath83 we need an equation of state to relate @xmath84 to @xmath85 . having in mind a quark gluon plasma , we choose the equation of state of a relativistic ideal gas with the appropriate number of massless fermions and bosons .
we set the chemical potential to zero and write all quantities in terms of the temperature , @xmath86 then eqs . ( [ eq : veom ] ) can be written in terms of the velocities and temperature as @xmath87 the advantage of a hydrodynamic description of the charged matter is that the fluid interacts directly with the field strengths and . if we were to consider classical or quantum field theory instead of hydrodynamics , we would need the vector potential @xmath88 . in our geometry , the simplest _ ansatz
_ would require two components , @xmath89 and @xmath90 .
the potential would necessarily be @xmath4-dependent in the central region ; moreover , the simplified treatment of the ends of the flux tube ( see below ) would be impossible . for study of the central region ,
all we need to know about the complex regions at the ends of the flux tube is @xmath52 , which represents the charge density built up near @xmath91 by the current @xmath92 .
( the region near @xmath93 is , of course , a mirror image of the region near @xmath91 . ) without dealing in detail with the motion of charges near the ends of the flux tube , we can guess at a few models : 1 .
_ point charge : _ here we just assume that all the current merely accumulates in a point charge at @xmath94 .
then @xmath95 is independent of @xmath48 , and charge conservation gives @xmath96 2 . _ surface charge density : _ here we let the charge pile up on a plate near @xmath94 , giving a surface charge density @xmath97
. then @xmath98 and @xmath99 initial conditions have to be specified for @xmath100 to represent the original source of the flux tube .
both : _ keeping the initial charge pointlike , we put only the accumulated charge into @xmath100
. then @xmath101 and @xmath100 is determined by ( [ charge_cons ] ) as above .
the initial condition is @xmath102 .
note that in this case the space charge will never exactly cancel @xmath103 and thus any plasma oscillations will be asymmetric .
the initial conditions must in any case satisfy gauss law , @xmath104 if there is no flux coming from infinity ( i.e. , @xmath67 ) .
we start off the system in the configuration of a static flux tube with a stationary fluid in it .
we combine the static limit of ( [ eq : fth ] ) with ( [ eq : ebq ] ) and ( [ eq : jgth ] ) to give @xmath105=2g^2\rho^2b \>,\ ] ] where we have chosen @xmath67 .
( [ eq : bqe ] ) gives the boundary condition @xmath106 unless @xmath68 includes a point charge @xmath103 ; in that case @xmath107 as @xmath108 .
( [ eq : bqe ] ) together with ( [ overallgauss ] ) gives @xmath109 at infinity .
the static limit of the klein - gordon equation ( [ eq : kg ] ) is @xmath110 \rho + \lambda\left(\rho^2-v^2\right)\rho = 0 \>,\ ] ] with the boundary conditions that @xmath64 is zero at the origin and tends to @xmath111 at infinity .
focusing on the case where @xmath68 is a point charge , we define the reduced field @xmath112 via @xmath113 giving @xmath114 with @xmath115 and @xmath116 , and @xmath117\rho=0 \>. \label{eq : rhostat}\ ] ] these equations determine @xmath64 and @xmath47 , and hence @xmath33 , for the static flux tube .
clearly @xmath118 here .
we determine the time evolution of the system through the following system of equations . the maxwell equations ( [ eq : az ] ) and ( [ eq : fth ] ) give and ; the mhd equations ( [ eq : vteom ] ) give and @xmath119 , and hence @xmath92 via ( [ eq : jez ] ) and ( [ eq : state ] ) . eq .
( [ eq : qptchg ] ) gives the charge @xmath68 whence ( [ eq : bqe ] ) gives the vector potential .
the scalar field @xmath64 evolves according to the klein - gordon equation ( [ eq : kg ] ) , and ( [ eq : jgth ] ) gives the magnetic current @xmath120 , the last ingredient for the maxwell equations .
the initial conditions , as noted , consist of the static flux tube with an initial value of @xmath68 , and of a static fluid distribution specified by @xmath121 . we choose four sets of parameters , listed in table [
table : params ] .
the magnetic charge @xmath122 and the vacuum expectation value @xmath123 determine the vector mass @xmath124 and its reciprocal , the london penetration depth @xmath125 .
the scalar self - coupling @xmath126 determines , along with @xmath127 , the ( dual ) higgs mass @xmath128 and the ginzburg - landau coherence length @xmath129 .
the ratio of these lengths is @xmath130 .
if this value is smaller than one then the superconductor is of type i ; otherwise it is of type ii . in ordinary superconductors , flux tubes
are observed only in type ii materials , when the applied magnetic field lies between the critical values @xmath131 and @xmath132 and penetrates via creation of an abrikosov lattice .
type i materials , on the other hand , expel the field entirely as long as superconductivity persists .
the situation would be different if one were to introduce a pair of magnetic monopoles .
then the magnetic field would be forced into a flux tube between the monopoles , even in a type i material . the difference between type i and ii would lie in the stability of the flux tube against splitting into a lattice ( a `` rope '' ) of smaller tubes : type i tubes would be stable , while type ii tubes would split .
this splitting is presumably inaccessible from our cylindrically symmetric _
the values of @xmath126 and @xmath122 fixed phenomenologically in @xcite are in the type ii region , and we choose these values for our type ii cases . for type
i , we increase @xmath122 and decrease @xmath126 to make @xmath133 .
we take @xmath134 mev from @xcite as well .
we fix the electric charge @xmath135 according to the dirac quantization condition @xmath136 , even though this is not meaningful for continuum hydrodynamics .
the charged fluid presents a conundrum .
our intuition about confinement suggests that the superconducting vacuum should expel this matter , and thus confine it to the flux tube .
( this happens in the friedberg
lee model @xcite . ) in this abelian theory , however , the locally neutral fluid is _ not _ confined . if the initial conditions contain a fluid inside the flux tube only , it will flow outward in a hydrodynamic rarefaction wave .
perhaps the fluid outside the flux tube may be interpreted as a hadronic fluid . in any case , we prefer not to superimpose radial hydrodynamic flow on the oscillations of the flux tube .
thus we choose a _ uniform _ initial fluid density , outside the flux tube as well as inside .
having in mind a quark gluon plasma , we set the initial temperature to be spatially uniform with @xmath137 = 150 mev .
the initial value @xmath103 of the charge at the ends of the flux tube gives the initial amount of electric flux . in a @xmath1 collision
, one would have @xmath138 , corresponding to the creation of a flux tube in the fundamental representation ; an @xmath139 collision between heavy nuclei would give@xcite @xmath140 , which reaches @xmath141 for large nuclei
. a large value of @xmath103 will give a thick flux tube and large - amplitude plasma oscillations , and thus enhance the non - linear effects due to the higgs coupling .
unfortunately , very large values of @xmath103 are outside the range of stability of our numerics ; we choose values that are as large as possible given this constraint .
the plasma frequency is @xmath142 where the factor of two reflects the fact that we have two fluids . for given @xmath85 and @xmath135 ,
we fix @xmath143 so as to tune @xmath144 to either side of the vector mass @xmath145 ; thus the plasma oscillations will occur at frequencies either above or below the threshold for radiation .
we expect _ a priori _ that @xmath143 will be in the neighborhood of the dynamical mass @xmath146 of light particles in a heat bath .
the parameters used in @xcite give a rather small vector mass , and with @xmath135 = @xmath147 the condition @xmath148 leads to an unreasonably large value of @xmath143 .
instead we choose in this case to lower the value of the electric charge to @xmath149 .
[ [ case-1 . ] ] case 1 .
+ + + + + + + fig .
[ case1e ] shows plasma oscillations in the on - axis electric field for the type i superconductor with @xmath150 .
the oscillations are clearly nonlinear , with varying amplitude and misshapen waveform .
[ case1fields ] contains snapshots taken at the times of maxima in @xmath151 .
these snapshots show radiation emanating from the flux tube in both @xmath33 and @xmath64 .
the plots of @xmath152 show that the oscillating electric flux in the outgoing wave is as large in amplitude as the initial flux distribution .
it is difficult to tell , however , whether the wave amplitude decays as @xmath153 , as expected for a true propagating wave , or as @xmath154 , which would make the wave evanescent .
since the frequency of the plasma oscillations is below the vector mass , this can not be simple linear radiation .
the third row of fig .
[ case1fields ] , indeed , shows strong coupling of the nonlinear @xmath64 field to the electromagnetic wave .
the origin of this coupling , shown in fig .
[ case1rho ] , is the collapse of the flux tube under the pressure of the @xmath64 field when the electric field is weak .
the figure shows snapshots of @xmath64 taken in the first half - cycle of the evolution . the middle snapshot , taken when @xmath155 , shows appreciable narrowing of the flux tube core even as @xmath64 is driven downward at larger radii in the first oscillation of the outgoing wave .
[ [ case-2 . ] ] case 2 .
+ + + + + + + oscillations in the type i superconductor with a larger plasma frequency , @xmath156 , are shown in figs .
[ case2e ] and [ case2fields ] . here
the irregularity of the waveform of @xmath151 is due to ordinary radiation in interaction with a more - or - less static flux tube wall .
again there is appreciable electric flux radiated outward , but there is little effect on the @xmath64 field .
this is because the latter can not respond when driven by the high - frequency electromagnetic oscillations .
we note that the larger value of @xmath157 here creates a thicker flux tube and keeps more flux in the flux tube despite the radiation .
[ [ case-3 . ] ] case 3 .
+ + + + + + + the type ii superconductor with @xmath150 is similar in behavior to the type i system .
the smaller plasma frequency makes it difficult to calculate through more than two or three oscillations , in spite of the high stability of the crank - nicholson algorithm we use . in fig .
[ case3e ] we see strong deformation of the waveform . fig .
[ case3fields ] shows radiation similar to the type i case , but the perturbation of the @xmath64 field is less apparent .
[ [ case-4 . ] ] case 4 .
+ + + + + + + finally , the corresponding plots for the type ii superconductor with @xmath156 show considerable structure . recall that this parameter set is that used in @xcite for phenomenological fits to static quantities .
the oscillations shown in fig .
[ case4e ] are too rapid to couple strongly to @xmath64 , and we see strong electromagnetic radiation ( figs . [ case4fields ] and [ case4fields_b ] ) accompanied by weak high - momentum disturbances in @xmath64 .
our numerical results raise difficulties for the dual superconductor picture of confinement . as a static model ,
the dual superconductor does indeed form flux tubes that confine charges in string - like configurations .
once the dynamics are examined , however , the lack of absolute color confinement becomes apparent . while the motion of neutral particle matter outside the flux tube may be passed off as the emission of color - neutral hadrons , the radiation of appreciable electric flux can not .
the electric field in t hooft s abelian projection is after all a _ color _ field , representing a coherent , colored gluon state .
it was predictable that oscillations with @xmath156 would radiate into the higgs vacuum , since the photon does have a less - than - infinite mass .
one might be tempted to restrict application of the dual superconductor model to situations where frequencies are much less than @xmath145 .
we have seen , however , that this is insufficient .
even low - frequency plasma oscillations , where radiation should be impossible , succeed in spreading out the electric flux via nonlinear effects .
thus the model is not reliable even in this regime .
it is possible that the model can be made applicable to low - frequency physics by choosing a monopole lagrangian more general than the @xmath158 of the simplest landau - ginzburg theory .
absolute confinement , however , will never be realized in this way .
( the introduction of a strongly nonlinear dielectric constant is what enables the friedberg
lee model to confine all color fields , but this model has not been derived from qcd . )
t hooft s abelian reduction rests on the identification of the important degrees of freedom in an abelian gauge .
it is supposed that the magnetic monopoles have a strong self - interaction , leading to their condensation ; that the abelian gluons , belonging to the cartan subalgebra , turn this condensate into a superconductor ; and , most important , that the off - diagonal gluons are irrelevant to confinement , except insofar as they screen zero - triality states .
these off - diagonal gluons , however , retain all the self - couplings of the original non - abelian gauge theory .
we conjecture that they are essential to understanding time - dependent phenomena related to confinement .
we thank dr .
alex kovner and prof .
eli turkel for their assistance .
this work was supported by the israel science foundation under grant no .
255/96 - 1 and by the basic research fund of tel aviv university .
flux quantization , as it turns out , affects only flux coming in from infinity and not the flux due to the charge @xmath68 at the ends of the flux tube .
the quantization condition comes from demanding that the total energy be finite @xcite .
the energy of the monopole field contains the term @xmath159 as @xmath160 , we have @xmath161 and @xmath162 , and thus the integrand approaches @xmath163 in order that @xmath164 be finite , we must have at large @xmath48 that @xmath165 since @xmath166 is only determined up to a multiple of @xmath167 , it can gain such a multiple when one goes around the circle . integrating ( [ circle ] ) over the circle at fixed
@xmath48 gives @xmath168 where @xmath62 is an integer . in view of ( [ eq : eb ] ) ,
@xmath169 where @xmath170 is the total electric flux .
thus @xmath171 only the _ external _ flux , the flux that does not end at @xmath68 , is quantized in units of @xmath172 .
choosing @xmath173 , i.e. , no external flux , means setting @xmath67 .
a. s. kronfeld , m. l. laursen , g. schierholz , and u. j. wiese , phys .
* 198b * , 516 ( 1987 ) ; a. s. kronfeld , g. schierholz , and u. j. wiese , nucl . phys .
* b293 * , 461 ( 1987 ) ; m. i. polykarpov , nucl . phys .
. suppl . ) * 53 * , 134 ( 1997 ) ; l. del debbio , m. faber , j. greensite , and .
olejnk , _ ibid . _ * 53 * , 141 ( 1997 ) , hep - lat/9607053 .
k. suzuki and h. toki , int .
a11 * , 2677 ( 1996 ) ; s. sasaki , h. suganuma , and h. toki , prog .
. phys . * 94 * , 373 ( 1995 ) ; i. ichie , h. suganuma , and h. toki , , 2944 ( 1995 ) ; _ ibid . _
* 54 * , 3382 ( 1996 ) ; h. monden , h. suganuma , h. ichie , h. toki , phys . rev .
c * 57 * , 2564 ( 1998 ) ; y. koma , h. suganuma , and h. toki , hep - ph/9902441 .
ldddd & type i & type i & type ii & type ii + & @xmath148 & @xmath174 & @xmath148 & @xmath174 + @xmath126 & 10 .
& 10 . & 25 . & 25 .
+ @xmath122 & 5.5 & 5.5 & 2.3 & 2.3 + @xmath135 & 2.3 & 2.3 & 1.0 & 5.5 + @xmath145 ( mev ) & 975 . & 975 . & 411 . & 411 .
+ @xmath175 ( mev ) & 565 . & 565 . & 893 . & 893 .
+ @xmath176 ( fm ) & 0.20 & 0.20 & 0.48 & 0.48 + @xmath177 ( fm ) & 0.35 & 0.35 & 0.22 & 0.22 + @xmath178 & 0.58 & 0.58 & 2.18 & 2.18 + @xmath143 ( mev ) & 250 . & 50 . & 250 . & 50
. + @xmath144 ( mev ) & 790 . & 1766 . & 343 .
+ @xmath179 ( fm ) & 1.57 & 0.70 & 3.60 & 0.30 + @xmath103 & 2.3 & 4.6 & 1.0 & 4.6 +
|
we study plasma oscillations in a flux tube of the dual superconductor model of t hooft and mandelstam . a magnetic condensate is coupled to an electromagnetic field by its dual vector potential , and fixed electric charges set up a flux tube .
an electrically charged fluid ( a quark plasma ) flows in the tube and screens the fixed charges via plasma oscillations .
we investigate both type i and type ii superconductors , with plasma frequencies both above and below the threshold for radiation into the higgs vacuum .
we find strong radiation of electric flux into the superconductor in all regimes , and argue that this invalidates the use of the simplest dual superconductor model for dynamical problems .
|
the use of idempotents in the construction of cyclic error correcting codes is well established and the resulting literature is extensive ( for example , see @xcite,@xcite,@xcite ) .
the basic building blocks for this theory are the primitive idempotents .
any cyclic code may be described by a unique idempotent and this idempotent is a sum of primitive idempotents . for binary cyclic codes ,
efficient algorithms exist for the calculation of these primitive idempotents .
another way of constructing idempotents in the binary case is by using cyclotomic cosets and it was this property which was exploited by shibuya and sakaniwa in @xcite .
their goal was to use idempotents to construct parity check matrices for ldpc codes which have no cycles of length 4 in their factor graphs . at the heart of their technique
is a lemma which is a variation of a result used by weldon @xcite , for the construction of difference set cyclic codes . using this lemma and a subsequent theorem
, they were able to simplify the problem of determining which of the idempotents that are constructed , using a single cyclotomic coset , do not have cycles of length 4 .
they then extended this theory to more general idempotents .
this approach to the construction of ldpc codes has the great advantage of simplicity , the parity check matrices depend only upon the correct choice of cyclotomic cosets and these are very easily calculated .
however , we believe that this advantage is offset by some fundamental weaknesses . whilst the absence of 4 cycles is a desirable objective in the construction of ldpc codes it is not mandatory @xcite , since there are some good codes which do not have this property .
an example of such a code is included in this paper .
the code rate is also an important property of codes but , as shibuya and sakaniwa admit in their conclusion , the codes which they construct in this way are _ expected _ to have a large minimum distance _ at the expense of rate_(our italics ) .
the minimum distance of a code is a crucial property but there is no indication in @xcite of how either a single cyclotomic coset , or combinations of more than one cyclotomic cosets , should be chosen to guarantee that the code constructed has a large minimum distance . in order to address the question of how to choose idempotents which will produce good ldpc codes we propose an entirely different route . as in @xcite
, we shall deal exclusively with binary cyclic codes . making effective use of the mattson - solomon polynomial
, we produce an algorithm which not only allows us to choose , in a systematic way , idempotents with low weight , and therefore a correspondingly sparse parity check matrix , but also with the desirable features that the corresponding codes have a high code rate and a large minimum distance .
this paper is organised as follows . in
we shall review the necessary theory and explain how it will be used to provide an algorithm for the determination of idempotents which may be used to construct good codes . in the design and implementation of this algorithm
is given and then , in , some of the results are displayed . finally , in , we draw our conclusions on this approach .
let @xmath0 , @xmath1 be a positive integer and @xmath2 be the splitting field for @xmath3 over @xmath4 . let @xmath5 be a primitive @xmath1th root of unity and let @xmath6 be the polynomials in @xmath7 $ ] of degree @xmath8 . if @xmath9 then the map @xmath10 is defined by @xmath11(z)=\sum_{j=1}^na(\alpha^j)z^{n - j}\]]and @xmath12 is the mattson - solomon polynomial of @xmath13 ( see @xcite ) .
( we use @xmath14 and @xmath15 for the polynomial variables to distinguish between the polynomials in the domain and codomain of @xmath16 . )
if @xmath17 is multiplication of polynomials@xmath18 and @xmath19 is defined on @xmath20 by the rule @xmath21 then it is well known @xcite,@xcite , that @xmath22 is an isomorphism of rings , in particular it is an isomorphism of the additive groups .
if @xmath23 is the subset of @xmath6 consisting of polynomials with coefficients in @xmath24 ( binary polynomials ) and @xmath25 is the subset of @xmath6 consisting of idempotents , both of these subsets are _ additive subgroups _ of @xmath6 .
it is easy to show ( see @xcite ) that @xmath26are both isomorphisms and from this it is obvious that @xmath27is also an isomorphism .
suppose that @xmath28 is a binary idempotent which is used to construct a parity check matrix for a cyclic code .
the parity check matrix is constructed from the @xmath1-cyclic shifts of @xmath28 @xcite , and so for the resulting code to be a ldpc code , @xmath28 must have low weight . if @xmath29 and @xmath30 , then @xmath31 is the generator of the cyclic code .
if the generator , @xmath31 , has degree @xmath32 , the dimension of the code is @xmath33 and the larger the value of @xmath33 , the better the code rate .
since @xmath31 is a divisor of @xmath3 , all of the zeros of @xmath31 are @xmath1th roots of unity , and there are @xmath32 of these .
further , gcd@xmath34=1 and @xmath35 , so that the number of distinct @xmath1th roots of unity which are also roots of @xmath28 is @xmath33 .
the dimension of the code is therefore the number of @xmath1th roots of unity which are also roots of @xmath28 .
the bch bound of the code is determined by the number of consecutive powers of @xmath36 , taken cyclically ( mod@xmath37 ) , which are also roots of @xmath31 . for the reasons outlined in the previous paragraph ,
this is precisely the same as the number of consecutive powers of @xmath36 , taken cyclically ( mod@xmath37 ) , which are _ not _ roots of @xmath28 .
the important features of the code are therefore determined by : [ cols="^ , < " , ] take @xmath38 and let @xmath39 be its ms polynomial .
the inverse mapping @xmath40 is defined as follows : if @xmath41(z)$ ] is the mattson - solomon polynomial of the polynomial @xmath42 then , for @xmath43 @xmath44 ( see @xcite ) .
let @xmath45 and let @xmath46 .
the three key properties relating to the idempotent @xmath28 , listed above , are easily gleaned from its mattson - solomon polynomial @xmath47 , and @xmath48 , as follows : the weight of @xmath28 is the number of @xmath1th roots of unity which are zeros of @xmath48 . to see this note that @xmath50 if and only if @xmath51 , since idempotents take only the values @xmath52 and @xmath53 in @xmath54 .
now @xmath55 and the coefficients of @xmath56 are given by @xmath57 ( _ cf _ equation ( [ inverse ] ) ) .
thus @xmath58 precisely when @xmath50 , giving the weight of @xmath28 as the degree of the polynomial @xmath48 . from the definition of the ms polynomial , ( [ ms ] ) ,
@xmath59 and the number of zeros of @xmath28 which are roots of unity is clearly @xmath60 .
the bch bound of the code is the largest number of consecutive powers of @xmath36 which are _ not _ roots of @xmath28 , i.e. the number of consecutive @xmath61 , taken ( mod@xmath37 ) , such that @xmath62 . from ( [ theta ] ) , this is the largest number of consecutive non - zero coefficients in @xmath63 , taken cyclically ( mod @xmath1 ) . using this information , a systematic search for idempotents
can now be made in increasing order of weight , with accompanying knowledge of the number of roots which are @xmath1th roots of unity and the corresponding bch bound .
this algorithm is constructed in the mattson - solomon domain .
let the decomposition of @xmath64 into irreducible ( over @xmath0 ) polynomials be @xmath65 . for @xmath66 let @xmath67 and let @xmath68 be the associated primitive idempotent ( see @xcite or @xcite ) .
these are displayed below in an array , together with other idempotents : @xmath69 here @xmath70 are the idempotents whose mattson - solomon polynomials are @xmath71 , respectively .
let @xmath72 and let @xmath73 and @xmath74 be defined as @xmath75 , @xmath76 and @xmath77 . from the properties of primitive idempotents ,
if @xmath45 then it follows that gcd@xmath78 and @xmath79 the idempotent @xmath80 will now have the following properties .
@xmath81 the bch bound is determined from @xmath82 as explained in [ bch - bound ] . since methods for finding the @xmath83 and @xmath84 are well documented ( see e.g. @xcite ) a search algorithm can be built around this observation to find a suitable weight idempotent with a known number of zeros and a known bch bound .
the rows of the array ( [ array1 ] ) , are ordered by the degree of the polynomials , i.e. @xmath85 for all @xmath61 , and a search can be made in increasing order of weight .
when a successful outcome has been obtained , _ only at this stage _ is the _ inverse fourier transform _ ( ms@xmath86 ) evaluated to find the corresponding idempotent .
all of the information which is required will already be known .
if @xmath87 denotes the number of cyclotomic cosets modulo @xmath1 , the complexity of an exhaustive search algorithm is @xmath88 .
we reduce this search complexity by targeting the search on the three key parameters : in @xcite , weldon introduced difference - set cyclic codes .
these codes have the desirable property that the parity check equations are orthogonal on all bits and have no cycles of length 4 in their factor graphs . a necessary condition for this
is that if @xmath89 is the polynomial which generates the parity check matrix then the weight of @xmath89 must satisfy the inequality @xmath90where @xmath1 is the code length . since the weights of the idempotents @xmath28 are related to the degrees of the @xmath84 by ( [ eqn : wt - of - u ] ) , a reasonable bound is @xmath91 in practice we have gone a little beyond this limit and this has enabled us to find some good codes which do have cycles of length 4 in their factor graph .
the code - rate is directly proportional to the number of roots of @xmath28 .
if we let @xmath92 represent our minimum desired code - rate then , following equation ( [ eqn : zeros - of - u ] ) , we can refine our search bound to @xmath93 let @xmath94 be the lowest desired minimum distance and let @xmath95 be the largest number of consecutive non - zero coefficients , taken cyclically @xmath96 , of @xmath63 .
then , following the discussion of [ bch - bound ] , we restrict our search algorithm to those @xmath63 for which @xmath97 we develop an efficient , but exhaustive recursive tree - search based on the above bounds .
the developed search algorithm , algorithm [ alg : search ] , is initialised by setting @xmath98 and @xmath99 to @xmath100 and @xmath101 respectively . @xmath102 minimum code - rate of interest + @xmath103 lowest expected minimum distance + @xmath104 small positive integer + @xmath105 @xmath106 sorted in ascending order of the degree + @xmath107 @xmath106 @xmath108 contains set of codes @xmath109 @xmath110 append @xmath61 to @xmath111 @xmath112 @xmath113 @xmath114 a cyclic code defined by @xmath28 add @xmath115 to @xmath116 codesearch(@xmath111 , @xmath99 , @xmath117 , @xmath118 ) @xmath119
since the algorithm is an exhaustive search , the code construction method presented in this paper is able to produce , in addition to new codes , many well - know cyclic codes , for example the difference - set cyclic codes and the euclidean and projective geometry codes .
some of the new codes which we have found using this technique are presented in table [ tbl : code - examples ] .
all codes in table [ tbl : code - examples ] , except those labelled with @xmath120 , have orthogonal parity - check equations .
@xmath121 & @xmath122 + ( 63,44)@xmath120 & @xmath123 & @xmath124 + ( 93,47 ) & @xmath125 & @xmath124 + ( 105,53 ) & @xmath126 & @xmath124 + ( 117,72)@xmath120 & @xmath127 & @xmath128 + ( 127,84)@xmath120 & @xmath129 & @xmath122 + ( 219,101 ) & @xmath130 & @xmath128 + ( 255,135 ) & @xmath131 & @xmath132 + ( 255,175 ) & @xmath133 & @xmath134 + ( 341,205 ) & @xmath135 & @xmath136 + ( 511,199 ) & @xmath137 & @xmath138 + ( 511,259 ) & @xmath139 & @xmath132 + ( 819,435 ) & @xmath140 & @xmath132 + ( 819,447 ) & @xmath141 & @xmath136 + frame error performance of the @xmath142 cyclic code , width=384 ] throughout the paper , it is assume that the codewords are transmitted across a noisy communication channel with bpsk modulation and at the receiving end is the modified belief - propagation decoder which approximates the maximum - likelihood decoder @xcite .
figure [ fig : fer - gf2 - 127 - 84 ] shows the frame - error - rate ( fer ) performance of the @xmath142 cyclic code , which is a code which does have cycles of length 4 .
neverthless , the performance of this code is outstanding and , at @xmath143 fer , it is within @xmath144db of the sphere - packing - bound constraint for binary transmission .
frame error performance of the @xmath145 codes , width=384 ] figure [ fig : fer - gf2 - 255 - 175 ] shows the performance of two @xmath145 codes , one constructed using our method and one an irregular computer generated code .
we can see that our code , which achieves a coding gain of around @xmath146db compared to the equivalent irregular code , performs approximately withinr @xmath147db of the sphere - packing - bound constraint for binary transmission .
frame error performance of the @xmath148 codes , width=384 ] our construction method can produce ldpc codes with high minimum distance and therefore they do not suffer from error - floor . figure [ fig : fer - gf2 - 341 - 205 ] demonstrates the performance of our @xmath148 code , which is inferior to the equivalent irregular code in the low signal - to - ratio region , but the irregular code exhibits early error - floor due to its low minimum distance .
a method of constructing binary cyclic codes from the finite - field transform ( mattson - solomon ) domain is able to produce a large number of codes which have high minimum - distance and code - rate .
these codes have sparse parity - check matrix and thus are applicable as ldpc codes . due to their cyclic property these ldpc codes have @xmath1 parity - check equations instead of @xmath32 equations as in the case of random ldpc codes . with these extra parity - check equations to iterate with , the performance of the iterative decoder is improved . in designing cyclic ldpc codes of length @xmath1 ,
our new method allows one to increase the @xmath149 of the code by combining additional irreducible factors of @xmath150 which , in turn , reduces the sparseness of the parity - check matrix .
the ability to control the sparseness of the parity - check matrix is a trade - off against the minimum distance of the code .
simulation results have shown that the our cyclic codes have outstanding performance which are superior to the equivalent irregular ldpc codes .
the high @xmath149 of these codes ensures the absence of an early error - floor in their performance .
99 f.j.macwilliams and n.j.a.sloane , _ the theory of error correcting codes _ , north holland , 1977 v.s.pless , w.c.huffman and r.a.brualdi , an introduction to algebraic codes , in _ handbook of coding theory _ , ( v.s.pless andw.c.huffman , eds . ) , elsevier , north holland , 1998 s.roman , _ coding and information theory _ , springer verlag , 1992 t.shibuya and k.sakaniwa , construction of cyclic codes sutiable for iterative decoding via generating idempotents , _ ieice trans .
fundamentals _ , * e86-a * , no.4 , april 2003 .
t.tian , c.jones , j.villasenor , and r. wesel , selective avoidance of cycles in irregular ldpc code construction , _ ieee transaction on communications _ , august 2004 j.h.van lint , _ coding theory _ , lecture notes in mathematics , 201 , springer - verlag , 1971 j.h.van lint , _ introduction to coding theory _ ,
second ed .
, springer - verlag , 1991 e. j. weldon , difference - set cyclic codes , _ bell system technical journal _ , september ( 1966 ) c. tjhai , e. papagiannis , m. tomlinson , m. ambroze and m. ahmed , `` improved iterative decoder for ldpc codes with performance approximating to a maximum - likelihood decoder '' , uk patent application 0409306.8
|
we show how to construct an algorithm to search for binary idempotents which may be used to construct binary ldpc codes .
the algorithm , which allows control of the key properties of sparseness , code rate and minimum distance , is constructed in the mattson - solomon domain .
some of the new codes , found by using this technique , are displayed .
* idempotents , polynomials and binary ldpc codes * + r. horan , c. tjhai , m. tomlinson , m. ambroze and m. ahmed + _ fixed and mobile communications research _ + _ university of plymouth , pl4 8aa , united kingdom _ + * *
|
the heavy ion collision experiments at rhic and lhc are designed with a hope to explore the existence of a new form of matter known as quark gluon plasma ( qgp ) and to explore its properties .
the estimation for the energy density @xcite attained in these collisions using the bjorken formula @xcite is well beyond the energy densities where qgp is expected to be formed .
the temperatures reached at rhic , as revealed from several studies ( see e.g. , @xcite for a compilation ) are also much larger than the values provided by lattice qcd calculations for the critical temperature for a transition to qgp @xcite .
strong confirmation of the formation of the qgp is given by observation of a large elliptic flow @xcite , jet - quenching @xcite , and the recombination of partons as the mechanism of production of hadrons at intermediate transverse momenta @xcite .
still higher temperatures are likely to be reached at lhc .
heavy quarks are produced from the initial fusion of gluons ( @xmath1 ) or light quarks ( @xmath2 ) .
this pair would be produced at @xmath3 0.1 fm / c .
their large mass ensures that their production can be treated using pqcd and that it is nearly negligible at later times .
these will traverse the qgp , colliding with quarks and gluons and radiating gluons before appearing as charm or bottom mesons or baryons .
thus the final spectra for these hadrons would contain information about the energy loss suffered by the heavy quarks .
the unique importance of heavy quarks as probes of qgp lies in their large mass .
this leads to a considerably reduced production of heavy quarks in comparison to light quarks and gluons which are produced copiously .
the strong interaction during the collision conserves flavour . even at the lhc , the reverse process of annihilation of heavy quarks ( @xmath4 , etc . ) can be safely ignored . thus heavy quarks and in turn charm and bottom hadrons will stand out in the back - ground of a multitude of light hadrons , and one can in principle track them . along with other reasons ,
an interest in the study of energy loss of heavy quarks was triggered by the large back - ground that correlated charm or bottom decay provides @xcite to the thermal dileptons which have been considered a signature of the formation of qgp for a long time @xcite .
it was pointed out by shuryak @xcite , lin _ et al . _
@xcite , kampfer _ et al . _
@xcite , and mustafa _ et al . _
@xcite that the correlated charm and bottom decay could be suppressed if the energy loss suffered by heavy quarks before they form d or b mesons was accounted for . since then several attempts have been made to estimate the energy loss of heavy quarks as they proceed through qgp .
of course the possibility to identify the vertex of d or b meson decay will further enrich this study .
these results have been put to a rigorous test by the measurement of single electrons from heavy ion collisions at rhic which show a clear evidence for the loss of energy by the heavy quarks @xcite .
their possible thermalization is also indicated by the elliptic flow that they show @xcite .
as indicated earlier , the temperature likely to be reached at lhc in collision of heavy nuclei could be even larger and thus this energy loss will play a more significant role .
the opening of a much wider window in rapidity at lhc is also likely to provide widely differing media at different rapidities through which the heavy quarks would propagate .
thus a valuable test of various theories for energy loss suffered by heavy quarks can be performed by studying it at rhic and lhc and at different rapidities .
we study these effects in terms of nuclear modification factor @xmath0 for heavy quarks . in these initial studies
we calculate the average energy loss suffered by them as they pass through the qgp and compare the resulting @xmath5 distribution with the same for proton - proton ( pp ) collisions to get @xmath0 . since the mass of charm or bottom quarks is quite large , the @xmath5 distribution of these quarks will closely reflect the @xmath5 distribution of d or b mesons .
we employ a local fluid approximation @xcite in order to picture the medium at larger rapidities .
we shall come back to this later .
the paper is organized as follows . as we need to compare the spectra of the heavy quarks from relativistic heavy ion collisions with those for pp collisions , as a first step we study the heavy quark production in lo pqcd and compare our results with a nlo pqcd calculation .
we find that single quark distribution calculated using lo pqcd supplemented with a k - factor adequately reproduces the nlo results as well as the available experimental data .
next we estimate the average energy loss suffered by heavy quarks of a given energy using various mechanisms discussed in the literature .
finally we perform a monte carlo calculation to obtain the average change in the transverse momentum spectra of heavy quarks for nucleus - nucleus collisions and get @xmath0 as a function of @xmath5 for different rapidities .
we add that this work is not intended as a complete review and the readers may see ref .
@xcite , for other treatments .
more detailed calculations where the consequences of energy loss of heavy quarks on the correlated charm or bottom decay and modification of the back - to - back correlation of heavy quarks are discussed , will be published shortly .
at lowest order in pqcd , heavy quarks in pp collisions are produced by fusion of gluons ( @xmath6 ) or light quarks ( @xmath7 ) @xcite .
the so - called flavour excitation process ( @xmath8 and @xmath9 ) is now known to be suppressed when the nlo processes are taken into account @xcite .
in addition , brodsky _ et al . _
@xcite have shown that the total contribution of intrinsic charm in the midrapidity region is small even though most of the heavy quarks are produced in this region .
the cross - section for the production of heavy quarks from pp collisions at lowest order is given by @xcite : @xmath10/(1+\delta_{ij } ) .
\label{ds}\end{aligned}\ ] ] in the above equation , i and j are the interacting partons , @xmath11 and @xmath12 are the partonic structure functions and @xmath13 and @xmath14 are the fractional momenta of the interacting hadrons carried by the partons i and j. the relation between @xmath5 and fractional momentum @xmath13 or @xmath14 through their respective rapidities can be written as @xmath15 where @xmath16 is the transverse mass , @xmath17 , of the produced heavy quark .
the function @xmath18 , the short range subprocess for the heavy quark production is defined as : @xmath19 @xmath20 for the heavy quark production processes @xmath21@xmath22@xmath23 and @xmath24@xmath22@xmath23 are expressed through the mass of the heavy quark and mandelstam variables @xmath25 , @xmath26 , and @xmath27 as @xmath28\end{aligned}\ ] ] and @xmath29.\end{aligned}\ ] ] the running coupling constant @xmath30 at lowest order is @xmath31 where @xmath32=3 is the number of active flavours and @xmath33 .
we use the factorization and renormalization scales as @xmath34=@xmath35 .
we refer the readers to vogt _ et al . _
@xcite for results on variations of these scales .
we also carry out the calculation of differential cross section for heavy quarks in pp collision at nlo in pqcd using the treatment developed by mangano , nason , and ridolfi ( mnr - nlo ) @xcite .
all the calculations are carried out by neglecting the intrinsic transverse momentum of the partons .
the effect of nuclear shadowing in high energy nucleus - nucleus collisions is well known @xcite . with the increase of the mass number of the nucleus and increasing contribution of terms having small x ,
the effect becomes more pronounced .
we introduce the shadowing effect in our calculations by using eks 98 parameterization @xcite for nucleon structure functions .
we take cteq4 m @xcite structure function set for nucleons .
we shall see that x dependence of the shadowing function introduces interesting structures in the nuclear modification factor as a function of @xmath5 , y , and the incident energy , because of the large mass of the quarks . in fig .
[ dsdy ] we compare our results for heavy quark @xmath5 distribution obtained using lowest order pqcd for pp collision with the results from nlo - mnr calculation at midrapidity for charm and bottom quarks at rhic and lhc energies .
these comparisons suggest a k factor of @xmath361.5 - 3 for our lowest order calculations for agreement with nlo results . in fig .
[ ptmin ] we compare the bottom quark production cross - section obtained using lowest order pqcd in pp collision at 630 and 1800 gev energies with ua1 , cdf and @xmath37 data @xcite .
we find a good description of these data using lowest order pqcd with a k - factor .
we calculate the total cross - section for charm quark production at lowest order for the process pp@xmath22@xmath38 as a function of @xmath39 considering the charm quark mass as 1.2 gev and 1.6 gev .
we have also included results for @xmath40 @xcite , suggested recently .
though the last value is not at the pole q=@xmath41 , it may serve as the lower limit to the mass of the charm quark . in the left panel of fig .
[ csc ] we compare these results and the results from nlo - mnr calculation with the experimental data points @xcite . here also our lowest order pqcd calculation show a good agreement with experimental data points for @xmath41=1.2 gev . in the right panel of fig .
[ csc ] we compare our results with the nlo - mnr calculations up to @xmath39=15000 gev .
we also calculate the total cross - section for bottom quark production for pp@xmath42 as a function of @xmath39 considering the bottom quark mass as 4.7 gev and 4.163 gev @xcite . in fig .
[ csb ] we compare these results and the results obtained from nlo - mnr calculation with the experimental data points @xcite .
this comparison is quite impressive as at @xmath43=4.7 gev our lowest order result accurately reproduce the result of nlo - mnr calculation with k=2.5 .
thus we see that the @xmath5 distribution and production cross - section for the heavy quarks calculated in lowest order pqcd and supplemented with a k - factor @xmath36 2.5 reproduces the results at nlo for pp collisions . in view of this
we feel that these distribution would be adequate for calculating @xmath0 .
the heavy quarks produced at the initial stage pass through the qgp , where they loose energy by colliding with quarks and gluons and also by radiating gluons .
the energy loss will depend upon the path - length of the heavy quarks in the plasma , the temperature evolution of the plasma , and the energy and mass of the heavy quarks . in order to proceed we make several simplifying assumptions .
it is expected that the heavy quarks will loose most of their energy when the temperature is still large , i.e. during the earliest times after the formation of qgp . during these early times
, we can neglect the transverse expansion of the plasma .
we assume a gaussian rapidity density distribution for the particles produced and further assume that the initial rapidity distribution of the quarks and gluons follows this distribution closely .
this would correspond to an isentropic expansion of the plasma at all rapidities , and should be sufficient for our initial studies .
thus we assume the rapidity distribution of the density of gluons as @xcite : @xmath44 we take @xmath45@xmath36900 and @xmath46 for au+au collisions at rhic @xcite and @xmath363300 and @xmath47 for pb+pb collisions at lhc @xcite .
the bjorken cooling is then assumed to work locally at different rapidities , and we consider the passage of a heavy quark having rapidity y in a fluid having an identical fluid rapidity .
this approximation , which corresponds to assuming a boost - invariant expansion along with a local fluid approximation , has been used earlier in literature @xcite . a more complete study would use a ( 3 + 1 ) dimensional hydrodynamics @xcite , which we plan to use in future publications . the distance , l , covered by a heavy quark while passing through the qgp . for central collisions the results for @xmath48 will not depend on @xmath49.,width=16 ] we consider a heavy quark produced in a central collision , at the point ( r , @xmath49 ) , and moving at an angle @xmath50 with respect to @xmath51 in the transverse plane . in general the distance covered by the heavy quark before it exists the qgp , will vary from 0 to 2r , where r is the radius of the colliding nuclei .
the distance covered by the heavy quark in the plasma , l , is given by @xcite : @xmath52 we can estimate the average distance travelled by the heavy quarks in the plasma as : @xmath53 in the above the nuclear overlap function @xmath54(r , b=0 ) provides the probability of production of heavy quarks in hard binary collisions .
we find that @xmath55 is 5.78 fm for au+au collisions at rhic and 6.14 fm for pb+pb collisions at lhc , and is about 20@xmath56 smaller than the radii of the colliding nuclei , as the appearance of the nuclear overlap function gives a larger weight to the points having smaller r. as the heavy quarks loose most of their energy in interaction with gluons , it is enough to consider only the distribution of gluons .
their density at the time @xmath57 can be written as @xcite : @xmath58 the corresponding temperature @xcite , assuming a chemically equilibrated plasma is @xmath59 the rapidity dependence of the temperature of the plasma at a typical @xmath60 at rhic and lhc is given in fig .
[ tr ] . assuming that the qgp is formed at @xmath61=0.2 fm / c
, we estimate the @xmath62 at y=0 for rhic as 377 mev and lhc as 555 mev .
more detailed studies do suggest a larger formation time of @xmath36 0.5 fm / c , which will correspondingly reduce the initial temperatures .
this will not affect our results as we approximate the expanding and cooling plasma , with one at a temperature determined at @xmath63 , which is much larger ( see the following discussion ) .
further assuming , bjorken s
cooling law , @xmath64=constant , this provides that the plasma would cool down to the transition or critical temperature @xmath65@xmath36160 mev by @[email protected] fm / c at rhic and 8 fm / c at lhc
. one can easily calculate the corresponding values at larger rapidities .
considering the velocity of the quark as @xmath67=@xmath68 , it would take a time @xmath69 = @xmath70 to cross the plasma . if @xmath66 @xmath71 @xmath69 the heavy quark would be inside qgp during the entire period , @xmath61 to @xmath69 .
however , if @xmath66 @xmath72 @xmath69 , only while covering the distance @xmath73 , would the heavy quark be in the qgp phase .
we further approximate the expanding and cooling plasma with one at a temperature of t at @xmath57 = @xmath74 , where @xmath75=min @xmath76 $ ] .
this procedure has been used frequently @xcite .
next we discuss the energy loss mechanisms that we have included .
as mentioned above , repeatedly , the heavy quarks loose energy both by collisions as well as radiation of gluons .
a number of formalisms have been proposed for the collisional as well as the radiative energy loss of heavy quarks in the literature .
we shall consider the following treatments for the collisional energy loss .
bjorken @xcite has considered the collisional energy loss of light quarks as analogous to the energy loss of a charged particle passing through a medium and losing energy by ionizing the medium .
his expression for massless quarks was adapted by braaten and thoma to the case of heavy quarks [ see eq .
[ bjorken ] ] .
we shall continue to label this mechanism as bjorken for clarity .
braaten and thoma ( bt ) @xcite also modified the expression for the energy loss suffered by muons while traversing qed plasma , to obtain the collisional energy loss of a heavy quark as it passes through the qgp [ see eqs .
[ bt1 ] and [ bt2 ] in the appendix a ] .
these results are valid for collisions where the momentum transfer q@xmath77e , where e is the energy of the heavy quark .
peigne and peshier ( pp ) @xcite have improved this treatment by including the u - channel , which becomes important for large energies [ see eq .
[ pp ] in the appendix a ] . for the calculation of radiative energy loss , we consider the treatment of djordjevic , gyulassy , levai , and vitev ( dglv ) @xcite using opacity expansion , the treatment of armesto , salgado , and wiedemann ( asw ) @xcite using path integral formalism for medium - induced gluon radiations off massive quarks , and the treatment of xiang , ding , zhou , and rohrich ( xdzr ) @xcite using light cone path integral approach .
detailed expressions for these formalisms are given in the appendix b.
we compare the results for transverse energy loss for a heavy quark using these different energy loss treatments for several rapidities .
we plot the transverse energy loss of charm and bottom quarks , @xmath78 as a function of transverse energy @xmath79 @xmath80 in figs .
[ eloss - c ] and [ eloss - b ] , at rhic and lhc energies .
several interesting features emerge .
we see that the collisional energy loss for charm quarks at rhic and lhc energies is only marginally dependent on the rapidity and the bt formalism gives largest energy loss , as expected . in our treatment ,
change of rapidity implies a change in the temperature of the plasma .
thus these results suggest a weaker dependence on the temperature and the average path length for the energy loss suffered by charm quarks due to collisions .
the radiative energy loss , on the other hand , shows a much more complex behaviour and is quite different for the different formalisms under consideration .
we note that the asw formalism for radiative energy loss gives largest degradation in the energy at all rapidities ( except for @xmath81 at lhc , where it is comparable to the collisional energy loss ) .
we also see that the dglv and the xdzr formalisms give similar results at rhic energies , at the three rapidities under consideration . on the other hand , at lhc energy ,
the asw and dglv formalisms provide nearly identical results for energy loss for charm quarks at y=0 and y=2 , and the corresponding results at y=4 differ by about 10@xmath56 .
this , we feel , is due to a more complex dependence on the average path length in the asw formulation .
the collisional energy loss for bottom quarks using the pp and the bjorken s formulation are seen to be quite similar at rhic and lhc energies for all rapidities under consideration .
the bt formulation due to the neglect of the u - channel , gives a much smaller energy loss , both for rhic and lhc energies and at all rapidities .
we have already mentioned that due to the numerical approximations used , the xdzr formulation is not valid for evaluation of the radiative energy loss for bottom quarks .
the asw and dglv radiative energy loss formalisms show a more complex dependence on the mass and the average path length .
the asw formulation gives a larger energy loss at rhic energy , though the results are again comparable at lhc energy at all rapidities .
we note that while the collisional and radiative energy losses for bottom quarks at rhic energy are comparable , the collisional energy loss dominates over the radiative energy loss in the @xmath79 range under consideration at lhc energy .
we have confirmed that at higher @xmath79 , the radiative energy loss again starts dominating .
this rich structure suggests that description of energy loss for one ( quark ) mass at one rapidity , and one energy may not be enough to identify the most reliable treatments , for this .
the nuclear modification factor @xmath0 for heavy quarks can be expressed as : @xmath82 where , as mentioned earlier , @xmath83 is the nuclear overlap function for impact parameter b , calculated using glauber model .
we get @xmath54@xmath36 280 @xmath84 for au+au collisions at rhic and @xmath36 290 @xmath84 for pb+pb collisions at lhc , for b=0 fm . as a first step ,
we give the results for @xmath0 with only nuclear shadowing effect for production of charm and bottom quarks at the rapidities considered earlier for rhic and lhc energies ( see fig .
[ raaws ] ) .
we see that the fairly large masses of the charm and bottom quarks , the kinematics , and the rich behaviour of the structure function with x and @xmath34 , lends interesting features to @xmath0 .
we see that @xmath0 for charm and bottom quarks for y=0 and y=1 are quite similar at rhic energy
. similarly , the results for y=0 and y=2 are only marginally different at lhc energy .
the results at larger y are more strongly affected due to increased variation in the x values ( see eq .
[ x1 ] ) which contribute . in order to do a full justice to these interesting results ,
we now discuss them individually . for charm quarks at rhic energy ,
we see a suppression at lower @xmath5 , an enhancement at intermediate @xmath5 , and again a suppression at larger @xmath5 , for y=0 and y=1 , while for y=2 , @xmath0 starts at about 0.8 , goes up to a value slightly more than 1 , then drops again to about 0.8 at @xmath5@xmath3620 gev , and rise again to beyond 1 at @xmath5@xmath3640 gev .
since the energy loss of the charm quarks always rises with increasing @xmath5 , this would introduce interesting features in @xmath0 after this is accounted for , unless of course the @xmath5 spectrum for the quarks drops too rapidly .
we shall come back to this point again .
the increased energy at lhc then provides a larger suppression at low @xmath5 for all the rapidities . in an interesting development , @xmath0 for y=0 and y=2 rises beyond 1 at @xmath5@xmath3620 gev , while it stays below 1 up to @xmath5@xmath3640 gev , for y=4 .
the results for bottom quarks are even more interesting . due to the large mass of the bottom quarks , at rhic energy , the @xmath0 for lower @xmath5 for y=0 and y=1 is already starts getting contributions from the region of x where anti - shadowing appears .
thus @xmath0 starts at a value which is more than 1 at lower @xmath5 , goes up , up to @xmath5@xmath365 gev and then drops again . for y=2 on the other hand , it starts at a value close to 1 , drops by about 10@xmath56 at @xmath5@xmath3620 gev and rises again . at lhc energy ,
@xmath0 for bottom quarks for y=0 and y=2 starts at @xmath3630@xmath56 below 1 and then rises steadily to about 1.1 at @xmath5@xmath3640 gev .
the results at y=4 remain close to 0.8 , rising slightly at intermediate @xmath5 .
now let us discuss our results for @xmath0 with the additional inclusion of collisional and radiative energy losses .
we shall restrict our consideration to inclusion of collisional energy loss using the pp formulation and the radiative energy loss using asw or dglv formulation .
[ raac - rhic ] gives our findings for charm quarks at rhic energy .
we see that the outcome of shadowing and energy loss gives an interesting structure to @xmath0 , as expected .
we see that the final @xmath0 starts at about 40@xmath56 below 1 , goes up to about 0.8 at @xmath5@xmath362 gev , and then drops to a value of @xmath360.3 at @xmath5@xmath3615 gev .
in an interesting development , we see that the combination of the shadowing and energy loss gives a marginally larger suppression at y=2 compared to y=0 , even though the fractional energy loss is higher at smaller y ( see fig .
[ eloss - c ] ) .
we have also given a comparison of @xmath0 by replacing the asw formulation for the radiative energy loss with the dglv treatment , and see that the former gives a larger suppression at all y ( see also fig .
[ eloss - c ] ) .
we recall that the single electrons produced from the semi - leptonic decay of charm mesons @xcite show @xmath0 of about [email protected] for @xmath5@xmath862 gev , in a very encouraging agreement with these results .
this will be pulbished shortly .
the results for the @xmath0 for charm quarks at lhc energy are shown in fig .
[ raac - lhc ] .
we show the results only for the dglv formulation for the radiative energy loss , as we have seen that it is quite similar to that for the asw formalism for charm quarks at lhc energy .
we see roughly similar behaviour , in that the @xmath0 starts from a lower value at @xmath5@xmath360 gev , rises up to @xmath5@xmath362 gev , and then drops to a level of about 0.2 at larger @xmath5 .
we also find a marginally larger suppression for y=0 compared to that for y=4 .
the results for y=0 for rhic energy are also given for a comparison which suggests a much larger suppression at lhc , as expected .
next we discuss our findings for nuclear suppression for bottom quarks at rhic energy ( see fig .
[ raab - rhic ] ) .
the shadowing and the large mass of the bottom quarks , with its consequences , gives an @[email protected] at @xmath5@xmath360 gev , which goes up to about 1.2 at @xmath5@xmath362 gev , and then drops to about [email protected] at larger @xmath5 .
the shadowing results in a larger suppression for y=2 compared to y=0 ( see fig . [ raaws ] ) , even though the energy loss is slightly lower for larger y ( see fig .
[ eloss - b ] ) .
results obtained by replacing the asw formulation with the dglv treatment show a smaller suppression , as for charm quarks ( see fig .
[ raac - rhic ] ) . finally in fig .
[ raab - lhc ] we have given our results for @xmath0 for bottom quarks at lhc energy for y=0 , 2 and 4 using shadowing , collisional energy loss and radiative energy loss using the dglv treatment .
the results using asw treatment are expected to be quite similar as seen from fig .
[ eloss - b ] .
the results for y=0 at rhic energy are also given for a ready reference .
we again see a trend which is common to our results that @xmath0 starts at @xmath360.6 at @xmath5@xmath360 gev , goes up to @xmath361.1 at @xmath5@xmath362 gev , and then slowly drops to about 0.4 at @xmath5@xmath3620 gev .
the overall effect of shadowing and energy loss is seen to lead to very similar values for @xmath0 from y=0 to y=4 .
of course the @xmath0 for rhic energy is about twice as large , showing a much reduced suppression .
we have made a detailed study of charm and bottom production from initial fusion of partons in relativistic collision of heavy nuclei .
as a first step we have checked the usefulness of lowest order pqcd to reproduce the nlo results for the @xmath5 distribution of charm and bottom quarks in pp collisions .
next we have checked our predictions against experimental results of the charm and bottom quarks production from such collisions .
we have further obtained the average energy loss suffered by heavy quarks at several rapidities for rhic and lhc energies due to collisions and gluon radiations .
we have obtained the nuclear suppression factor @xmath0 by additionally incorporating nuclear shadowing for the cases under study .
a rich picture of dependence of @xmath0 on y , @xmath5 , incident energy , and the mass of the heavy quarks emerges . we have noted that our findings would support the suppression of single electrons seen at rhic . before concluding , we discuss some of the short - comings of the present work .
these initial calculations can be improved in several ways . we have used ( 1 + 1 ) dimensional bjorken hydrodynamics and assumed it to apply at all y. since the heavy quarks will loose most of their energy at very early times , this may not be a serious short - coming .
still we are looking at the possibility of using a full fledged ( 3 + 1 ) dimensional hydrodynamics calculations , also at b@xmath870 . we are incorporating the single electron decay of the resulting d and b mesons , along with the results for their back to back correlation .
we expect this to be rewarding , especially in conjunction with nlo results for pp collisions , as it may throw up an interesting detail about differences of nlo results and the results with energy loss .
these will be published shortly .
finally , we conclude that the description for energy loss for one quark mass at one rapidity for a particular incident energy may not be sufficient to identify the most reliable energy loss treatment for either collisional or radiative energy loss valid for all cases .
we gratefully acknowledge the use of mnr - nlo code @xcite made available to us by the authors .
we also thank m. g. mustafa for useful discussions .
bjorken argued that the elastic energy loss by partons in the qgp is very similar to the energy loss due to ionization due to passage of charged particles in ordinary matter @xcite .
this treatment was adapted by braaten and thoma for heavy quarks @xcite . the fractional energy loss suffered by the heavy quark due to collisions with the quarks and gluons given as : @xmath88 \log\frac { q_{\rm { max}}}{q_{\rm{min } } } , \label{bjorken}\end{aligned}\ ] ] where @xmath89 is the velocity of the heavy quark .
as suggested by braaten and thoma we use the upper limit of the momentum transfer @xmath90 as @xmath91 and the lower limit of the momentum transfer @xmath92 as @xmath93 .
the thermal gluon mass @xmath94 can be expressed as @xmath94=@xmath95 where @xmath96=@xmath97 is the debye screening mass .
braaten and thoma first developed a theoretical formalism to find the collisional energy loss of a muon propagating through a plasma of electrons , positrons and photons to leading order in qed @xcite .
this work was further extended by them to calculate the collisional energy loss of heavy quarks propagating through qgp @xcite .
the energy loss formulation is given in two energy regimes : the fractional collisional energy loss of heavy quarks with energy @xmath98 is @xmath99\times \nonumber \\ & & \log \left(2^{\frac{n_f}{6\,+\,n_f } } \ , b(v ) \
, \frac{e\,t}{m_g\,m}\right ) \label{bt1}\end{aligned}\ ] ] the fractional collisional energy loss of heavy quarks with energy @xmath100 is @xmath101 where b(@xmath89 ) is a smooth function of velocity having value in the range 0.6 - 0.7 .
braaten and thoma have shown the crossover energy between these energy regimes as @xmath102 for @xmath32=2 . in bt formalism
, it was assumed that the momentum exchange in the elastic scattering process is much less than the energy carried by the heavy quarks .
peigne and peshier pointed out that this assumption is not reliable in the energy regime @xmath103 , and corrected it in the qed case while calculating the collisional energy loss of a muon in qed plasma @xcite .
this work in qed is then used by them to derive the collisional energy loss suffered by heavy quarks while passing through qgp @xcite .
the fractional collisional energy loss suffered by heavy quarks as proposed by peigne and peshier is @xmath104 \label{pp}\end{aligned}\ ] ] and @xmath105 .
for massless quarks , gyulassy , levai and vitev ( glv ) calculated the induced radiation to arbitrary order in opacity @xmath106 ( @xmath107 ) of the plasma @xcite where @xmath108 is the mean free path of the quark . in djordjevic ,
gyulassy , levai , and vitev ( dglv ) formulation @xcite , the glv method is generalized to estimate the first order induced radiative energy loss including the kinematic effect for heavy quarks .
et al . _ @xcite present a simplified form of the dglv formalism for the average radiative energy loss of heavy quarks : @xmath109 where @xmath110 , \nonumber \\
q_k^{\pm}=q^2\,\pm \,k,\nonumber \\
f_\beta = f\left(\beta,\ , q_\mu^-,\,q_\mu^+\right ) , \nonumber \\
f_\mu = f\left(\mu , \,q_k^+,\,q_k^-\right ) \nonumber \\ { \rm { and } } \nonumber\\ f(x,\ , y , \,z)=\sqrt{x^4\,+\,2\,x^2\,y\,+\,z^2}.\end{aligned}\ ] ] in armesto , salgado , and wiedemann ( asw ) , formulation @xcite path integral method for medium - induced gluon radiation is employed to calculate the radiative energy loss of heavy quarks .
this formalism provides the analysis of the double differential medium - induced gluon distribution by the heavy quarks as a function of transverse momentum .
the average radiative energy loss is @xmath111}{\left(q^2\,+\,\bar{m}^2\right)^2 } \nonumber \\ & & \times \frac{q^2}{q^2\,+\,\bar{m}^2}\nonumber\\ & & \ , \times \frac{\left(k^2\,+\,\bar{m}^2\right)\,+\,\left(k^2\,-\,\bar{m}^2\right ) \left(k^2\,-\,q^2\right ) } { \left(k^2\,+\,\bar{m}^2\right)\left[\left(1\,+\,k^2\,+\,q^2\right)^2\,-\,4\ , k^2\,q^2\right]^\frac{3 } { 2}}. \label{asw}\end{aligned}\ ] ] in the above equation the gluon energy , transverse momentum and heavy quark mass are expressed as dimensionless parameters .
the rescaled dimensionless parameters : @xmath112 we use the parameter @xmath113=4 . in xiang , ding , zhou , and rohrich ( xdzr ) formulation @xcite ,
light cone - path integral method is used to calculate the gluon radiation from heavy quarks where an analytical expression is obtained for the heavy quark radiative energy loss .
the average radiative energy loss is @xmath114 , \label{xdzr}\end{aligned}\ ] ] where @xmath115=2.5 gev .
this analytical expression is derived by using an expansion for bessel function @xcite , which is valid only for not too large mass of the quarks .
we use it only for charm quark .
90 s. s. adler _ et al . _
[ phenix collaboration ] , , 182301 ( 2003 ) ; s. esumi [ for the phenix collaboration ] , , 599 ( 2003 ) ; c. adler _ et al . _ [ star collaboration ] , , 032301 ( 2003 ) ; _ ibid . _ * 89 * 132301 ( 2002 ) ; _ ibid . _ * 87 * 182301 ( 2001 ) , p. huovinen , p. f. kolb , u. w. heinz , p. v. ruuskanen and s. a. voloshin , , 58 ( 2001 ) , d. teaney , j. lauret and e. v. shuryak , [ arxiv : nucl - th/0110037 ] .
b. kampfer , o. p. pavlenko , and k. gallmeister , , 412 ( 1998 ) ; k. gallmeister , b. kampfer , and o. p. pavlenko , , 3276 ( 1998 ) .
m. g. mustafa , d. pal , and d. k. srivastava , , 889 ( 1998 ) , erratum : , 3499 ( 1998 ) .
k. j. eskola , v. j. kolhinen , and c. a. salgado , , 61 ( 1999 ) , jyfl-8/98 , us - ft/14 - 98 , hep - ph/9807297 ; k. j. eskola , v. j. kolhinen , and p. v. ruuskanen , , 351 ( 1998 ) , cern - th/97 - 345 , jyfl-2/98 , hep - ph/9802350 .
|
we calculate nuclear suppression @xmath0 of heavy quarks produced from the initial fusion of partons in nucleus - nucleus collisions at rhic and lhc energies .
we take the shadowing as well as the energy loss suffered by them while passing through quark gluon plasma into account .
we obtain results for charm and bottom quarks at several rapidities using different mechanisms for energy loss , to see if we can distinguish between them .
|
the copula provides an elegant margin - free description of the dependence structure of a random variable . by the famous theorem of @xcite
, it follows that the distribution function @xmath4 of a bivariate random variable @xmath5 can be represented in terms of the marginal distributions @xmath6 and @xmath7 of @xmath8 and @xmath9 , that is , @xmath10 where @xmath11 denotes the copula , which characterizes the dependence between @xmath8 and @xmath9 .
extreme - value copulas arise naturally as the possible limits of copulas of component - wise maxima of independent , identically distributed or strongly mixing stationary sequences [ see @xcite and @xcite ] .
these copulas provide flexible tools for modeling joint extremes in risk management .
an important application of extreme - value copulas appears in the modeling of data with positive dependence , and in contrast to the more popular class of archimedean copulas they are not symmetric [ see @xcite or @xcite ] .
further applications can be found in @xcite or @xcite among others .
a copula @xmath11 is an extreme - value copula if and only if it has a representation of the form @xmath12,\ ] ] where @xmath13 \to[1/2,1]$ ] is a convex function satisfying @xmath14 , which is called pickands dependence function .
the representation of ( [ 11 ] ) of the extreme - value copula @xmath11 depends only on the one - dimensional function @xmath15 and statistical inference on a bivariate extreme - value copula @xmath11 may now be reduced to inference on its pickands dependence function @xmath15 .
the problem of estimating pickands dependence function nonparametrically has found considerable attention in the literature .
roughly speaking , there exist two classes of estimators .
the classical nonparametric estimator is that of @xcite [ see @xcite for its asymptotic properties ] and several variants have been discussed .
alternative estimators have been proposed and investigated in the papers by @xcite , @xcite , @xcite , @xcite and @xcite , where the last - named authors also discussed the multivariate case . in most references ,
the estimators of pickands dependence function are constructed assuming knowledge of the marginal distributions .
recently @xcite proposed rank - based versions of the estimators of @xcite and @xcite , which do not require knowledge of the marginal distributions . in general , all of these estimators are neither convex nor do they satisfy the boundary restriction @xmath16 , in particular the endpoint constrains @xmath17 . however , the estimators can be modified without changing their asymptotic properties in such a way that these constraints are satisfied , see , for example , @xcite.=-1 before the specific model of an extreme - value copula is selected , it is necessary to check this assumption by a statistical test , that is a test for the hypotheses @xmath18 where @xmath19 denotes the class of all copulas satisfying ( [ 11 ] ) . throughout this paper ,
we call ( [ hypextr ] ) the hypothesis of extreme - value dependence
. the problem of testing this hypothesis has found much less attention in the literature . to our best knowledge ,
only two tests of extremeness are currently available in the literature .
the first one was proposed by @xcite .
it exploits the fact that for an extreme - value copula the random variable @xmath20 satisfies the identity @xmath21 - 9 \mathbb{e}[w^2 ] = 0.\ ] ] the properties of this test have been studied by @xcite , who determined the finite- and large - sample variance of the test statistic . in particular , the test proposed by @xcite is not consistent against alternatives satisfying ( [ how ] ) .
the second class of tests was recently introduced by @xcite who proposed to compare the empirical copula and a copula estimator which is constructed from the estimators proposed by @xcite under the assumption of an extreme - value copula .
these tests are only consistent against alternatives that are left tail decreasing in both arguments and satisfy strong smoothness assumptions on the copula and convexity assumptions on an analogue of pickands dependence function , which are hard to verify analytically .
the present paper has two purposes .
the first is the development of some alternative estimators of pickands dependence function using the principle of minimum distance estimation .
we propose to consider the best approximation of the logarithm of the empirical copula @xmath22 evaluated in the point @xmath23 , that is , @xmath24 , by functions of the form @xmath25 with respect to a weighted @xmath1-distance .
it turns out that the minimal distance and the corresponding optimal function can be determined explicitly . on the basis of this result , and by choosing various weight functions in the @xmath1-distance , we obtain an infinite - dimensional class of estimators for the function @xmath15 .
our approach is closely related to the theory of @xmath26-estimation and in section [ secmindist ] we indicate how this point of view provides several interesting relationships between the different concepts for constructing estimates of pickands dependence function .
the second purpose of the paper is to present a new test for the hypothesis of extreme - value dependence , which is consistent against a much broader class of alternatives than the tests which have been proposed so far . here
our approach is based on an estimator of a weighted minimum @xmath1-distance between the true copula and the class of functions satisfying ( [ 12 ] ) and the corresponding tests are consistent with respect to all positive quadrant dependent alternatives satisfying weak differentiability assumptions of first order . to our best knowledge , this method provides the first test in this context which is consistent against such a general class of alternatives .
moreover , in contrast to @xcite and @xcite we also provide a weak convergence result under fixed alternative which can be used for studying the power of the test .
the remaining part of the paper is organized as follows . in section [ secevd ] , we consider the approximation problem from a theoretical point of view .
in particular , we derive explicit representations for the minimal @xmath1-distance between the logarithm of the copula and its best approximation by a function of the form ( [ 12 ] ) , which will be the basis for all statistical applications in this paper .
the new estimators , say @xmath27 , are defined in section [ secmindist ] , where we also prove weak convergence of the process @xmath28}$ ] in the space of uniformly bounded functions on the interval @xmath29 $ ] under appropriate assumptions on the weight function used in the @xmath1-distance .
furthermore , we give a theoretical and empirical comparison of the new estimators with the estimators proposed in @xcite .
we will also determine `` optimal '' estimators in the proposed class by minimizing the asymptotic mse with respect to the choice of the weight function used in the @xmath1-distance .
in particular , we demonstrate that some of the new estimators have a substantially smaller asymptotic variance than the estimators proposed by the last - named authors .
we also provide a simulation study in order to investigate the finite sample properties of the different estimates . in section [ sectestevd ] , we introduce and investigate the new test of extreme - value dependence .
in particular , we derive the asymptotic distribution of the test statistic under the null hypothesis as well as under the alternative . in order to approximate the critical values of the test
, we introduce a multiplier bootstrap procedure , prove its consistency and study its finite sample properties by means of a simulation study . finally ,
most of the technical details are deferred to the .
let @xmath30 denote the set of all functions @xmath31\rightarrow[1/2,1 ] , $ ] and define @xmath32 as the copula corresponding to independent random variables , that is , @xmath33 . throughout this paper , we assume that the copula @xmath11 satisfies @xmath34 which holds for any extreme - value copula due to the lower bound for the function @xmath15 .
as pointed out by @xcite , this property is equivalent to the concept of positive quadrant dependence , that is , @xmath35 for a copula with this property , we define the weighted @xmath1-distance @xmath36 where @xmath37\rightarrow{\mathbb{r}}^+$ ] is a continuous weight function . the following result is essential for our approach and provides an explicit expression for the best @xmath1-approximation of the logarithm of the copula by the logarithm of a function of the form ( [ 11 ] ) and as a by - product characterizes the function @xmath38 minimizing @xmath39 .
[ theobestapprox ] assume that the given copula satisfies @xmath40 for some @xmath41 and that the weight function @xmath42 satisfies @xmath43 . then the function @xmath44 is unique and given by @xmath45 where the associated weight function @xmath46 is defined by @xmath47 and @xmath48 moreover , the minimal @xmath1-distance between the logarithms of the given copula and the class of functions of the form ( [ 12 ] ) is given by @xmath49 since @xmath40 , we get @xmath50 and thus@xmath51 and all integrals exist . rewriting the @xmath1 distance in ( [ 21 ] ) gives @xmath52 and the assertion is now obvious . note that @xmath53 if @xmath11 is an extreme - value copula of the form ( [ 11 ] ) with pickands dependence function @xmath15 .
furthermore , the following lemma shows that the minimizing function @xmath38 defined in ( [ astar ] ) satisfies the boundary conditions of pickands dependence functions .
[ conditionpick ] assume that @xmath11 is a copula satisfying @xmath54 .
then the function @xmath38 defined in ( [ astar ] ) has the following properties : @xmath55 , @xmath56 , @xmath57 .
assertion ( i ) is obvious . for a proof of ( ii ) , one uses the frchet hoeffding bound @xmath58 [ see , e.g. , @xcite ] and obtains the assertion by a direct calculation . similarly ,
assertion ( iii ) follows from the inequality @xmath34 .
unfortunately , the function @xmath38 is in general not convex for every copula satisfying @xmath34 .
a counterexample can be derived from theorem 3.2.2 in @xcite and is given by the following shuffle of the copula @xmath59 : @xmath60 ^ 2 $ } , \vspace*{2pt}\cr \min\bigl\{u , v+1/2-\sqrt{1/2}\bigr\ } , \vspace*{2pt}\cr \hspace*{91pt}(u , v)\in\bigl[0,\sqrt{1/2}\bigr]\times\bigl[\sqrt{1/2},1\bigr],\vspace*{2pt}\cr \min\bigl\{u+1/2-\sqrt{1/2},v\bigr\ } , \vspace*{2pt}\cr \hspace*{91pt}(u , v)\in \bigl[\sqrt{1/2},1\bigr]\times\bigl[0,\sqrt{1/2}\bigr],\vspace*{2pt}\cr \min\bigl\{u , v , u+v+1/2 - 2\sqrt{1/2}\bigr\ } , \vspace*{2pt}\cr \hspace*{91pt}\mbox{$(u , v)\in\bigl[\sqrt{1/2},1\bigr]^2$},}\ ] ] for which an easy calculation shows that the mapping @xmath61 is not convex .
consequently , one can find a weight function @xmath42 such that the corresponding best approximating function @xmath38 is not convex .
with the notation @xmath62 the function @xmath38 is convex ( for every weight function @xmath42 ) if and only if the function @xmath63 is convex for every @xmath64 .
the following lemma is now obvious .
[ convex ] if the function @xmath65 is twice differentiable and the inequality @xmath66 ^ 2 \geq f_y''(t ) f_y(t)\ ] ] holds for every @xmath67 , then the best approximation @xmath38 defined by ( [ astar ] ) is convex .
it is worthwhile to mention that the function @xmath38 is convex for some frequently considered classes of copulas , which will be illustrated in the following examples .
[ claytonastar ] consider the clayton copula @xmath68 then a tedious calculation yields @xmath69 ^ 2-f_y''(t)f_y(t ) \\ & & \qquad= \theta\log^2(y ) \ { c_{\mathrm { clayton}}(y^{1-t},y^t;\theta)\ } ^{2 + 2\theta}\bigl ( 4y^{-\theta}-y^{-\theta t}-y^{-\theta(1-t ) } \bigr)\\ & & \qquad\geq \theta\log^2(y ) \
{ c_{\mathrm { clayton}}(y^{1-t},y^t;\theta)\}^{2 + 2\theta } ( 3y^{-\theta}- 1 ) \geq0,\end{aligned}\ ] ] where the inequalities follow observing that @xmath70 and @xmath71 .
therefore , we obtain from lemma [ convex ] that the best approximation @xmath38 is convex and corresponds to an extreme - value copula .
[ exhalphahk ] in the following , we discuss the weight function @xmath72 ( @xmath73 ) with associated function @xmath74 , which will be used later for the construction of the new estimators of pickands dependence function . on the one hand
this choice is made for mathematical convenience , because it allows an explicit calculations of the asymptotic variance @xmath38 in specific examples .
on the other hand , estimates constructed on the basis of this weight function turn out to have good asymptotic and finite sample properties ( see the discussion in section [ simulation ] ) .
it follows that @xmath75 and @xmath76 which simplifies in the case @xmath77 to the representation @xmath78 @xmath79 in the following , we calculate the minimal distance @xmath80 and its corresponding best approximation @xmath38 for two copula families and the associated weight function @xmath81 from example [ exhalphahk ] .
first , we investigate the gaussian copula defined by @xmath82 where @xmath83 is the standard normal distribution function and @xmath84 is the distribution function of a bivariate normal random variable with standard normally distributed margins and correlation @xmath85 $ ] . for the limiting cases @xmath86 and @xmath87 , we obtain the independence and perfect dependence copula , respectively , while for @xmath88 the copula @xmath89 is not an extreme - value copula .
the minimal distances are plotted as a function of @xmath90 in the left part of the first line of figure [ picdist ] . in the right part
, we show some functions @xmath38 corresponding to the best approximation of the logarithm of the gaussian copula by a function of the form ( [ 12 ] ) .
we note that all functions @xmath38 are convex although @xmath89 is only an extreme value copula in the case @xmath86 . for the gaussian copula ( as a function of its correlation coefficient ) and for the convex combination of a gumbel and a clayton copula ( as a function of the parameter @xmath91 in the convex combination ) .
right : the functions @xmath38 corresponding to the best approximations by functions of the form ( [ 12 ] ) . ] in the second example , we consider a convex combination of a gumbel copula with parameter @xmath92 ( corresponding to a coefficient of tail dependence of @xmath93 ) and a clayton copula with parameter @xmath94 , that is , @xmath95,\ ] ] where the clayton copula is given in ( [ copclay ] ) and the gumbel copula is defined by @xmath96 note that only the gumbel copula is an extreme - value copula and obtained for .
the minimal distances are depicted in the left part of the lower panel of figure [ picdist ] as a function of @xmath91 . in the right part , we show the functions @xmath38 corresponding to the best approximation of the logarithm of @xmath97 by a function of the form ( [ 12 ] ) . again all approximations are convex , which means that @xmath38 corresponds in fact to an extreme value copula .
let @xmath98 denotea sample of independent identically distributed bivariate random variables with copula @xmath11 and marginals @xmath6 and @xmath7 .
most of the estimates which have been proposed in the literature so far are based on the fact that the random variable @xmath99 is exponentially distributed with parameter @xmath100 . in particular , we have @xmath101 = 1/ a(t)$ ] .
if the marginal distributions would be known , an estimate of @xmath100 could be obtained by the method of moments . in the case of unknown marginals
, @xcite proposed to replace @xmath6 and @xmath7 by their empirical counterparts and obtained @xmath102 as a rank - based version of pickands estimate , where @xmath103 and @xmath104 denote the ( slightly modified ) empirical distribution functions of the samples @xmath105 and @xmath106 at the points @xmath107 and @xmath108 , respectively .
similarly , observing the identity @xmath109 = - \log a(t ) - \gamma$ ] ( here @xmath110 denotes euler s constant ) , they obtained a rank - based version of the estimate proposed by @xcite , that is , @xmath111 for illustrative purposes , we finally recall two integral representations for the rank - based version of pickands and cfg estimate , which we use in section [ subsecmundzest ] to put all estimates considered in this paper in a general context , that is , @xmath112 \label{cfgdef } \gamma+ \log\hat a^{\mathrm{cfg}}_{n , r } ( t ) & = & \int^1_0 \frac{\hat c_n ( y^{1-t } , y^t ) - i \ { y > e^{-1 }
\}}{\log y}\,dy,\end{aligned}\ ] ] where @xmath113 denotes the empirical copula and @xmath114 , @xmath115 are defined in ( [ 31a ] ) [ see @xcite for more details ] .
theorem [ theobestapprox ] suggests to define a class of new estimators for pickands dependence function by replacing the unknown copula in ( [ astar ] ) through the empirical copula defined in ( [ 31 ] ) .
the asymptotic properties of the corresponding estimators will be investigated in this section . for technical reasons , we require that the argument in the logarithm in the representation ( [ astar ] ) is positive and propose to use the estimator @xmath116 where the constant @xmath117 satisfies @xmath118 and the empirical copula @xmath119 is defined in ( [ 31 ] ) . for the subsequent proofs
, we will need a result on the weak convergence of the empirical copula process with estimated margins . while this problem has been considered by many authors [ see , e.g. , @xcite , @xcite or @xcite among others ] , all of them assume that the copula has continuous partial derivatives on the whole unit square @xmath29 ^ 2 $ ] .
however , as was pointed out by @xcite , there is only one extreme - value copula that has this property .
luckily , in a remarkable paper @xcite was able to show that the following condition is sufficient for weak convergence of the empirical copula process @xmath120 ^ 2 | u_j \in(0,1)\}\ ] ] ( @xmath121 ) .
this condition can be shown to hold for any extreme - value copula with continuously differentiable pickands function @xmath15 [ see @xcite ] . moreover , under this assumption , the process @xmath122 shows the same limiting behavior as the empirical copula process @xmath123 , that is , @xmath124 where the symbol @xmath125 denotes weak convergence in @xmath126 ^ 2 $ ] . here , @xmath127 is a gaussian field on the square @xmath29 ^ 2 $ ] which admits the representation @xmath128 where @xmath129 is a bivariate pinned @xmath11-brownian sheet on the square @xmath29 ^ 2 $ ] with covariance kernel given by @xmath130 and the minimum @xmath131 is understood component - wise . observing the representation ( [ astar ] )
, we obtain the estimator @xmath132 for pickands dependence function , where @xmath133 is defined in ( [ tildecn ] ) .
note that this relation specifies an infinite - dimensional class of estimators indexed by the set of all admissible weight functions .
the following results specify the asymptotic properties of these estimators .
we begin with a slightly more general statement , which shows weak convergence for the weighted integrated process @xmath134 where the weight function @xmath135 ^ 2 \rightarrow\bar{\mathbb{r}}$ ] depends on @xmath136 and @xmath137 .
the result ( and some arguments in its proof ) are also needed in section [ sectestevd ] .
[ theoh1 ] assume that for the weight function @xmath138 ^ 2 \rightarrow\bar{\mathbb{r}}$ ] there exists a function @xmath139 \rightarrow\bar{\mathbb{r}}_0^+$ ] such that @xmath140 ^ 2\qquad|w(y , t)|&\leq&\bar w(y),\\[-2pt ] \label{cw2 } \forall\varepsilon>0\qquad\sup_{y \in[\varepsilon,1 ] } \bar w(y ) & < & \infty,\\[-2pt ] \label{cw3 } \int_0 ^ 1 \bar w(y)y^{-\lambda}\,dy&<&\infty\end{aligned}\ ] ] for some @xmath141 .
if the copula @xmath11 satisfies ( [ asproc ] ) and @xmath142 , then we have for any @xmath143 as @xmath144 @xmath145\\[-10pt ] & & \hspace*{4pt}\quad{\stackrel{w}{\rightsquigarrow}}\quad{\mathbb{w}}_{c , w}(t ) = \int_0 ^ 1 \frac{{\mathbb{g}}_c(y^{1-t } , y^t)}{c(y^{1-t},y^t ) } w(y , t ) \,dy \nonumber\end{aligned}\ ] ] in @xmath146 $ ] .
the following result is now an immediate consequence of theorem [ theoh1 ] using @xmath147 [ recall the definition of the associated weight function @xmath46 in ( [ hstar ] ) ] and yields the weak convergence of the process for a broad class of weight functions .
[ theopickest1 ] if the copula @xmath142 satisfies condition ( [ asproc ] ) and the weight function @xmath42 satisfies the conditions @xmath148 } \biggl| \frac{h^*(y)}{\log y } \biggr| & < & \infty,\\ \label{20c } \int_0 ^ 1 h^*(y ) ( -\log y)^{-1 } y^{-\lambda } \,dy&<&\infty\end{aligned}\ ] ] for some @xmath149 , then we have for any @xmath150 as @xmath151 @xmath152,\ ] ] where the process @xmath153 is given by @xmath154
\(a ) conditions ( [ 20a ] ) and ( [ 20c ] ) restrict the behavior of the function @xmath46 near the boundary of the interval @xmath29 $ ] .
a simple sufficient condition for ( [ 20a ] ) and ( [ 20c ] ) is given by @xmath155 } \biggl| \frac{h^*(x)}{x^\alpha(1-x)^\beta } \biggr| < \infty\ ] ] for some @xmath156 .
in this case , @xmath157 can be chosen as @xmath158 .
\(b ) in the construction discussed so far , it is also possible to use weight functions that depend on @xmath137 , that is , functions of the form @xmath159 .
as long as @xmath160 for @xmath161 $ ] , the corresponding best approximation @xmath38 will still be well defined and correspond to the pickands dependence function if @xmath11 is an extreme - value copula . theorem [ theoh1 ]
provides the asymptotic properties of the corresponding estimator @xmath15 if we set @xmath162 and assume that @xmath163 for all @xmath137 .
however , for the sake of a clear presentation , we will only use weight functions that do not depend on @xmath137 .
note that theorem [ theopickest1 ] is also correct if the given copula is not an extreme - value copula . in other words
: it establishes weak convergence of the process @xmath164 to a centered gaussian process , where @xmath38 denotes the function corresponding to the best approximation of the logarithm of the copula @xmath11 by a function of the form ( [ 12 ] ) .
if @xmath38 is convex , it corresponds to an extreme - value copula and coincides with pickands dependence function .
note also that theorem [ theopickest1 ] excludes the case @xmath165 , because condition ( [ 20c ] ) is not satisfied for this weight function . nevertheless , under the additional assumption that @xmath11 is an extreme - value copula with twice continuously differentiable pickands dependence function @xmath15 , the assertion of the preceding theorem is still valid .
[ theopickest2 ] assume that @xmath11 is an extreme - value copula with twice continuously differentiable pickands dependence function @xmath15 .
for the weight function @xmath166 , we have for any @xmath167 as @xmath144 @xmath168 in @xmath146 $ ] , where @xmath169 .
\(a ) if the marginals of @xmath5 are independent the distribution of the random variable @xmath170 coincides with the distribution of the random variable @xmath171 , which appears as the weak limit of the appropriately standardized pickands estimator ; see @xcite .
in fact , a much more general statement is true : by using weight functions @xmath159 depending on @xmath137 it is possible to obtain for any extreme - value copula estimators of the form ( [ 33 ] ) which show the same limiting behavior as the estimators proposed by @xcite .
this already indicates that for any extreme - value copula it is possible to find weight functions which will make the new minimum distance estimators asymptotically at least as efficient ( in fact better , as will be shown in section [ subsecoptweight ] ) as the estimators introduced by @xcite .
\(b ) a careful inspection of the proof of theorem [ theoh1 ] reveals that the condition @xmath34 can be relaxed to @xmath40 for some @xmath172 , if one imposes stronger conditions on the weight function .
\(c ) the estimator depends on the parameter @xmath117 which is used for the construction of the statistic @xmath173 .
this modification is only made for technical purposes and from a practical point of view the behavior of the estimators does not change substantially provided that @xmath117 is chosen larger than @xmath174 .
[ bayes ] the new estimators can be alternatively motivated observing that the identity ( [ 11 ] ) yields the representation @xmath175 for any @xmath176 .
this leads to a simple class of estimators , that is , @xmath177 where @xmath178 is the dirac measure at the point @xmath136 and @xmath179 is defined in ( [ tildecn ] ) . by averaging these estimators with respect to a distribution , say @xmath180 , we obtain estimators of the form @xmath181 which coincide with the estimators obtained by the concept of best @xmath1-approximation . in this subsection , we illustrate the results investigating example [ exhalphahk ] discussed at the end of section [ secevd ] . for the associated weight function @xmath182 with @xmath73 , we obtain @xmath183 the process @xmath184}$ ] converge weakly in @xmath185 $ ] to the process@xmath186}$ ] , which is given by @xmath187 consequently , for @xmath188 , the asymptotic variance of @xmath189 is obtained as @xmath190 where the function @xmath191 is given by @xmath192 in order to find an explicit expression for these variances , we assume that the function @xmath15 is differentiable and introduce the notation @xmath193 where @xmath194 denotes the derivative of @xmath15
. the following results can be shown by similar arguments as given in @xcite ; for details , see @xcite .
[ propvarak ] for @xmath195 $ ] , let @xmath196 and @xmath197 .
if @xmath11 is an extreme - value copula with pickands dependence function @xmath15 , then the variance of the random variable @xmath198 is given by @xmath199 note that the limiting process in ( [ 34 ] ) is a centered gaussian process .
this means that , asymptotically , the quality of the new estimators [ as well as of the estimators of @xcite , which show a similar limiting behavior ] is determined by the variance .
based on these observations , we will now provide an asymptotic comparison of the new estimators @xmath200 with the estimators investigated by @xcite .
some finite sample results will be presented in the following section for various families of copulas . for the sake of brevity , we restrict ourselves to the independence copula @xmath32 , for which @xmath201 . in the case
@xmath77 , we obtain from proposition [ propvarak ] the same variance as for the rank - based version of pickands estimator , that is , @xmath202 [ see corollary 3.4 in @xcite ] while the case @xmath203 yields @xmath204 investigating the derivative in @xmath205 , it is easy to see that @xmath206 is strictly decreasing in @xmath205 with @xmath207 therefore , we have @xmath208 for all @xmath209 with strict inequality for all @xmath203 . this means that for the independence copula all estimators obtained by our approach with associated weight function @xmath210 , @xmath211 , have a smaller asymptotic variance than the rank - based version of pickands estimator . on the other hand
, a comparison with the cfg estimator proposed by @xcite does not provide a clear picture about the superiority of one estimator and we defer this comparison to the following section , where optimal weight functions for the new estimates @xmath212 are introduced . in this section
, we discuss asymptotically optimal weight functions corresponding to the class of estimates introduced in section [ subsecnewest ] .
as pointed out in the previous section , from an asymptotic point of view the mean squared error of the estimates is dominated by the variance and therefore we concentrate on weight functions minimizing the asymptotic variance of the estimate @xmath213 .
the finite sample properties of the mean squared error of the various estimates will be investigated by means of a simulation study in section [ simulation ] .
note that an optimal weight function depends on the point @xmath137 where pickands dependence function has to be estimated and on the unknown copula .
therefore , an estimator with an optimal weight function can not be implemented in concrete applications without preliminary knowledge about the copula . however , it can serve as a benchmark for user - specified weight functions . to be precise ,
observe that by theorem [ theopickest1 ] the variance of the limiting process @xmath214 is of the form @xmath215 where @xmath216 denotes a probability measure on the interval @xmath29 $ ] defined by @xmath217 and the kernel @xmath218 is given by @xmath219.\ ] ] it is easy to see that @xmath220 defines a convex function on the space of all probability measures on the interval @xmath29 $ ] and the existence of a minimizing measure follows if the kernel @xmath221 is continuous on @xmath29 ^ 2 $ ] .
the following result characterizes the minimizer of @xmath220 and is proved in the .
[ effic ] a probability measure @xmath222 on the interval @xmath29 $ ] minimizes @xmath220 if and only if the inequality @xmath223 is satisfied for all @xmath224 } $ ] .
theorem [ effic ] can be used to check the optimality of a given weight function .
for example , if the copula @xmath11 is given by the independence copula @xmath32 we have @xmath225 and it is easy to see that none of the associated weight functions @xmath210 with @xmath73 is optimal in the sense that it minimizes the asymptotic variance of the estimate @xmath226 with respect to the choice of the weight function .
on the other hand , the result is less useful for an explicit computation of optimal weight functions .
deriving an analytical expression for the optimal weight function seems to be impossible , even for the simple case of the independence copula .
however , approximations to the optimal weight function can easily be computed numerically . to be precise
we approximate the double integral appearing in the representation of @xmath227 by the finite sum @xmath228 where @xmath229 , @xmath230 denotes an @xmath231 matrix , @xmath232 is an vector of length @xmath233 and @xmath234 $ ] represents the mass of @xmath216 allocated to the interval @xmath235 $ ] ( @xmath236 ) . minimizing the right - hand side of the above equation with respect to @xmath237 under the constrains @xmath238 , @xmath239 is a quadratic ( convex ) optimization problem which can be solved by standard methods ; see , for example , @xcite and approximations of the optimal weight function can be calculated with arbitrary precision by increasing @xmath233 . in the remaining part of this section
, we will compare the asymptotic variance of the pickands- , the cfg - estimator proposed by @xcite and the new estimates , where the new estimators are based on the weight functions @xmath240 discussed in section [ exampleexp ] for two values of @xmath205 as well as on the optimal weights minimizing the right - hand side of ( [ diskret ] ) , where we set @xmath241 . in order to compute the solution @xmath242
, we used the routine _ ipop _ from the r - package _ kernlab _ by @xcite . in the left part of figure [ asymvar ]
, we show the asymptotic variances of the different estimators for the independence copula .
we observe that pickands estimator has the largest asymptotic variances ( this curve is not displayed in the figure ) , while the cfg estimator of @xcite yields smaller variances than the estimator @xmath243 , but larger asymptotic variances than the estimators @xmath244 . on the other hand , the estimate @xmath245 corresponding to the numerically determined optimal weight function yields a substantially smaller variance than all other estimates under consideration . in the right - hand part of figure [ asymvar ]
, we display the corresponding results for the asymmetric negative logistic model [ see @xcite ] @xmath246 with parameters @xmath247 and @xmath248 chosen such that the coefficient of tail dependence is @xmath249 .
we observe that the estimate @xmath250 yields the largest asymptotic variance .
the cfg estimate proposed by @xcite and the estimate @xmath243 show a similar behavior ( with minor advantages for the latter ) , while the best results are obtained for the new estimate corresponding to the optimal weight function .
we conclude this section with the remark that we have presented a comparison of the different estimators based on the asymptotic variance which determines the mean squared error asymptotically . for finite samples , minimizing only the variance might increase the bias and therefore the asymptotic results can not directly be transferred to applications . in the finite sample study presented in section [ simulation ]
, we will demonstrate that not all of the asymptotic results yield good predictions for the finite - sample behavior of the corresponding estimators . in general ,
all of the estimates discussed so far [ including those proposed by @xcite ] will neither be convex , nor will they satisfy the other characterizing properties of pickands dependence functions .
however , the literature provides many proposals on how to enforce these conditions .
various endpoint corrections have been proposed by @xcite , @xcite or @xcite among others .
@xcite proposed an @xmath1-projection of the estimate of pickands dependence function on a space of partially linear functions which is arbitrarily close to the space of all convex functions in @xmath30 satisfying the conditions of lemma [ conditionpick ] .
they also showed that this transformation decreases the @xmath1-distance between the `` true '' dependence function and the estimate .
an alternative concept of constructing convex estimators is based on the greatest convex minorant , which yields a decrease in the sup - norm , that is , @xmath251 where @xmath252 is any initial estimate of pickands dependence function and @xmath253 its greatest convex minorant [ see , e.g. , @xcite , @xcite , @xcite among others ] .
it is also possible to combine this concept with an endpoint correction calculating the greatest convex minorant of the function @xmath254 [ see @xcite who also proposed alternative special endpoint corrections for their estimators ] .
all these methods can be used to produce an estimate of @xmath15 which has the characterizing properties of a pickands dependence function .
as mentioned in the , a broader class of estimates could be obtained by minimizing more general distances between the given copula and the class of functions defined by ( [ 11 ] ) and in this paragraph we briefly indicate this principle .
consider the best approximation of the copula @xmath11 by functions of the form ( [ 11 ] ) with respect to the distance @xmath256 where @xmath257 \times[0,1 ] \rightarrow{\mathbb{r}}^+_0 $ ] denotes a `` distance '' and @xmath258 is a given weight function .
note that the minimization in ( [ mabstand ] ) can be carried out by separately minimizing the inner integral for every value of @xmath137 .
consequently , the problem reduces to a one - dimensional minimization problem and assuming differentiability it follows that for fixed @xmath137 the optimal value @xmath0 minimizing the interior integral in ( [ mabstand ] ) is obtained as a solution of the equation @xmath259 under suitable assumptions , integration and differentiation can be exchanged and we have @xmath260 where @xmath261 denotes the derivative of @xmath83 with respect to the second argument . in general , the solution of ( [ zest ] ) is only defined implicitly as a functional of the copula @xmath11 .
therefore , if @xmath11 is replaced through the empirical copula the analysis of the stochastic properties of the corresponding process turns out to be extremely difficult because in many cases one has to control improper integrals ( see the proofs of theorems [ theoh1 ] and [ theopickest2 ] in the ) . for the sake of a clear exposition ,
we do not discuss details in this paper and defer these considerations to future research .
nevertheless , equation ( [ zest ] ) yields a different view on the estimation problem of pickands dependence function .
note that the estimate introduced in section [ subsecnewest ] is obtained by the choice @xmath262 and @xmath263 in ( [ zest ] ) .
this estimate corresponds to a minimum distance estimate .
similarly , an estimate corresponding to the classical @xmath1-distance is obtained for the choice @xmath264 this yields for ( [ zest ] ) the equation @xmath265 which can not be solved analytically .
the rank - based versions of pickands and the cfg estimator proposed by @xcite do not correspond to @xmath255-estimates , but could be considered as @xmath26-estimates obtained from ( [ zest ] ) for the function @xmath266 with @xmath267 with @xmath268 and @xmath269 , @xmath270 , respectively .
in fact , this choice leads to a general class of estimators which relates the pickands and the cfg estimate in an interesting way . to be precise ,
note that for @xmath271 equation ( [ zest ] ) yields @xmath272 here the case @xmath270 has to be interpreted as the limit @xmath273 , which yields a generalization of the defining equation for the cfg estimate , that is , @xmath274 observing the relation @xmath275 we obtain the defining equation for the estimate proposed by @xcite [ see ( [ cfgdef ] ) ] .
similarly , if @xmath271 it follows from ( [ oneclass ] ) @xmath276 and we obtain a defining equation for a generalization of the pickands estimate . the classical case is obtained for @xmath277 [ see @xcite or equation ( [ pickdef ] ) ] , but ( [ oneclass1 ] ) defines many other estimates of this type . therefore , the pickands and the cfg estimate correspond to the extreme cases in the class @xmath278 \}$ ] .
we finally note that there are numerous other functions @xmath279 , which could be used for the construction of alternative @xmath26-estimates , but most of them do not lead to an explicit solution for @xmath0 . in this sense
the cfg - estimator , pickands - estimator and the estimates proposed in this paper could be considered as attractive special cases , which can be explicitly represented in terms of an integral of the empirical copula . in this subsection , we investigate the small sample properties of the new estimators by means of a simulation study .
especially , we compare the new estimators with the rank - based estimators suggested by @xcite , which are most similar in spirit with the method proposed in this paper .
we study the finite sample behavior of the greatest convex minorants of the endpoint corrected versions of the various estimators .
the new estimators are corrected in a first step by @xmath280 and in a second step the greatest convex minorant of @xmath281 is calculated .
for the rank - based cfg and pickands estimators , we first used the endpoint corrections proposed in @xcite , then applied ( [ merk ] ) and finally calculated the greatest convex minorant .
hereby , we compare the performance of the different statistical procedures which will be used in concrete applications and apply the corrections , that are most favorable for the respective estimators .
the greatest convex minorants are computed using the routine _ gcmlcm _ from the package _ fdrtool _ by @xcite .
all results presented here are based on 5,000 simulation runs and the sample size is @xmath282 . as estimators
, we consider the statistics defined in ( [ 33 ] ) with the weight function @xmath283 and the optimal weight function determined in section [ subsecoptweight ] .
an important question is the choice of the parameter @xmath205 for the statistic @xmath189 in order to achieve a balance between bias and variance .
for this purpose , we first study the performance of the estimator @xmath189 with respect to different choices for the parameter @xmath205 and consider the asymmetric negative logistic model defined in ( [ asyneg ] ) and the symmetric mixed model [ see @xcite ] defined by @xmath284.\ ] ] the results for other copula models are similar and are omitted for the sake of brevity .
for the pickands dependence function ( [ asyneg ] ) , we used the parameters @xmath285 and @xmath286 such that the coefficient of tail dependence is given by @xmath287 and varies in the interval @xmath288 , while the parameter @xmath289 $ ] used in ( [ symmix ] ) yields @xmath290 $ ] .
the quality of an estimator @xmath291 is measured with respect to mean integrated squared error @xmath292,\ ] ] which was computed by taking the average over 5,000 simulated samples .
the new estimators turned out to be rather robust with respect to the choice of the parameter @xmath117 in the definition of the process @xmath173 provided that @xmath293 .
for this reason , we use @xmath294 throughout this section . analyzing the impact of choosing different values for @xmath205 , in figure [ picmisek ] we display simulated curves @xmath295 for the asymmetric negative logistic and the mixed models with different coefficients of tail dependence @xmath90
, as well as the maximum over such curves for different values of @xmath90 ( solid curves ) , that is , @xmath296 where by @xmath297 we denote the mise for the tail dependence coefficient @xmath90 . the curves in ( [ effmise ] ) attain their minima in the optimal @xmath205 for the respective @xmath90 , and their shapes provide information about the performance of the estimators for nonoptimal values of @xmath205 .
the solid curve gives an impression about the `` worst case '' scenario ( with respect to @xmath298 ) in every model .
the simulations indicate , that for @xmath282 the optimal values of @xmath205 for different models and tail dependence coefficients lie in the interval @xmath299 $ ] . moreover , for values of @xmath205 in this interval the quality of the estimators remains very stable . for @xmath300 , @xmath301 and additional models
the picture remains quite similar and these results are not depicted for the sake of brevity .
we thus recommend using @xmath302 in practical applications .
note that the asymptotic analysis in section [ subsecoptweight ] suggests that the asymptotically optimal @xmath205 should differ substantially for various models .
however , this effect is not visible for sample size up to @xmath301 . in these cases ,
the optimal values for @xmath205 usually varies in the interval @xmath303 $ ] . ) for various models and coefficients of tail dependence .
the minimum corresponds to the optimal value of @xmath205 in the weight function @xmath304 .
the solid curve corresponds to the worst case defined by ( [ solid ] ) .
the sample size is @xmath282 and the mise is calculated by 5,000 simulation runs .
left panel : asymmetric negative logistic model .
right panel : mixed model . ] next , we compare the new estimators with rank - based versions of pickands and the cfg estimator proposed by @xcite . in figure [ picmise ] , the normalized @xmath305 is plotted as a function of the tail dependence parameter @xmath90 for the asymmetric negative logistic and the mixed model , where the parameter @xmath306 is chosen in such a way , that the coefficient of tail dependence @xmath307 varies over the specific range of the corresponding model . for each sample , we computed the rank - based versions of pickands estimator , the cfg estimator [ see @xcite ] and two of the new estimators @xmath189 ( @xmath308 , @xmath249 ) . in this comparison
, we also include the estimator @xmath309 which uses the optimal weight function determined in section [ subsecoptweight ] . for various estimators , models and coefficients of tail dependence , based on 5,000 samples of size @xmath282 .
] summarizing the results , one can conclude that in general the best performance is obtained for our new estimator based on the weight function @xmath304 with @xmath308 and @xmath310 , in particular if the coefficient of tail dependence is small .
a comparison of the two estimators @xmath311 and @xmath312 shows that the choice @xmath308 performs slightly better than the choice @xmath310 in both models . in both settings ,
the mise obtained by @xmath311 and @xmath312 is smaller than the mise of the cfg estimator proposed in @xcite if the coefficient of tail dependence is small .
on the other hand , the latter estimators yield sightly better results for a large coefficient of tail dependence .
the results for rank - based version of the pickands estimator are not depicted , because this estimator yields a uniformly larger mise .
simulations of other scenarios show similar results and are also not displayed for the sake of brevity .
it is remarkable that the optimal weight function usually yields an estimator with a substantially larger mise than all other estimates if the coefficient of tail dependence is small .
similar results can be observed for the sample size @xmath301 ( these results are not depicted ) .
this indicates that the advantages of the asymptotically optimal weight function only start to play a role for rather large sample sizes .
from the definition of the functional @xmath313 in ( [ 21 ] ) it is easy to see that , for a strictly positive weight function @xmath42 with @xmath314 , a copula function @xmath11 is an extreme - value copula if and only if @xmath315 where @xmath38 denotes the best approximation defined in ( [ astar ] ) .
this suggests to use @xmath316 as a test statistic for the hypothesis ( [ hypextr ] ) , that is , @xmath317 recalling the representation ( [ mcastar ] ) @xmath318 with @xmath319 and defining @xmath320 we obtain the decomposition @xmath321 where the last identity defines the terms @xmath322 and @xmath323 in an obvious manner . note that under the null hypothesis of extreme - value dependence we have @xmath324 and thus @xmath325
this means that under @xmath326 the term @xmath327 will vanish and the asymptotic distribution will be determined by the large sample properties of the random variable @xmath328 . under the alternative , the equality @xmath325 will not hold anymore and it turns out that in this case the statistic is asymptotically dominated by the random variable @xmath327 . with the following results , we will derive the limiting distribution of the proposed test statistic under the null hypothesis and the alternative .
[ testh0 ] assume that the given copula @xmath11 satisfies condition ( [ asproc ] ) and is an extreme - value copula with pickands dependence function @xmath38 . if the function @xmath329 fulfills conditions ( [ cw2 ] ) and ( [ cw3 ] ) for some @xmath330 and the weight function @xmath42 is strictly positive and satisfies assumptions ( [ 20a ] ) , ( [ 20c ] ) for @xmath331 , then we have for any @xmath332 and @xmath144 @xmath333 where the random variable @xmath334 is defined by @xmath335 with @xmath336 and the process @xmath337}$ ] is defined in theorem [ theopickest1 ] .
the next theorem gives the distribution of the test statistic @xmath338 under the alternative .
note that in this case we have @xmath339 .
[ testh1 ] assume that the given copula @xmath11 satisfies @xmath34 , condition ( [ asproc ] ) and that @xmath339 . if additionally the weight function @xmath42 is strictly positive and @xmath42 and the function @xmath329 satisfy the assumptions ( [ 20a ] ) , ( [ 20c ] ) and ( [ cw2 ] ) , ( [ cw3 ] ) for some @xmath149 , respectively , then we have for any @xmath340 and @xmath144 @xmath341 where the random variable @xmath342 is defined as @xmath343 with @xmath344 [ remm ] ( a ) note that the weight functions @xmath345 satisfy the assumptions of theorems [ testh0 ] and [ testh1 ] for @xmath346 and @xmath347 , respectively .
\(b ) the preceding two theorems yield a consistent asymptotic level @xmath91 test for the hypothesis of extreme - value dependence by rejecting the null hypothesis @xmath326 if @xmath348 where @xmath349 denotes the @xmath350-quantile of the distribution of the random variable @xmath334 .
\(c ) by theorem [ testh1 ] , the power of the test ( [ testext ] ) is approximately given by @xmath351 where the function @xmath38 is defined in ( [ astar ] ) corresponding to the best approximation of the logarithm of the copula @xmath11 by a function of the form ( [ 12 ] ) , @xmath191 is the standard deviation of the distribution of the random variable @xmath342 and @xmath83 is the standard normal distribution function .
thus , the power of the test ( [ testext ] ) is an increasing function of the quantity @xmath352 . in general , the distribution of the random variable @xmath334 can not be determined explicitly , because of its complicated dependence on the ( unknown ) copula @xmath11 .
we hence propose to determine the quantiles by the multiplier bootstrap approach as described in @xcite . to be precise ,
let @xmath353 denote independent identically distributed random variables with @xmath354 we define @xmath355 as the mean of @xmath356 and consider the multiplier statistics @xmath357 where @xmath358 and @xmath359 denotes the marginal empirical distribution functions .
if we estimate the partial derivatives of the copula @xmath11 by @xmath360 where @xmath361 , we can approximate the distribution of @xmath127 by the distribution of the process @xmath362 where @xmath363 .
more precisely , it was shown by @xcite that we have weak convergence conditional on the data in probability toward @xmath127 , that is , @xmath364
^ 2,\ ] ] where the symbol @xmath365 denotes weak convergence conditional on the data in probability as defined by @xcite , that is , @xmath366 if @xmath367 ^ 2 ) } | { { \mathbb{e}}_\xi h(\alpha_n^{\mathrm{pdm } } ) - { \mathbb{e}}h({\mathbb{g}}_c)}| \stackrel{{\mathbb{p}}}{\longrightarrow } 0\ ] ] and @xmath368 ^ 2).\ ] ] here @xmath369 ^
2 ) \\ & & \qquad=\ { f\dvtx l^\infty[0,1]^2 \rightarrow{\mathbb{r}}\dvtx \|f\|_\infty\leq1 , |f(\beta)-f(\gamma)| \leq\|\beta -\gamma\|_\infty\\ & & \hspace*{223pt}\forall\gamma,\beta\in l^\infty[0,1]^2 \}\end{aligned}\ ] ] is the class of all uniformly bounded functions which are lipschitz continuous with constant smaller one , and @xmath370 denotes the conditional expectation with respect to the weights @xmath371 given the data @xmath372 . as a consequence ,
we obtain the following bootstrap approximation for @xmath334 .
[ multitest ] if condition ( [ asproc ] ) is satisfied , the weight function @xmath42 satisfies the conditions of theorem [ testh0 ] and @xmath373 is uniformly bounded then @xmath374 converges weakly to @xmath334 conditional on the data , that is , @xmath375.\ ] ] by theorem [ multitest ] , @xmath376 is a valid bootstrap approximation for the distribution of @xmath334 .
consequently , repeating the procedure @xmath377 times yields a sample @xmath378 that is approximately distributed according to @xmath334 and we can use the empirical @xmath350-quantile of this sample , say @xmath379 , as an approximation for @xmath349 . therefore , rejecting the null hypothesis if @xmath380 yields a consistent asymptotic level @xmath91 test for extreme - value dependence .
note that the condition on the boundedness of the function @xmath381 is not satisfied for any member of the class @xmath382 from example [ exhalphahk ] . nevertheless , mimicking the procedure from @xcite and using @xmath383}(y)$ ] instead of @xmath384 is sufficient for the boundedness . since this is the procedure being usually performed in practical applications , theorem [ multitest ] is still valuable for the weight functions investigated in this paper . in this subsection
, we investigate the finite sample properties of the test for extreme - value dependence .
we consider the asymmetric negative logistic model ( [ asyneg ] ) , the symmetric mixed model ( [ symmix ] ) and additionally the symmetric model of gumbel @xmath385 with parameter @xmath386 [ see @xcite ] and the model of hsler and reiss @xmath387 where @xmath248 and @xmath83 is the standard normal distribution function [ see @xcite ] .
the coefficient of tail dependence in ( [ huesler ] ) is given by @xmath388 , that is , independence is obtained for @xmath389 and complete dependence for @xmath390 .
for the gumbel model ( [ gumbel ] ) , complete dependence is obtained in the limit as @xmath306 approaches infinity while independence corresponds to @xmath391 .
the coefficient of tail dependence @xmath307 is given by @xmath392 .
we generated 1,000 random samples of sample size @xmath300 from various copula models and calculated the probability of rejecting the null hypothesis . under the null hypothesis
, we chose the model parameters in such a way that the coefficient of tail dependence @xmath90 varies over the specific range of the corresponding model . under the alternative
, the coefficient of tail dependence does not need to exist and we therefore chose the model parameters , such that kendall s @xmath393 is an element of the set @xmath394 .
the weight function is chosen as @xmath395 and the critical values are determined by the multiplier bootstrap approach as described in section [ mulitplier ] with @xmath396 bootstrap replications .
the results are stated in table [ tabtest ] .
@c@ .simulated rejection probabilities of the test ( [ testboot ] ) for the null hypothesis of an extreme - value copula for various models .
the first four columns deal with models under the null hypothesis , while the last four are from the alternative [ cols="<,^,^,^",options="header " , ] @l@c@c@c@ @xmath397**-model * * & @xmath398 & * 0.05 * & * 0.1 * + clayton & 0.25 & 0.874 & 0.916 + & 0.5 & 1 & 1 + & 0.75 & 0.999 & 1 + frank & 0.25 & 0.291 & 0.396 + & 0.5 & 0.73 & 0.822 + & 0.75 & 0.783 & 0.898 + gaussian & 0.25 & 0.168 & 0.240 + & 0.5 & 0.237 & 0.336 + & 0.75 & 0.084 & 0.156 + @xmath399 & 0.25 & 0.105 & 0.187 + & 0.5 & 0.158 & 0.263 + & 0.75 & 0.046 & 0.092 + we observe from the left part of table [ tabtest ] that the level of test is accurately approximated for most of the models , if the tail dependence is not too strong . for a large tail dependence coefficient
the bootstrap test is conservative .
this phenomenon can be explained by the fact that for the limiting case of random variables distributed according to the upper frchet hoeffding the empirical copula @xmath119 does not converge weakly to a nondegenerate process at a rate @xmath400 , rather in this case it follows that @xmath401 .
consequently , the approximations proposed in this paper , which are based on the weak convergence of @xmath402 to a nondegenerate process , are not appropriate for small samples , if the tail dependence coefficient is large .
considering the alternative , we observe reasonably good power for the frank and clayton copulas , while for the gaussian or @xmath137-copula deviations from an extreme - value copula are not detected well with a sample size @xmath300 . in some cases ,
the power of the test ( [ testboot ] ) is close the nominal level .
this observation can be again explained by the closeness to the upper frchet hoeffding bound .
indeed , we can use the minimal distance @xmath403 as a measure of deviation from an extreme - value copula . calculating the minimal distance @xmath403 ( with kendall s @xmath404 and @xmath405 )
, we observe that the minimal distances are about ten times smaller for the gaussian and @xmath399 than for the frank and clayton copula , that is , @xmath406 moreover , as explained in remark [ remm](b ) the power of the tests ( [ testext ] ) and ( [ testboot ] ) is an increasing function of the quantity @xmath407 . for the four copulas considered in the simulation study ( with @xmath404 ) ,
the corresponding ratios are approximately given by @xmath408 which provides some theoretical explanation of the findings presented in table [ tabtest ] .
loosely speaking , if the value @xmath352 is very small a larger sample size is required to detect a deviation from an extreme - value copula .
this statement is confirmed by further simulations results .
for example , for the gaussian and @xmath399 copula ( with kendall s @xmath409 ) we obtain for the sample size @xmath301 the rejection probabilities @xmath410 @xmath411 and @xmath412 @xmath413 for the bootstrap test with level @xmath414 ( @xmath415 ) , respectively .
proof of theorem [ theoh1 ] fix @xmath149 as in ( [ cw3 ] ) and @xmath150 . due to lemma 1.10.2(i ) in @xcite , the process
@xmath416 will have the same weak limit ( with respect to the @xmath125 convergence ) as @xmath417 .
for @xmath418 we consider the following random functions in @xmath126 $ ] : @xmath419 we prove the theorem by an application of theorem 4.2 in @xcite , adapted to the concept of weak convergence in the sense of hoffmann jrgensen , see , for example , @xcite .
more precisely , we will show in lemma [ lema1 ] in appendix [ seclem ] that the weak convergence @xmath420 in @xmath126 $ ] follows from the following three assertions : @xmath421 , \nonumber\\ & & \hphantom{\mbox{i}}\mbox{(ii)}\quad w_i { \stackrel{w}{\rightsquigarrow}}w\qquad \mbox{for } i\rightarrow\infty \mbox { in } l^\infty[0,1],\nonumber\\[-8pt]\\[-8pt ] & & \mbox{(iii ) } \quad\mbox{for every } \varepsilon>0\nonumber\\ & & \hphantom{\mbox{(iii)}}\quad\qquad\lim_{i\rightarrow \infty}\limsup_{n\rightarrow\infty } { \mathbb{p}}^*\bigl({\sup_{t\in[0,1 ] } } main part of the proof now consists in the verification assertion ( iii ) .
we begin by proving assertion ( i ) . for this purpose , set @xmath422
^ 2 $ ] and consider the mapping @xmath423 where its domain @xmath424 is defined by @xmath425 . by lemma 12.2 in @xcite
, it follows that @xmath426 is hadamard - differentiable at @xmath11 , tangentially to @xmath427 , with derivative @xmath428 . since @xmath429 and @xmath34 we have @xmath430 and the functional delta method
[ see theorem 2.8 in @xcite ] yields @xmath431 in @xmath432 .
next , we consider the operator @xmath433\times[0,1]),\cr f \mapsto f\circ\varphi , } \ ] ] where the mapping @xmath434\times[0,1]\rightarrow t_i$ ] is defined by @xmath435 .
observing @xmath436\times[0,1 ] } |f\circ\varphi(y , t ) - g\circ \varphi(y , t ) | \leq\sup_{{\mathbf{x}}\in t_i } |f({\mathbf{x}})-g({\mathbf{x}})|\ ] ] we can conclude that @xmath437 is lipschitz - continuous . by the continuous mapping theorem
[ see , e.g. , theorem 7.7 in @xcite ] and conditions ( [ cw1 ] ) and ( [ cw2 ] ) , we immediately obtain @xmath438 in @xmath439\times[0,1])$ ] . the assertion in ( i )
now follows by continuity of integration with respect to the variable @xmath136 . for the proof of assertion ( ii ) , we simply note that @xmath127 is bounded on @xmath29 ^ 2 $ ] and that @xmath440 is uniformly bounded with respect to @xmath441 $ ] by the integrable function @xmath442 . for the proof of assertion ( iii )
, choose some @xmath443 such that @xmath444 and consider the decomposition @xmath445 where @xmath446 and @xmath447\\[-8pt ] i_{b^{(2)}_i ( t ) } & = & ( 0,1 ) \setminus i_{b^{(1)}_i ( t)}.\nonumber\end{aligned}\ ] ] the usual estimate @xmath448 } |w_{i , n}(t)-w_n(t)| > \varepsilon\bigr)\nonumber\\[-8pt]\\[-8pt ] & & \qquad\leq { \mathbb{p}}^*\bigl(\sup_{t\in[0,1]}\bigl|b^{(1)}_i(t)\bigr| > \varepsilon /2\bigr ) + { \mathbb{p}}^*\bigl(\sup_{t\in[0,1]}\bigl|b^{(2)}_i(t)\bigr| > \varepsilon/2\bigr)\nonumber\end{aligned}\ ] ] allows for individual investigation of both expressions , and we begin with the term @xmath449 } } |b^{(1)}_i(t)|$ ] . by the mean value theorem applied to the logarithm
, we have @xmath450\\[-8pt ] & = & ( \tilde c_n - c ) ( y^{1-t } , { y^t } ) \frac{1}{c^*(y , t ) } , \nonumber\end{aligned}\ ] ] where @xmath451 is some intermediate point satisfying @xmath452 . especially , observing @xmath142 we have @xmath453 and therefore @xmath454}\bigl|b^{(1)}_i(t)\bigr| & \leq & \sup_{t\in[0,1 ] } \int_{i_{b^{(1)}_i(t ) } } \sqrt{n}| ( \tilde c_n - c ) ( y^{1-t},y^t ) | \\ & & \hspace*{55.5pt}{}\times\biggl| 1\vee\frac { c}{\tilde c_n } ( y^{1-t},y^t)\biggr| w(y , t ) y^{-1 } \,dy \\ & \leq & \sup_{{\mathbf{x}}\in[0,1]^2}\sqrt{n}|\tilde c_n({\mathbf{x}})-c({\mathbf{x}})|\\ & & { } \times \biggl ( 1\vee\sup_{{\mathbf{x}}\in[0,1]^2 \dvtx c({\mathbf{x}})>n^{-\alpha } } \biggl| \frac{c}{\tilde c_n}({\mathbf{x } } ) \biggr| \biggr ) \times\psi(i)\end{aligned}\ ] ] with @xmath455 for @xmath456 .
this yields for the first term on the right - hand side of ( [ iiihalf ] ) @xmath457}\bigl|b^{(1)}_i(t)\bigr| > \varepsilon\bigr ) & \leq & { \mathbb{p}}^*\biggl(\sup_{{\mathbf{x}}\in[0,1]^2}\sqrt{n}|\tilde c_n({\mathbf{x}})-c({\mathbf{x}})| > \sqrt{\frac{\varepsilon}{\psi(i ) } } \biggr ) \nonumber\\[-8pt]\\[-8pt ] \qquad & & { } + { \mathbb{p}}^*\biggl ( 1\vee\sup_{c({\mathbf{x}})>n^{-\alpha } } \biggl| \frac{c}{\tilde c_n}({\mathbf{x } } ) \biggr| > \sqrt{\frac{\varepsilon}{\psi ( i ) } } \biggr).\nonumber\end{aligned}\ ] ] since @xmath458 ^ 2}\sqrt{n}|\tilde c_n({\mathbf{x}})-c({\mathbf{x}})|$ ] is asymptotically tight , we immediately obtain @xmath459 ^ 2}\sqrt{n}|\tilde c_n({\mathbf{x}})-c({\mathbf{x}})|>\sqrt{\frac { \varepsilon}{\psi(i ) } } \biggr ) = 0.\ ] ] for the estimation of the second term in ( [ iiisqrt ] ) , we note that @xmath460 ^ 2 \dvtx c({\mathbf{x}})>n^{-\alpha } } \biggl| \frac{\tilde c_n({\mathbf{x}})-c({\mathbf{x}})}{c({\mathbf{x } } ) } \biggr| < n^{\alpha } \sup_{{\mathbf{x}}\in[0,1]^2 } |\tilde c_n({\mathbf{x}})-c({\mathbf{x } } ) | \stackrel{{\mathbb{p}}^*}{\longrightarrow } 0,\ ] ] which in turn implies @xmath461\\[-8pt ] & & { } + \biggl ( \sup_{c({\mathbf{x}})>n^{-\alpha } } \biggl| 1+\frac{\tilde c_n - c}{c}({\mathbf{x } } ) \biggr|^{-1}\biggr ) { \mathbb{i}}_{\omega \setminus a_n } \nonumber\\ & \stackrel{{\mathbb{p}}^*}{\longrightarrow } & 1 , \nonumber\end{aligned}\ ] ] where @xmath462 .
this implies that the function @xmath463 can be bounded by a function that converges to one in outer probability , and thus @xmath464 observing ( [ iiisqrt ] ) and ( [ 510a ] ) it remains to estimate the second term on the right - hand side of ( [ iiihalf ] ) .
we make use of the mean value theorem again [ see ( [ mvt ] ) ] but use the estimate @xmath465 [ recall that @xmath149 by assumption ( [ cw3 ] ) ] .
this yields @xmath454}\bigl|b^{(2)}_i(t)\bigr| & \leq & \sup_{t\in[0,1 ] } \int _ { i_{b^{(2)}_i(t ) } } \sqrt{n}| ( \tilde c_n - c ) ( y^{1-t},y^t ) & & \hspace*{56.2pt}{}\times\biggl| 1\vee\frac{c^\lambda}{\tilde c_n } ( y^{1-t},y^t)\biggr| w(y , t ) y^{-\lambda } \,dy \\ & \leq & \sup_{{\mathbf{x}}\in[0,1]^2}\sqrt{n}|\tilde c_n({\mathbf{x}})-c({\mathbf{x}})|\\ & & { } \times \biggl ( 1\vee\sup_{{\mathbf{x}}\in[0,1]^2 \dvtx c({\mathbf{x}})\leq n^{-\alpha } } \biggl| \frac{c^\lambda}{\tilde c_n}({\mathbf{x } } ) \biggr| \biggr ) \times\phi(i),\end{aligned}\ ] ] where @xmath466 for @xmath456 by condition ( [ cw3 ] ) . using analogous arguments as for the estimation of @xmath467}}|b^{(1)}_i(t)| $ ]
the assertion follows from @xmath468 ^ 2 \dvtx c({\mathbf{x}})\leq n^{-\alpha } } \biggl| \frac { c^\lambda}{\tilde c_n}({\mathbf{x } } ) \biggr| \leq\sup_{{\mathbf{x}}\in[0,1]^2 \dvtx c({\mathbf{x}})\leq n^{-\alpha } } | n^\gamma c^\lambda({\mathbf{x } } ) | \leq n^{\gamma-\lambda\alpha } = o(1)\ ] ] due to the choice of @xmath117 and @xmath91 .
proof of theorem [ theopickest2 ] the proof will also be based on lemma [ lema1 ] in appendix [ seclem ] verifying conditions ( i)(iii ) in ( [ cond ] ) .
a careful inspection of the previous proof shows that the verification of condition ( i ) in ( [ cond ] ) remains valid .
regarding condition ( ii ) , we have to show that the process @xmath469 is integrable on the interval @xmath470 . for this purpose , we write @xmath471=0 and consider each term separately . from theorem g.1 in @xcite , we know that for any @xmath472 the process @xmath473 has continuous sample paths on @xmath29 ^ 2 $ ]
. considering @xmath474 and using the notation @xmath475 with @xmath476 it remains to show that there exist integrable functions @xmath477 with @xmath478 for all @xmath195 $ ] @xmath479 . for @xmath480
this is immediate because @xmath481 . for @xmath482 , note that @xmath483 , with @xmath484 .
therefore , @xmath485 where the second estimate follows from the inequality @xmath486 and holds for @xmath487 .
a similar argument works for the term @xmath488 . for the verification of condition ( iii ) , we proceed along similar lines as in the previous proof .
we begin by choosing some @xmath489 and some @xmath490 in such a way that @xmath491 .
first , note that @xmath492 implies @xmath493 for all @xmath195 $ ] .
this yields @xmath494 uniformly with respect to @xmath441 $ ] , and therefore it is sufficient to consider the decomposition in ( [ aibi ] ) with the sets @xmath495 we can estimate the term @xmath496 analogously to the previous proof by @xmath497 let @xmath498 denote the empirical distribution function of the standardized sample @xmath499 . by the results in segers [ ( @xcite ) , section 5 ]
we can decompose @xmath500 as follows : @xmath501 where @xmath502 and the remainder satisfies @xmath503 ^ 2 } } | \tilde r_n({\mathbf{x } } ) |=o\bigl(n^{1/2-\gamma } + n^{-1/4}(\log n)^{1/2}(\log\log n)^{3/4}\bigr ) \qquad\mbox{a.s.}\ ] ] note that the estimate of ( [ tildern ] ) requires validity of condition 5.1 in @xcite .
this condition is satisfied provided that the function @xmath15 is assumed to be twice continuously differentiable ; see example 6.3 in @xcite . with ( [ stute ] )
, we can estimate the term @xmath504 analogously to decomposition ( [ aibi ] ) by @xmath505 , where @xmath506 the decomposition in ( [ stute ] ) , theorem g.1 in @xcite and the inequality @xmath507 may be used to conclude @xmath508 which in turn implies @xmath509 analogously to ( [ opbruch1 ] ) . together with ( [ tildern ] ) and observing the inequality @xmath510 , we obtain , for @xmath144 @xmath511 } b^{(1)}_{i,4}(t ) = o_{{\mathbb{p}}^*}\bigl(n^{1/2-\gamma}\log
n + n^{-1/4}(\log n)^{3/2}(\log\log n)^{1/4}\bigr)=o_{{\mathbb{p}}^*}(1),\ ] ] which implies @xmath512 } b^{(1)}_{i,4}(t ) > \varepsilon/4 \bigr ) = 0.\ ] ] observing that @xmath513 the first term @xmath514 can be estimated by @xmath454 } b^{(1)}_{i,1}(t ) & \leq & \sup_{{\mathbf{x}}\in[0,1]^2 } \frac{|\alpha_n({\mathbf{x}})|}{q_\omega ( x_1\wedge x_2 ) } \\ & & { } \times \biggl ( 1\vee\sup_{(y , t)\dvtx c(y^{1-t},y^t)>n^{-\alpha } } \biggl| \frac{c}{\tilde c_n } ( y^{1-t},y^t)\biggr| \biggr ) \times\psi(i),\end{aligned}\ ] ] where @xmath515 for @xmath456 . using analogous arguments as in the previous proof
we can conclude , using of ( [ opbruch2 ] ) and theorem g.1 in @xcite , that @xmath516 } b^{(1)}_{i,1}(t ) > \varepsilon/4 ) = 0 .
$ ] for the second summand , we note that @xmath454 } b^{(1)}_{i,2}(t ) & \leq&\sup_{x_1\in[0,1 ] } \frac { |\alpha_n(x_1,1)|}{q_\omega(x_1 ) } \times \biggl ( 1\vee\sup_{(y , t)\dvtx c(y^{1-t},y^t)>n^{-\alpha } } \biggl| \frac{c}{\tilde c_n } ( y^{1-t},y^t)\biggr| \biggr)\\ & & { } \times\sup_{t\in [ 0,1 ] } \int_0^{1/i } k_2(y , t ) \,dy,\end{aligned}\ ] ] where @xmath517 is defined in ( [ k2 ] ) . observing the estimate in ( [ k2 * ] ) , we easily obtain @xmath518 } \int_0^{1/i } k_2(y , t ) \,dy = 0 $ ] . again , under consideration of ( [ opbruch2 ] ) and theorem g.1 in @xcite , we have @xmath519 } b^{(1)}_{i,2}(t)\,{>}\,\varepsilon/4 ) \,{=}\,0 $ ] .
a similar argument works for @xmath520 and from the estimates for the different terms the assertion @xmath521}\bigl|b^{(1)}_i(t)\bigr| > \varepsilon\bigr ) = 0\ ] ] follows . considering the term @xmath449 } } |b^{(2)}_i(t ) |$ ] , we proceed along similar lines as in the proof of theorem [ theoh1 ] . for the sake of brevity
, we only state the important differences : in estimation ( [ mvt2 ] ) replace @xmath157 by @xmath522 , then make use of decomposition ( [ stute ] ) , calculations similar to ( [ k2 * ] ) , and theorem g.1 in @xcite again and for the estimation of the remainder note that @xmath523 .
proof of theorem [ effic ] let @xmath222 denote a probability measure minimizing the functional @xmath220 defined in ( [ asymvarall ] ) .
note that @xmath220 is convex and define for @xmath524 $ ] and a further probability measure @xmath216 on @xmath29 $ ] the function @xmath525 because @xmath220 is convex it follows that @xmath222 is optimal if and only if the directional derivative of @xmath222 in the direction @xmath526 satisfies @xmath527 for all probability measures @xmath216 . using dirac measures for @xmath216 yields that this inequality is equivalent to ( [ lowbound ] ) , which proves theorem [ effic ] . proof of theorem [ testh0 ] since the integration mapping is continuous , it suffices to establish the weak convergence @xmath528 in @xmath126 $ ] where we define @xmath529 we prove this assertion along similar lines as in the proof of theorem [ theoh1 ] . for @xmath530 , we recall the notation @xmath531 and consider the following random functions in @xmath126 $ ] : @xmath532 by an application of lemma [ lema1 ] in appendix [ seclem ]
, it suffices to show the conditions listed in ( [ cond ] ) . by arguments similar to those in the proof of theorem [ theoh1 ]
, we obtain @xmath533 in @xmath439\times[0,1])$ ] . assertion ( i )
now follows immediately by the boundedness of the functions @xmath534 and @xmath535 on @xmath536 $ ] [ see conditions ( [ cw1 ] ) , ( [ cw2 ] ) and ( [ 20a ] ) ] and the continuous mapping theorem . for the proof of assertion ( ii ) , we simply note that @xmath537 and @xmath127 are bounded on @xmath29 ^ 2 $ ] and @xmath538 and @xmath539 are bounded uniformly with respect to @xmath441 $ ] by the integrable functions @xmath540 and @xmath541 . for the proof of assertion ( iii )
, we fix some @xmath443 such that @xmath542 and consider the decomposition @xmath543 where @xmath544 @xmath545 and @xmath546 are defined in ( [ hd1 ] ) and @xmath547 by the same arguments as in the proof of theorem [ theoh1 ] , we have for every @xmath548 @xmath549 } } |i(t,1/i)|>\varepsilon\bigr ) = 0,\ ] ] and @xmath550 } |i(t,1)| = o_{{\mathbb{p}}^*}(1)$ ] , which yields the asymptotic negligibility@xmath551 } |b^{(3)}_i(t)|>\varepsilon ) = 0 $ ] . for @xmath496
, we obtain the estimate @xmath552}\bigl|b^{(1)}_i(t)\bigr|\\ & & \qquad\leq \sup_{t\in[0,1 ] } \int_{i_{b^{(1)}_i(t ) } } n| ( \tilde c_n - c ) ( y^{1-t},y^t ) |^2 \biggl| 1 \vee\frac{c^2}{\tilde c_n^2 } ( y^{1-t},y^t)\biggr| \bar w(y ) y^{-2 } \,dy \\ & & \qquad\leq \sup_{{\mathbf{x}}\in[0,1]^2}n|\tilde c_n({\mathbf{x}})-c({\mathbf{x}})|^2\times \biggl ( 1\vee\sup_{{\mathbf{x}}\in[0,1]^2 \dvtx c({\mathbf{x}})>n^{-\alpha } } \biggl| \frac { c^2}{\tilde c_n^2}({\mathbf{x } } ) \biggr| \biggr ) \times\psi(i),\end{aligned}\ ] ] where @xmath553 , which can be handled by the same arguments as in the proof of theorem [ theoh1 ] .
finally , the term @xmath554 can be estimated by @xmath555}\bigl|b^{(2)}_i(t)\bigr| \\ & & \qquad\leq \sup_{t\in[0,1 ] } \int_{i_{b^{(2)}_i(t ) } } n| ( \tilde c_n - c ) ( y^{1-t},y^t ) |^2
\biggl| 1\vee\frac{c^{\lambda } } { \tilde c_n^2 } ( y^{1-t},y^t)\biggr| \bar w(y ) y^{-\lambda } \,dy \\ & & \qquad\leq \sup_{{\mathbf{x}}\in[0,1]^2}n|\tilde c_n({\mathbf{x}})-c({\mathbf{x}})|^2\times \biggl ( 1\vee\sup_{{\mathbf{x}}\in[0,1]^2 \dvtx c({\mathbf{x}})\leq n^{-\alpha } } \biggl| \frac{c^{\lambda}}{\tilde c_n^2}({\mathbf{x } } ) \biggr| \biggr ) \times\phi(i),\end{aligned}\ ] ] where @xmath556 for @xmath456 by condition ( [ cw3 ] ) . mimicking the arguments from the proof of theorem [ theoh1 ] completes the proof .
proof of theorem [ testh1 ] recall the decomposition @xmath557 , @xmath558 where @xmath322 and @xmath323 are defined in ( [ decomps1 ] ) . with the notation
@xmath559 it follows that @xmath560 and the assumptions on @xmath42 yield the validity of ( [ cw1])([cw3 ] ) for @xmath561 .
this allows for an application of theorem [ theoh1 ] and together with the continuous mapping theorem we obtain @xmath562 , where @xmath342 is the limiting process defined in ( [ testh1 ] ) .
thus , it remains to verify the negligibility of @xmath328 .
for @xmath323 , we note that by theorem [ theopickest1 ] and the continuous mapping theorem we have @xmath563 and it remains to consider @xmath564 . to this end , we fix some @xmath443 such that @xmath565 and consider the decomposition=1 @xmath566=0 where the sets @xmath567 are defined in ( [ hd1 ] ) . on the set @xmath568 , we use the estimate @xmath569 ^ 2 \dvtx c({\mathbf{x}})>n^{-\alpha}}\frac{c(\mathbf { x})}{\tilde c_n(\mathbf{x})}\biggr),\end{aligned}\ ] ] where @xmath570 . by arguments similar to those used in the proof of theorem [ theoh1 ] , it is now easy to see that @xmath571 ^ 2}n^{\alpha + 1/2}|\tilde c_n({\mathbf{x}})-c({\mathbf{x}})|^2\\ & & { } \times\biggl ( 1\vee\sup_{{\mathbf{x}}\in [ 0,1]^2 \dvtx c({\mathbf{x}})>n^{-\alpha } } \biggl| \frac{c}{\tilde c_n}({\mathbf{x } } ) \biggr| \biggr)^2 \times k \\ & = & o_{{\mathbb{p}}^*}(1),\end{aligned}\ ] ] where @xmath572 denotes a finite constant [ see condition ( [ cw3 ] ) ] .
now set @xmath573 . from the estimate @xmath574
we obtain by similar arguments as in the proof of the negligibility of @xmath575 in the proof of theorem [ theoh1 ] ( note that on @xmath576 we have @xmath577 ) @xmath578 } } |t_2(t , n)| & \leq&\log(n ) n^{-\beta\alpha}\sup _ { { \mathbf{x}}\in[0,1]^2}\sqrt{n}|\tilde c_n({\mathbf{x}})-c({\mathbf{x}})|\\ & & { } \times\biggl ( 1\vee\sup_{{\mathbf{x}}\in[0,1]^2 \dvtx c({\mathbf{x}})\leq n^{-\alpha } } \biggl| \frac{c^{1+\beta}}{\tilde c_n}({\mathbf{x } } ) \biggr| \biggr ) \times\tilde k,\end{aligned}\ ] ] where @xmath579 denotes a finite constant [ see conditions ( [ cw3 ] ) and ( [ 20c ] ) ] and we used the estimate @xmath580 which holds for sufficiently large @xmath581 .
finally , we observe that @xmath468 ^ 2 \dvtx c({\mathbf{x}})\leq n^{-\alpha } } \biggl| \frac { c^{1+\beta}}{\tilde c_n}({\mathbf{x } } ) \biggr| \leq\sup_{{\mathbf{x}}\dvtx c({\mathbf{x}})\leq n^{-\alpha } } | n^\gamma c^{1+\beta}({\mathbf{x } } ) | \leq n^{\gamma-(1+\beta)\alpha } = o(1).\ ] ] now the proof is complete .
proof of theorem [ multitest ] the conditions on the weight function imply that all integrals in the definition of @xmath334 are proper and therefore the mapping @xmath582 is continuous .
hence , the result follows by the continuous mapping theorem for the bootstrap [ see , e.g. , theorem 10.8 in @xcite ] provided the conditional weak convergence in ( [ weakpdm ] ) holds under the nonrestrictive smoothness assumption ( [ asproc ] ) . to see this , proceed similar as in @xcite and show hadamard - differentiability of the mapping @xmath583 , which is defined for some distribution function @xmath4 on the unit square whose marginals @xmath584 and @xmath585 satisfy @xmath586
this can be done by similar arguments as in @xcite and the details are omitted for the sake of brevity .
[ lema1 ] let @xmath587 for @xmath588 be arbitrary maps with values in the metric space @xmath589 and @xmath590 be borel - measurable .
suppose that : @xmath591 then @xmath592 for @xmath144 .
let @xmath593 be closed and fix @xmath594 .
if @xmath595 denotes the @xmath596-enlargement of @xmath6 we obtain @xmath597 by hypothesis ( i ) and the portmanteau theorem [ see @xcite ] @xmath598 by conditions ( ii ) and ( iii ) @xmath599 and since @xmath600 for @xmath601 and closed @xmath6 the result follows by the portmanteau theorem .
the authors would like to thank martina stein , who typed parts of this manuscript with considerable technical expertise .
the authors would also like to thank christian genest for pointing out important references and johan segers for many fruitful discussions on the subject .
we are also grateful to two unknown referees and an associate editor for their constructive comments on an earlier version of this manuscript , which led to a substantial improvement of the paper .
|
we propose a new class of estimators for pickands dependence function which is based on the concept of minimum distance estimation .
an explicit integral representation of the function @xmath0 , which minimizes a weighted @xmath1-distance between the logarithm of the copula @xmath2 and functions of the form @xmath3 is derived .
if the unknown copula is an extreme - value copula , the function @xmath0 coincides with pickands dependence function .
moreover , even if this is not the case , the function @xmath0 always satisfies the boundary conditions of a pickands dependence function .
the estimators are obtained by replacing the unknown copula by its empirical counterpart and weak convergence of the corresponding process is shown .
a comparison with the commonly used estimators is performed from a theoretical point of view and by means of a simulation study .
our asymptotic and numerical results indicate that some of the new estimators outperform the estimators , which were recently proposed by genest and segers [ _ ann .
statist . _
* 37 * ( 2009 ) 29903022 ] . as a by - product of our results
, we obtain a simple test for the hypothesis of an extreme - value copula , which is consistent against all positive quadrant dependent alternatives satisfying weak differentiability assumptions of first . , and .
|
excess infrared ( ir ) emission from t tauri stars is thought to originate from circumstellar disks ( mendoza 1968 ; shu , adams , & lizano 1987 ; and references therein ) .
passive disks are the simplest to consider . by definition ,
they lack intrinsic luminosity and reradiate the energy they absorb from the central star .
hydrostatic , radiative equilibrium models for passive t tauri disks are derived by chiang & goldreich ( 1997 , hereafter cg ) .
the disk surface flares outward with increasing radius and intercepts more stellar radiation , especially at large distances from the central star , than a flat disk would ( kenyon & hartmann 1987 ) .
an optically thin layer of superheated dust grains ensheaths the entire disk .
dust grains in this surface layer are directly exposed to central starlight and reradiate to space about half the stellar energy they absorb .
the other half is emitted towards the midplane and regulates the temperature of the cooler disk interior .
vertical temperature gradients in externally illuminated disk atmospheres are calculated in detail by calvet et al .
( 1991 ) , malbet & bertout ( 1991 ) , and dalessio et al .
( 1998 ) .
cg compute spectral energy distributions ( seds ) of passive disks viewed face - on .
the calculated sed is fairly constant over the thermal ir , in accord with the observed flattish excesses of t tauri stars such as gm aur .
spectral features from dust grains in the superheated layer appear in emission when the disk is viewed face - on , similar to solid - state emission lines evinced by t tauri stars and herbig ae / be stars ( cohen & witteborn 1985 ; waelkens et al . 1996 ; waters & waelkens 1998 ) . what is the spectrum of a passive disk whose midplane is inclined at an arbitrary angle , @xmath5 , to the plane of the sky ? in [ contsed ] , model assumptions and results pertaining to the continuum sed are set forth . in [ specdust ] ,
we study how spectral signatures of some dust grain resonances change from emission to absorption as the disk is viewed increasingly edge - on .
applications to observations are contained in [ appobs ] .
there , we assess the possibility that the relatively low 12@xmath1 flux of gm aur might be caused by a non - zero disk inclination , as opposed to a central au - sized gap
. we close by considering whether differences between class i and class ii seds might reflect differences in viewing angle rather than evolutionary status . ,
i.e. , @xmath6 where @xmath7 . for class
ii sources , @xmath8 . ]
we consider a passive disk in radiative and hydrostatic balance around a t tauri star .
the model employed is identical to that derived in 2.1 and 2.3 of cg .
symbols and values of free parameters are listed here in table 1 , and henceforth are used without explanation .
we measure position by cylindrical radius , @xmath9 , and vertical distance above the disk midplane , @xmath10 .. ] cll @xmath11 & stellar mass & @xmath12 @xmath13 & stellar radius & @xmath14 @xmath15 & stellar effective temperature & @xmath16 @xmath17 & disk surface density & @xmath18 @xmath19 & inner disk radius & @xmath20 @xmath21 & outer disk radius & @xmath22 @xmath23 & dust opacity at visual wavelengths & @xmath24 @xmath25 & grain emissivity & @xmath26 @xmath27 & grain radius & @xmath28 @xmath29 & grain emissivity index & 1 derived disk properties relevant to our present investigation are summarized as follows . at each radius
we distinguish two regions : the superthermal surface layer which is directly exposed to light from the central star , and the cooler , diffusively heated interior which the surface encases . variables evaluated in the former region are denoted by a subscript @xmath30 , while those in the latter region carry a subscript @xmath31 .
pertinent results from cg include the dust temperature at the surface @xmath32 and the gas density in the interior @xmath33 both the gas scale height , @xmath34 , and interior temperature , @xmath35 , take on different forms depending on the vertical optical depth of the interior .
inside 84@xmath36 , the interior is opaque to its own reprocessed radiation , and @xmath37 @xmath38 between 84 and 209 @xmath36 , the interior is optically thin to its own radiation but still thick to radiation from the surface ; here @xmath39 @xmath40 finally , in the outermost regions of the disk , the encased material is transparent to radiation from the surface , and @xmath41 @xmath42 boundary conditions are as follows .
the superheated layer is located at @xmath43 .
its visual optical depth normal to its flared surface is equal to the angle @xmath44 at which stellar rays penetrate the disk ( refer to figure 3 of cg ) : @xmath45 radially , the disk extends from the silicate condensation boundary at @xmath19 , to an outer radius , @xmath21 , at which @xmath46 . the occulting angle , @xmath47 is a natural angle with which to compare the viewing angle @xmath5 . for @xmath48 , both the central star and
most of the disk surface are in direct view . for @xmath49 , star and inner disk
are blocked from view by the flared outer `` wall '' . for clarity ,
we first consider only the disk interior .
figure [ incschem ] depicts the viewing geometry and coordinate system .
define @xmath51 and @xmath52 to be orthogonal spatial axes in the plane of the sky , centered on the star .
the sed is the integral of specific intensity over projected disk area , viz .
, @xmath53 where @xmath54 traces the disk boundary on the sky , @xmath55 measures distance to the source from earth , and @xmath56 is the specific intensity ;
@xmath57 with @xmath58 here @xmath59 is the planck function , and @xmath60 , @xmath61 both measure line - of - sight distance from the observer . since temperature , density , and opacity
are specified by our model , calculating the sed at arbitrary inclination is primarily an exercise in multidimensional integration .
we perform these integrals numerically using standard techniques ( press et al .
optical depths are evaluated using a romberg integrator with a mandated fractional accuracy of @xmath62 .
specific intensities are computed on an adaptive stepsize grid , where the source function ( @xmath59 ) changes by no more than 20% between adjacent grid points .
integrals over @xmath52 are performed using either a romberg integrator or , in cases where the integrand is very sharply peaked and the romberg integrator takes prohibitively many steps , a 60-point gaussian quadrature routine .
the final integration over @xmath51 employs the trapezoidal rule on a 60-point logarithmic grid .
answers are routinely checked for robustness against changes in stepsize .
we verify that the sed for @xmath63 computed using the multi - dimensional integration code matches the sed for face - on disks computed using the 1-dimensional integrator of cg .
figure [ intersed ] displays a family of interior seds for different @xmath5 .
first consider seds for angles @xmath64 .
radiation at wavelengths shorter than @xmath0100 @xmath1 comes mainly from optically thick regions ; consequently , @xmath65 is approximately proportional to the areal projection factor @xmath66 . at longer wavelengths ,
the radiation is emitted by increasingly transparent material , so that @xmath65 tends to be independent of orientation . for angles
@xmath67 , short wavelength emission from the inner disk is strongly extincted by the intervening outer disk . however , millimeter ( mm ) wavelength fluxes only drop by a factor of 1.8 as the disk is viewed increasingly edge - on , since about half of the outer disk remains visible in that limit .
next we compute the total sed .
the superheated layer is treated as a plane - parallel atmosphere having visual optical depth @xmath44 normal to the surface . whenever a line - of - sight intersects the surface , we increment the specific intensity from the interior by , where @xmath68 is the geometrical thickness of the surface , and @xmath69 is the radius of curvature of @xmath70 . since this fraction is less than @xmath71 , for computational simplicity we let the intersected optical depth of the surface be @xmath72 . ]
@xmath73 \ , \exp(-\tau_{\nu , l } ) \ , .\ ] ] here , @xmath74 and @xmath75 are unit vectors normal to the surface and parallel to the line - of - sight , respectively , @xmath76 is the planck - averaged dust emissivity at the surface , and @xmath77 is the intervening optical depth between disk surface and observer .
figure [ totsed ] displays a family of total disk seds labeled by @xmath5 .
the behavior of the sed for @xmath78 is similar to that described in [ intsed ] , since the interior dominates emission at those wavelengths .
shortward of @xmath0@xmath79 , emission from the superheated surface makes a qualitative difference to the appearance of the seds .
one difference is that for angles @xmath80 , the flux is remarkably insensitive to viewing geometry . , both because surface layers intermediate in radius are seen through the outer disk s atmosphere and because emission from the optically thick interior is proportional to the projected area . ]
this is because shorter wavelength radiation emerges primarily from the superheated surface layers , which are optically thin along most lines of sight . for angles
@xmath81 , the intervening outer disk dramatically extinguishes surface radiation at the shortest wavelengths .
however , the total disk sed between 30 and 80@xmath1 is relatively robust to changes in inclination ; it decreases by less than a factor of 3 between face - on and edge - on orientations .
radiation at these wavelengths originates from surface layers at large radii and is only slightly obscured by the tenuous interior at still larger radii .
finally , in figure [ starsed ] , we add the contribution from the central star to the sed .
the stellar flux fades rapidly with increasing @xmath5 beyond @xmath0@xmath82 .
when viewed nearly face - on , passive disks exhibit emission lines associated with dust grain resonances .
these features arise from material along the line - of - sight which is optically thin : line - to - continuum flux ratios are of the same order as the percentage enhancement in line opacity
. emission lines may originate from both the optically thin superheated surface ( as discussed in cg ) and the optically thin interior .
how do such lines vary as the disk is viewed increasingly edge - on ?
might they appear in absorption instead , as cohen & witteborn ( 1985 ) find for 7 of the 32 t tauri stars they survey spectrophotometrically ? to address these questions , we add 6 gaussian - shaped spectral resonances to our grain emissivity law . in choosing resonant wavelengths and strengths , we aim to illustrate the range of effects which occur with variable viewing angle .
roughly , our 6 lines can be associated with crystalline h@xmath83o ice at 60 and 100@xmath1 ( waters & waelkens 1998 ) , amorphous silicates at 10 and 20@xmath1 ( mathis 1990 ) , and small carbon - rich grains or pahs ( polycyclic aromatic hydrocarbons ) at 3.3 and 6.2@xmath1 ( waelkens et al .
1996 ) .
solid - state line spectra at various disk inclinations are displayed in figure [ specline ] .
emission at all wavelengths varies little with viewing geometry for angles @xmath48 .
this is true particularly at the shortest wavelengths where lines arise largely from the optically thin surface layers . at wavelengths progressively longer than @xmath030@xmath1 , the optically thin interior at large radius contributes increasingly to line emission .
the drop in the long wavelength continuum with increasing inclination reflects the @xmath84 dependence of radiation emitted by the optically thick interior .
emission lines at @xmath060 and @xmath85 persist at angles @xmath49 ,
although their fluxes are lower by a factor @xmath02 for edge - on as compared to face - on disks . for edge - on orientations , only about half of the optically thin interior remains visible .
emission around @xmath86 is more robust to changes in inclination ; most of this radiation emerges from the superheated surface at large radii and passes through the rarefied disk interior at yet larger radii with little attenuation . at the shortest wavelengths , where occultation of the inner disk is significant ,
lines indeed appear in absorption , though only over a limited range of inclinations between @xmath87 and 90@xmath3 . for a dust grain resonance to appear in absorption , disk inclinations must be sufficiently high that line emission from the superheated surface is extinguished
. however , @xmath5 can not be so high that line and continuum optical depths both exceed unity at large radius where the disk interior is nearly isothermal . only over an intermediate range of inclinations
does radiation on and off the resonant wavelength probe a variety of interior disk temperatures . for the @xmath88 resonance to appear in absorption , @xmath89 . at @xmath90 , opacities are lower , and inclinations needed for absorption higher : @xmath91 . in principle , near - infrared line spectra may provide a sensitive diagnostic of disk inclination . while the trends outlined above apply qualitatively to all passive disks , quantitative conclusions are model - dependent .
for example , in our standard model , no absorption line at @xmath92 appears because line - of - sight optical depths are too small to extinguish emission from the superthermal surface .
however , such an absorption line would appear in the sed of a sufficiently massive , inclined disk .
gm aur is a particularly clean , well - studied t tauri system to which we can apply our sed models .
the single central star has mass @xmath93 , luminosity @xmath94 , and age @xmath95 ( beckwith et al .
aperture synthesis maps in @xmath96 evince a rotating circumstellar gas disk having approximate projected dimensions 950 @xmath97 @xmath98 @xmath99 , inclined at @xmath100 ( koerner , sargent , & beckwith 1993 ) .
hubble space telescope ( hst ) images in scattered visible light reveal the disk surface to be flared , and suggest @xmath31 is closer to @xmath0@xmath4 ( koerner 1997 ; stapelfeldt et al .
these same observations detect no outflow or remnant envelope surrounding this relatively evolved system .
cg model the sed of gm aur , but with a face - on disk .
they suggest that a non - zero disk inclination might cause line - of - sight radiation from the inner disk to be absorbed by the outer disk , thereby explaining the relatively low @xmath101 flux without invoking a central hole . here
we investigate this possibility by refitting the sed with an inclined disk .
figure [ gmsed ] displays our new fit with a disk inclined at @xmath102 . despite the non - zero inclination ,
none of our fit parameters changes significantly from those given by cg for the face - on case .
fluxes at iras ( infrared astronomical satellite ) wavelengths emerge mainly from the optically thin surface layers , and those at mm wavelengths emerge mostly from the optically thin interior ; neither is sensitive to viewing angles @xmath5 near or less than @xmath87
. for this disk equals @xmath103 . ]
our model reproduces the observed modest extinction to the central star of @xmath104 magnitudes , but at the cost of still requiring a large central gap having a radius @xmath0@xmath105 times greater than the dust sublimation radius : @xmath106 , where the uncertainty reflects that of the @xmath101 iras point .
there is no inclination for which the outer edge of our model disk attenuates the @xmath101 flux without also extinguishing central starlight by several magnitudes .
line - of - sight column densities to surface regions inside @xmath05 au ( where most of the @xmath101 emission originates ) are only @xmath0@xmath107 greater than the column density to the star .
this small difference in the amount of obscuring material is completely insignificant ; the opacity at @xmath101 is @xmath108 times lower than at visible wavelengths .
thus , the low visual extinction to the central star implies that we should have a clear view of disk regions inside a few au at near - infrared wavelengths . in spite of these considerations ,
we remain skeptical of the existence of such a large central gap devoid of dust .
a more palatable alternative might be that the inner disk s aspect ratio does not increase monotonically with radius .
such undulations in the height of the surface would shadow annular regions from the central star and lower their temperatures .
in addition , those ripples nearest the observer would hide their warmer , starlit sides when viewed at non - zero inclination .
accretional heating may be responsible for such changes in surface geometry , as mentioned in cg ( see their 3.5 ) ; as estimated there , enhancements in disk thickness due to heating of the midplane may become significant inside a few @xmath99 . for an accretion disk which derives its luminosity solely from local viscous dissipation , the aspect ratio indeed decreases with radius whenever opacities increase steeply with temperature , as demonstrated by bell et al .
( 1997 ) .
might some class i spectra represent t tauri stars obscured by inclined disks ?
the possibility should be entertained for sources such as 04108 + 2803b ; figure [ class1sed ] demonstrates that a passive disk inclined at @xmath109 provides a reasonable fit ( to within factors of 2 ) to this class i sed .
fit parameters are similar to those of our standard model , and are listed in the figure caption . for this particular model ,
near - ir emission at j , h , and k is interpreted as central starlight extincted by the disk s outer edge ; inclusion of starlight scattered off the disk surface would imply a larger inclination . to highlight the contribution from the superheated surface
, we also plot the sed with the surface emission removed .
fluxes between 3 and 60@xmath1 arise primarily from the superthermal surface , seen both through and over the lip of the outer disk atmosphere ( refer to figure [ incschem ] ) .
clearly , a class i sed does not imply a unique distribution of circumstellar material .
high angular resolution images provide additional clues .
many class i sources exhibit near - ir nebulosity on scales ranging from 1500 to 3000 au ( kenyon et al .
1993 ; and references therein ) ; for these sources , an additional , non - disk component of dust is needed .
kenyon , calvet , & hartmann ( 1993 ) model class i seds using rotationally flattened , infalling envelopes that are passively heated by central stars .
these envelopes extend out to larger radii than do our disks , typically 3000 au , but contain roughly the same amount of mass , about 0.1@xmath2 .
bipolar holes empty of material , presumably evacuated by outflows , are invoked so that central starlight may scatter off cavity walls towards earth , thereby explaining the observed near - ir fluxes . in imaging observations of three class i sources using hst / nicmos ( near infrared camera and multi - object spectrometer ) ,
padgett et al . ( 1998 ) discover nearly edge - on , flared circumstellar disks having diameters 300750 au .
near - ir emission is observed to be scattered not only off upper and lower disk surfaces , but also off `` dusty material within or on the walls of outflow cavities . '' in general , a combination of an inclined , passively heated disk , and a dusty bipolar outflow or partially evacuated envelope may best describe class i sources . since 04108
+ 2803b reveals no optical or near - ir emission beyond @xmath0140 au from its central star ( kenyon , calvet , & hartmann 1993 ) , it is consistent with being a limiting example of a simple inclined disk .
two other examples for which this limiting case scenario may also apply include the embedded sources 04295 + 2251 and 04489 + 3042 .
both sources exhibit nearly flat excesses between @xmath03 and 100 @xmath1 which may arise from flared circumstellar disks .
neither source betrays extended near - ir emission on scales greater than a few hundred au , or high - velocity molecular gas from an outflow ( kenyon et al .
1993 ) .
near - ir polarimetry also addresses the possible presence of envelopes .
whitney , kenyon , & gmez ( 1997 ) determine that linear polarizations greater than @xmath020% at j , h , and k characterize the largest reflection nebulae associated with class i sources .
such large polarizations are interpreted as arising from the scattering of central starlight off cavity walls in envelopes .
in contrast , 04108 + 2803b evinces fractional polarizations of 5.1% and 1.6% at h and k , respectively ( whitney , kenyon , & gmez 1997)levels more comparable to those in class ii sources than in truly younger , more embedded protostars , and which may arise from starlight scattered off the disk surface .
resolved polarimetry maps of 04108 + 2803b can test our hypothesis .
degeneracies inherent in models deduced from a sed may also be broken by kinematic studies .
interferometric radial velocity maps in molecular lines may distinguish between an infalling envelope ( e.g. , hayashi , ohashi , & miyama 1993 ; ohashi et al .
1997 ) , and a rotating disk ( e.g. , koerner & sargent 1995 ) .
finally , we note that the inverse problem to 04108 + 2803b is presented by the star hk tau b : a system observed to possess an edge - on ( @xmath110 ) circumstellar disk and no observable envelope , but whose infrared sed is not well measured because of confusion from an infrared - bright companion ( stapelfeldt et al .
1998 ; koresko 1998 ) .
we await sirtf ( space infrared telescope facility ) which can provide both the sed between 3 and 180 microns and also images of the superheated disk surface in thermal emission .
financial support for this research was provided by nsf grant 94 - 14232 , nasa grant nag5 - 7008 , and an nsf graduate fellowship .
we thank an anonymous referee for a careful reading of our paper and for providing suggestions which helped to improve its presentation .
beckwith , s. v. w. , sargent , a. i. , chini , r. s. , & gusten , r. 1990 , , 99 , 924 bell , k. r. , cassen , p. m. , klahr , h. h. , & henning , th .
1997 , , 486 , 372 calvet , n. , patio , a. , magris , g. , & dalessio , p. 1991
, , 380 , 617 chiang , e. i. , & goldreich , p. 1997
, , 490 , 368 ( cg ) cohen , m. , & witteborn , f. c. 1985 , , 294 , 345 dalessio , p. , cant , j. , calvet , n. , & lizano , s. 1998 , , 500 , 411 hayashi , m. , ohashi , n. , & miyama , s. m. 1993 , , 418 , l71 kenyon , s. j. , calvet , n. , & hartmann , l. 1993 , , 414 , 676 kenyon , s. j. , & hartmann , l. 1987 , , 323 , 714 kenyon , s. j. , whitney , b. a. , gomez , m. , & hartmann , l. 1993 , , 414 , 773 koerner , d. w. 1997 , origins of life and evolution of the biosphere , 27 , 157 koerner , d. w. , & sargent , a. i. 1995 , , 223 , 169 koerner , d. w. , sargent , a. i. , & beckwith , s. v. w. 1993 , icarus , 106 , 2 koresko , c. d. 1998 , , 507 , l145 lada , c. j. 1987 , in iau symp .
115 , star forming regions , ed .
m. peimbert & j. jugaku ( dordrecht : reidel ) , 1 lada , c. j. , & wilking , b. a. 1984 , , 287 , 610 malbet , f. , & bertout , c. 1991 , , 383 , 814 mathis , j. m. 1990 , , 28 , 37 mendoza , e. e. v. 1968 , , 151 , 977 ohashi , n. , hayashi , m. , ho , p.t.p . , & momose , m. 1997 , , 475 , 211 padgett , d. , brandner , w. , stapelfeldt , k. , koerner , d. , terebey , s. , strom , s. , & kenyon , s. 1998 , american astronomical society meeting , 192 , # 10.04 press , w. h. , teukolsky , s. a. , vetterling , w. t. , & flannery , b. p. 1992
, numerical recipes in fortran ( cambridge university press ) shu , f. h. , adams , f. c. , & lizano , s. 1987 , , 25 , 23 stapelfeldt , k. , burrows , c. j. , koerner , d. , krist , j. , watson , a. m. , trauger , j. t. , & wfpc2 idt 1995 , american astronomical society meeting , 187 , # 113.04 stapelfeldt , k. r. , krist , j. e. , mnard , f. , bouvier , j. , padgett , d. l. , & burrows , c. j. 1998 , , 502 , l65 waelkens , c. , et al .
1996 , , 315l , 245 waters , l.b.f.m . , & waelkens , c. 1998 , , in press .
whitney , b. a. , kenyon , s. j. , & gmez , m. 1997 , , 485 , 703
|
we compute spectral energy distributions ( seds ) for passive t tauri disks viewed at arbitrary inclinations .
semi - analytic models of disks in radiative and hydrostatic equilibrium are employed . over viewing angles for which the flared disk does not occult the central star , the sed varies negligibly with inclination . for such aspects ,
the sed shortward of @xmath080@xmath1 is particularly insensitive to orientation , since short wavelength disk emission is dominated by superheated surface layers which are optically thin .
the sed of a nearly edge - on disk is that of a class i source .
the outer disk occults inner disk regions , and emission shortward of @xmath030@xmath1 is dramatically extinguished .
spectral features from dust grains may appear in absorption .
however , millimeter wavelength fluxes decrease by at most a factor of 2 from face - on to edge - on orientations .
we present illustrative applications of our sed models . the class i source 04108 + 2803b is considered a t tauri star hidden from view by an inclined circumstellar disk
. fits to its observed sed yield model - dependent values for the disk mass of @xmath00.015 @xmath2 and a disk inclination of @xmath065@xmath3 relative to face - on .
the class ii source gm aur represents a t tauri star unobscured by its circumstellar disk .
fitted parameters include a disk mass of @xmath00.050 @xmath2 and an inclination of @xmath0@xmath4 .
|
the _ spitzer _ space telescope opens a unique mid- and far - infrared window to the exploration of massive star formation not only in our galaxy but also in our neighboring dwarf galaxies , the large and small magellanic cloud ( lmc , smc ) .
stellar associations in both the magellanic clouds ( mcs ) contain the richest sample of young bright stars in these galaxies , and consequently our knowledge on their young massive stars has been collected from studies of such stellar systems ( see e.g. massey 2002 ) .
furthermore , almost every mcs association coincide with one or more h ii regions , as they have been earlier cataloged ( henize 1956 ; davies et al .
1976 ) , and it has been shown that the measured h fluxes are in excellent agreement with those expected from the ionizing flux of the detected massive stars ( massey 1993 ) .
however , mcs associations are not mere aggregates of young bright stars alone , but they also host large numbers of faint pre - main sequence ( pms ) stars , as recent hst studies showed for both the lmc ( gouliermis et al .
2006 ) and the smc ( nota et al .
both , the existence of h ii regions and pms stars in stellar associations of the mcs indicate that star formation may be still active in their vicinity , and _ spitzer
_ offers a unique opportunity to test this hypothesis . indeed , using _ spitzer _
observations , jones et al .
( 2005 ) reported four young stellar objects ( ysos ) in the h ii complex n 159 ( located in the vicinity of the association lh 105 ; lucke & hodge 1970 ) in the lmc .
chu et al .
( 2005 ) obtained _ spitzer _ observations of the super - bubble n 51d ( associations lh 51/54 ) in the lmc and they found three ysos projected within the super - bubble .
images from the _ hubble _ space telescope ( hst ) in h and [ s ii ] allowed them to identify dust globules associated with two of them .
the third yso was associated with the first herbig - haro object detected outside the galaxy .
furthermore , meixner et al . ( 2006 ) in their presentation of the first results from the _ spitzer _ space telescope survey of the lmc legacy project _ surveying the agents of a galaxy s evolution _ ( sage ) , show as a test - case epoch 1 data of the area of two h ii regions , n 79/n 83 ( associations lh 2/lh 5 ) .
they found that mips 70 and 160 images of the diffuse dust emission of the regions reveal a distribution similar to the gas emissions , especially the h i 21 cm emission , and they note that the measured point - source sensitivity for the epoch 1 data is consistent with expectations for the survey .
bolatto et al .
( 2007 ) , recently , presented their results from the _ spitzer _ survey of the small magellanic cloud ( s@xmath1mc ) , which imaged the star - forming body of the smc in all seven mips and irac wave bands .
these authors compiled a photometric catalog of 400,000 mid- and far - infrared point sources in the smc , from which they identified 282 bright candidate ysos as bright 5.8 sources with very faint optical counterparts and very red mid - infrared colors ( [ 5.8]@xmath2[8.0 ] @xmath3 1.2 mag ) .
more recently , simon et al .
( 2007 ) used the same observations and applied spectral energy distribution ( sed ) fits based on models by whitney et al .
( 2003a , b , 2004 ) to enhance the catalog of candidate ysos in the brightest h ii region in the smc , n 66 . the present paper deals with mid - infrared imaging from the _ spitzer _ space telescope of the h ii region lha 115-n90 ( henize 1956 ) or in short n 90 , located in the wing of the smc .
this region , also cataloged by davies et al .
( 1976 ) and named dem s 166 , is related to the young star cluster ngc 602 , which was classified as a stellar association by hodge ( 1985 ; system no 68 in his catalog ) .
we present the results of our investigation on the candidate ysos in the region of ngc 602/n 90 , identified with our photometry according to criteria based on galactic templates and recent _ spitzer _ observations of other star forming regions in the smc .
specifically , in 2 we discuss the retrieval of the _ spitzer _ data used in our study and their photometry .
general description of the mid - infrared characteristics of the region is given in 3 , and in 4 the point sources detected with our photometry , and the selection of candidate ysos are presented .
the recent release of high - resolution data on ngc 602/n 90 obtained with the _ advanced camera for surveys _ on _ hubble _ space telescope allowed us a detailed photometric study of the stellar association and its surroundings down to the faint red pre - main sequence stars of the system ( schmalzl et al .
2007 ) . in
4 we also combine the results from both _ hubble _ and _ spitzer _ photometries to identify the optical counterparts of the candidate ysos found in the region . based on our findings
we discuss the current star forming process that probably took place in the region of ngc 602/n 90 with emphasis to the clustered behavior of star formation in 5 .
finally , general conclusions of this study are given in 6 .
the imaging data of our study have been obtained with the _ infrared array camera _ ( irac ; fazio et al .
2004 ) on - board _ spitzer
_ space telescope ( werner et al . 2004 ) .
the images are publicly available and are downloaded from the _ spitzer _ space telescope archive using the leopard package .
these observations ( dataset archive i d : aor 12485120 ) were obtained within the gto science program 125 ( pi : g. fazio ) on 28 november 2004 .
we retrieved the post basic calibrated data ( pbcd ) for all four irac channels ( 3.6 , 4.5 , 5.8 and 8.0 ) , which are already fully co - added and calibrated .
after visual inspection of the images we carried out psf - photometry with the daophot package provided within the iraf environment . in the case of the region of ngc 602/n 90 , psf photometry is required as the field is too crowded for aperture photometry to be appropriately applied . consequently , for each filter we constructed a reference psf by combining the psf of at least 5 objects with a high signal to noise ratio and we used this psf to fit all objects identified with a 3-@xmath4 confidence level over the local background .
we converted the pixel values in each filter from mjysr@xmath5 to dns@xmath5 according to `` irac data handbook '' version 5.1.1 and we computed the corresponding magnitudes as @xmath6 with @xmath7 denoting the flux measured in dns@xmath5 and @xmath8 being the zero point for each filter .
zero points were taken from hartmann et al .
( 2005 ) : 19.66 ( 3.6 ) , 18.94 ( 4.5 ) , 16.88 ( 5.8 ) and 17.39 ( 8.0 ) .
for the initial psf fit we used a psf size of 2 pixel , and applied aperture corrections as described in the `` irac data handbook '' to obtain the final magnitudes . finally , we cross - matched the obtained catalogs for each filter to identify the sources that are detected in more than one irac bands , by allowing a maximum offset of 2 pixels for the center of each point source .
_ spitzer _ covers a suitable wavelength range ( 3.6 - 160 ) for defining the dust properties of the interstellar medium ( ism ) .
the mid - infrared channels of irac ( 3.6 , 4.5 , 5.8 and 8 ) are good tracers of the smallest grains through polycyclic aromatic hydrocarbon ( pah ) emission .
very small grains ( typical sizes 12 - 150 ) and pahs ( typical sizes 4 - 12 ; desert et al .
1990 ) are very efficient in heating the gas through the photoelectric effect ( bakes & tielens 1994 ) and therefore their analysis is very important for understanding the thermodynamics of the ism .
iras data on both lmc and smc suggest a deficit of the smallest dust grains in the low - metallicity environments of these galaxies due to the lower 12 emission in comparison to the milky way and other higher metallicity galaxies ( schwering 1989 ; sauvage et al .
a study in the smc by stanimirovi et al .
( 2000 ) concluded that the dust in the smc is dominated by large grains .
the lower abundance of pahs and very small grains in the smc is also supported by detailed modeling of smc uv absorption and fir emission data by weingartner & draine ( 2001 ) , li & draine ( 2002 ) , and clayton et al .
a different grain size distribution with respect to the galaxy , or grain destruction , possibly due to the intense uv radiation and/or by supernova shocks have been proposed to explain the very small grain paucity in both galaxies .
the absence of pahs and very small grains has a profound influence on the gas heating and the existence of cold and warm phases in the interstellar medium ( wolfire et al .
1995 ) .
the images of ngc 602/n 90 taken with _ spitzer _ , which we present here , can be used for a qualitative analysis on the dust properties and its relation to stellar sources of radiation in the region of ngc 602 . a color composite image from irac channels 1 , 2 and 4
is shown in fig .
[ irac - maps ] . the 3.6 and 4.5 wave - bands are mostly sensitive to stellar photospheres and very hot circumstellar dust , and therefore the blue ( 3.6 ) and green colors ( 4.5 ) in fig .
[ irac - maps ] reveal the stars , whereas the red color ( 8 ) shows mainly the diffuse dust emission .
specifically , the origin of diffuse emission at 3.6 is likely to originate partly from bound - free transitions in the ionized gas with some contribution from the 3.29 emission near molecular clouds ( thought to be due to c@xmath2h bond stretching in pahs ) and very small dust grain emission ( e.g. engelbracht et al .
the 4.5 diffuse emission is a combination of bound - free continuum and brackett-@xmath9 recombination emission in h ii regions .
both the 5.8 and 8.0 bands show molecular material in the interstellar medium ( ism ) and circumstellar envelopes , as well as increasingly faint stellar photospheres .
the origin of the 5.8 diffuse emission is most likely very warm dust emission ( @xmath10 k ) .
it encompass the 5.6 and 6.2 pah band emission .
the origin of the 8 diffuse emission is most likely the very bright emission complex at 7 - 9 thought to be dominated by c@xmath2c stretching modes ( with some contribution from in - plane c@xmath2h bending ) of the bonds in pahs and very small carbonaceous dust grains ( jger et al .
2006 ) .
the postscript file is removed due to size limitations .
+ to see the figure download the corresponding jpg file : `` f4.jpg ''
our photometry revealed 57 objects detected in all four irac channels .
44 sources were identified in channels 1 , 2 and 3 , but not in channel 4 , while only two in channels 2,3 and 4 and not in channel 1 . for the classification of the sources we make use of the [ 3.6]@xmath2[8.0 ] , [ 8.0 ] color - magnitude diagram ( cmd ) in comparison with templates and models compiled for a wide range of ir sources in the milky way .
these templates have been developed by cohen ( 1993 ) for iras data on galactic ir sources , adapted for the irac and 2mass color classification scheme for the galactic legacy infrared mid - plane survey extraordinaire ( glimpse ) _ spitzer _ project ( benjamin et al .
we also make use of the yso models that have been developed by whitney et al .
( 2004 ) also for glimpse .
both papers include several types of ir sources such as main - sequence stars , red giants , o - rich and c - rich agb stars , h ii regions , and ysos . in fig .
[ iractempl ] ( _ left _ )
, we show the position of such sources in the [ 3.6]@xmath2[8.0 ] , [ 8.0 ] cmd according to the templates and models .
we also show the corresponding cmd for the point - sources we found in all four irac channels in the region of ngc 602/n 90 ( fig .
[ iractempl ] , _ right _ ) . from the comparison of the cmds of fig .
[ iractempl ] it is shown that there are 27 objects in our photometry with positions in the [ 3.6]@xmath2[8.0 ] , [ 8.0 ] cmd similar to galactic ysos of class 0 , i and ii .
however , class 0 ysos are very rare due to their short evolutionary phase , and such objects would be extremely faint in the irac bands at the distance of the smc .
consequently , one should consider that , although few of our sources match the positions of class 0 ysos in the cmd according to the models , probably there are no ysos of this class in our sample . a criterion adopted by bolatto et al .
( 2007 ) within the s@xmath1mc survey ( see also 1 ) for the selection of ysos in their sample is based on very red [ 5.8]@xmath2[8.0 ] color index ( @xmath3 1.2 mag ) of bright 5.8 sources .
support for their selection is lent by the location of these sources in the [ 3.6 ] @xmath2 [ 8.0 ] , [ 8.0 ] cmd , and the color - color diagrams in near- , mid- and far - infrared wavelengths as modeled by whitney et al .
( 2003 ) for different objects .
bolatto et al .
note that the highly reddened 8.0 sources occupy the region in those plots that is predicted for late class 0 to class ii ysos with different geometrical parameters .
we further select the most probable ysos in the region of ngc 602/n 90 by applying this criterion to the sources detected with our photometry .
the [ 5.8]@xmath2[8.0 ] , [ 3.6]@xmath2[4.5 ] color - color diagram of our sources is shown in fig .
[ fig - ccd ] with the limit of bolatto et al .
( 2007 ) for candidate ysos overplotted as a vertical dashed line .
there are 24 sources in our photometry that have colors [ 5.8]@xmath2[8.0 ] @xmath3 1.2 mag , meeting this selection criterion .
all , except two of them , have been selected as candidate ysos also according to their positions in the [ 3.6]@xmath2[8.0 ] , [ 8.0 ] cmd ( fig .
[ iractempl ] ) .
consequently , we refine the number of candidate ysos in the region of ngc 602/n 90 to 22 sources , which meet both criteria . the classification scheme of bolatto et al .
predicts only @xmath11 2% contamination of this part of the mid - infrared color - color diagram by b stars with 24 emission ( their figure 12 ) .
other types of evolved sources , such as carbon stars , rsgs and o - rich agb stars , are located to bluer colors . in addition , bolatto et al .
estimated the contamination by background , unresolved galaxies , which can be seen throughout the smc at infrared wavelengths .
they expect 20 contaminants in their sample of 282 candidate ysos ( @xmath12 7% ) , which translates to 1.5 sources in our sample .
however , detailed inspection of the candidate ysos on the acs image of ngc 602/n 90 revealed no suspicious background galaxies in our sample .
the characteristics of the candidate ysos revealed with our photometry in the region of ngc 602/n 90 are presented in table [ tab - yso ] , where we give the i d number ( from our irac photometric catalog ) , the position and the magnitudes and corresponding photometric uncertainties in the four irac channels for each candidate yso .
it should be noted that the classification of candidate ysos can not , in any case , be reduced to a single criterion , and therefore the technique described above should be considered a first - order selection .
moreover , taking into account that the selection of objects with [ 5.8]@xmath2[8.0 ] @xmath3 1.2 mag may compromise the actual number of ysos , this limit should be considered a tentative one . indeed , allen et al .
( 2004 ) plot known stars with yso classifications in the irac color - color diagram ( their fig .
4 ) and their class i ysos clearly stretch to red [ 3.6]@xmath2[4.5 ] colors with [ 5.8]@xmath2[8.0 ] colors less than 1.2 mag . our selection , based on bolatto et al . ( 2007 ) scheme , apparently causes such sources to be left out , but its use is being considered in order to reduce as much as possible the number of misidentifications .
furthermore , sources with colors [ 3.6]@xmath2[4.5 ] @xmath13 0.0 mag and [ 5.8]@xmath2[8.0 ] @xmath3 1.2 mag are not well represented in whitney et al .
( 2003 , 2004 ) nor in allen et al .
based on their colors , some of them are suspected to be small h ii regions , but there are no high resolution radio observations of n 90 ( unlike n 159 in the lmc ; jones et al .
2005 ) , which would provide information on the ionizing flux for these sources and help us clarify their nature .
the postscript file is removed due to size limitations .
+ to see the figure download the corresponding jpg file : `` f5.jpg '' cccccccccccccccc 1 & 01:29:35.66&@xmath273:33:39.61&15.90&0.22&16.24&0.30&13.64&0.24&11.99&0.15 & 2&24.62 & 22.85 & 24.46 & 22.63 + 2 & 01:29:25.95&@xmath273:34:46.69&16.52&0.23&16.82&0.32&14.17&0.31&12.50&0.19 & 5&25.45 & 23.62 & 24.66 & 22.45 + 5 & 01:29:26.52&@xmath273:32:30.31&16.27&0.15&16.45&0.21&14.57&0.29&12.91&0.15 & 2&17.84 & 17.72 & 17.82 & 17.66 + 22&01:29:44.77&@xmath273:33:25.31&16.46&0.21&16.50&0.26&14.70&0.32&13.06&0.17&15&23.26 & 21.76 & 21.82 & 20.28 + 24&01:29:39.23&@xmath273:33:18.80&16.66&0.20&16.69&0.31&14.45&0.36&12.79&0.22 & 5&24.54 & 21.85 & 23.12 & 20.93 + 26&01:29:19.80&@xmath273:33:12.46&16.06&0.23&16.04&0.29&13.84&0.30&12.17&0.17 & & & & & + 29&01:29:35.55&@xmath273:33:30.20&14.13&0.09&14.09&0.11&12.22&0.13&10.61&0.08 & 8&17.49 & 17.30 & 16.73 & 16.65 + 34&01:29:23.08&@xmath273:32:51.51&16.00&0.16&15.93&0.17&14.28&0.26&12.88&0.13 & 4&22.82 & 20.95 & 22.06 & 20.13 + 35&01:29:22.34&@xmath273:32:48.17&16.67&0.27&16.60&0.36&14.63&0.35&12.95&0.22 & & & & & + 36&01:29:28.61&@xmath273:34:12.14&16.88&0.31&16.80&0.42&15.32&0.50&13.68&0.29 & 1&25.15 & 24.08 & & + 38&01:29:34.08&@xmath273:34:11.66&16.67&0.28&16.58&0.39&14.84&0.39&13.10&0.18 & & & & & + 42&01:28:47.20&@xmath273:31:31.69&16.70&0.24&16.54&0.32&15.37&0.51&14.02&0.29 & & & & & + 43&01:29:41.38&@xmath273:36:24.53&15.92&0.11&15.71&0.15&14.67&0.33&12.87&0.11 & & & & & + 45&01:29:43.34&@xmath273:33:43.21&16.50&0.20&16.21&0.20&14.19&0.29&12.97&0.18 & 2&21.90 & 21.22 & 21.89 &
21.16 + 46&01:29:30.92&@xmath273:34:05.74&16.51&0.23&16.17&0.27&14.80&0.38&13.36&0.21 & 1&22.45 & 20.92 & & + 49&01:29:20.44&@xmath273:33:16.90&16.68&0.21&16.19&0.23&15.16&0.50&13.65&0.21 & 1&23.16 & 21.55 & & + 50&01:29:37.52&@xmath273:33:51.34&15.00&0.18&14.50&0.20&13.09&0.22&11.70&0.19 & 1&20.96 & 20.42 & & + 51&01:29:06.45&@xmath273:33:48.73&13.91&0.08&13.35&0.10&12.15&0.11&10.69&0.06&10&20.16 & 19.56 & 20.05 & 19.19 + 52&01:29:30.21&@xmath273:33:10.55&15.49&0.09&14.87&0.09&13.83&0.16&12.53&0.09 & 1&19.30 & 18.62 & & + 53&01:29:31.67&@xmath273:34:09.39&15.36&0.14&14.60&0.13&13.42&0.17&11.94&0.07 & 1&26.44 & 24.28 & & + 54&01:29:35.67&@xmath273:33:02.11&13.07&0.07&12.26&0.05&11.18&0.08 & 9.86&0.06 & 9&19.34 & 18.54 & 19.29 & 18.42 + 57&01:29:35.62&@xmath273:33:01.52&15.14&0.20&12.26&0.05&11.18&0.08 & 9.86&0.06 & 7&24.20 & 22.37 & 22.67 & 20.78 + we use the results of the photometry by schmalzl et al .
( 2007 ) on acs / wfc imaging of ngc 602 to identify the optical counterparts of the candidate ysos found with _
spitzer _ located in this region .
we cross - correlate the positions of the candidate ysos with the ones of all stars found with acs . for the comparison of the two catalogs we take into account
the fact that the resolution provided by irac is almost 25 times lower than the one of wfc , and therefore the selected area around each candidate yso to be searched for optical counterparts in most of the cases revealed more than one optical sources . specifically , we selected a search circle with diameter of 24 , which corresponds to the size of two irac pixels , a typical fwhm of the point - sources found with our psf photometry . in table
[ tab - yso ] the results of this search are given .
specifically , in the last five columns of this table we give the number of stars from the acs photometry , found within a radial distance of 12 ( 2 irac pixels ) around the center of each candidate yso .
we also give the magnitude of the brightest optical source in the filters @xmath14 ( @xmath15 ) and @xmath16 ( @xmath17 ) respectively , as well as the corresponding total magnitudes from all optical sources found within the searched area around each candidate yso . in table
[ tab - yso ] there are five candidate ysos for which no optical counterpart was found .
however , only three of them ( no 26 , 35 and 38 ) are located within the acs / wfc field - of - view for which the cross - corelation was possible .
the remaining two candidate ysos ( no 42 and 43 ) are located outside this field - of - view , since the selected area observed by irac ( fig .
[ irac - maps ] ) is a bit larger than the one covered by acs .
as shown in table [ tab - yso ] , almost all of the detected candidate ysos in the area of ngc 602/n 90 are related to stars visible at optical wavelengths . in half of the cases
there are more than two optical sources related to each candidate yso ( within a region corresponding to two irac pixels ) .
the loci of the detected candidate ysos are shown marked with red squares on a color composite image from the observations taken with acs / wfc of the area of ngc 602/n 90 in fig .
[ fig - ysomap ] . in this figure
it is shown that there is a higher concentration of candidate ysos to the east and south , where also the nebula is brighter and where the brighter 8 emission is found in the _ spitzer _ images ( fig .
[ irac - maps ] ) .
more detailed color composite images from acs imaging of the loci around 13 selected candidate ysos are shown in fig .
[ fig - ysosamples ] .
circles of radius corresponding to 12 mark the centers of the candidate ysos . in this figure
it can be seen that there are candidate ysos coinciding with stellar clusterings , which could be true compact clusters ( objects no 22 , 24 , 29 , 54/57 ) . the nature of these compact clusters and if they are signatures of on - going clustered star formation are two very important issues .
we attempt to address these issues by plotting the cmd of all stars from our acs photometry , that are associated with the candidate ysos found with irac . the part of a @xmath18 , @xmath15 cmd , which is expected to be occupied by pre - main sequence ( pms ) stars of star - forming regions in the mcs observed with _
hubble _ is fairly known ( e.g. gouliermis et al .
2006 , nota et al .
this cmd for the optical sources related to the candidate ysos in the region of ngc 602/n 90 , plotted in fig .
[ fig - cmdpms ] , shows that all of the stars found to coincide with the candidate ysos in our sample are actually pms stars , suggesting that either we probably observe the existence of compact pms star clusters recently formed ( or currently under formation ) around typical ysos , or that the sources identified as candidate ysos _ are _ themselves unresolved young pms clusters . in either case
it is almost certain that these stellar concentrations demonstrate possible events of extragalactic clustered star formation on the act .
if so , it would be interesting to know how long ago this formation took place . to answer this question , on the cmd of fig .
[ fig - cmdpms ] we also plot pms evolutionary tracks from the models by palla & stahler ( 1999 ) , and two indicative isochrones ( for 0.1 and 5 myr ) and the zero age main sequence ( zams ) from the same models
. simulations of the positions at the @xmath18 , @xmath15 cmd of all pms stars found in the region of ngc 602/n 90 by schmalzl et al .
( 2007 ) indicate that reddening , variability and binarity are important factors , which cause a spread of the loci of pms stars in this cmd and that even modest interstellar reddening ( schmalzl et al .
find @xmath19 0.04 - 0.06 mag for the association ngc 602 ) can make these stars appear younger ( being `` pushed '' to redder colors ) . considering these results and the plot of fig .
[ fig - cmdpms ] , one may conclude that the pms stars , which coincide with the candidate ysos of our sample do not seem to be older than about 5 myr , giving an upper limit to the time - scale within which these clusters were formed . for a more detailed analysis one should concentrate on the stellar content of each of the individual clusters , but in most of the cases very low numbers of stars per cluster
do not allow such an analysis .
we selected the most populous among the identified clusters ( the ones corresponding to candidate ysos no 22 , 29 , 51 and 54 ) to plot their individual cmds , but the stellar numbers were still not enough to define a more narrow age range .
pms stars are mainly red sources , and therefore near - infrared imaging with cameras equipped with adaptive optics ( ao ) systems would provide the necessary resolution for a richer sample of pms stellar members in these clusters .
indeed , high resolution near - infrared observations of these clusters would certainly be very useful , because they would 1 ) provide better statistics with higher numbers of detected members , enough to plot the cmds of individual clusters and 2 ) fill the gap between optical and mid - infrared wavelengths , providing near - infrared magnitudes for these pms stars .
the later information will offer the opportunity to construct accurate spectral energy distributions ( seds ) of candidate ysos , to be compared to models . as far as candidate ysos that coincide with no more than two pms stars concerns , in fig .
[ fig - ysosamples ] we also show several different cases of `` single '' candidate ysos ( e.g. objects no 1 , 5 , 49 , 53 ) , but as their images show it is unclear if such sources are indeed single isolated objects or embedded clusters ( as the faint magnitudes given in table [ tab - yso ] suggest ) , or even one massive embedded protostar .
modeling of the seds of such objects is quite difficult , in the sense that many different models seem to fit the seds , due to their small number of points , and near - infrared ao observations would certainly add three more points to the seds , throwing more light toward this direction as well .
it is worth noting that fig .
[ fig - ysosamples ] also demonstrates few interesting cases of candidate ysos ( objects no 26 , 49 , 34 , 35 , 46 and 53 ) , both with and without optical counterparts , which are located at the edge of features similar to the `` pillars of creation '' or `` elephant trunks '' in the galactic h ii region m 16 ( the eagle nebula ; hester et al . 1996 ) .
the sample of candidate ysos derived from our irac photometry in the region of ngc 602/n 90 ( table [ tab - yso ] ) is based on a single classification scheme .
as a consequence , although it is almost certain that this sample does include true ysos , there is always a possibility that it also includes sources misidentified as ysos , like small h ii regions , or that it is quite incomplete not including all possible true ysos , like the class i ysos in the sample of allen et al .
the identification of pms stars in the areas of the candidate ysos as their optical counterparts indicates that half of these objects are not single sources .
but , considering the difficulty of modeling the seds of these objects , it is not straightforward to clarify which of them represent embedded proto - clusters . in the case of candidate ysos
coinciding with one or two pms stars , there is no information if there should be more optical counterparts , which simply were not detected in our acs photometry , or if these are indeed single proto - stellar objects . in any case , the composite acs image of the area of ngc 602/n 90 shown in fig .
[ fig - ysomap ] provides evidence that sequential star formation does take place in the area and especially on the dust ridges of a cavity , which seems to extend outwards .
this feature is most probably the product of photoionization by the young association , ngc 602 , located almost at its center . the combination of both acs ( fig .
[ fig - ysomap ] ) and irac ( fig .
[ irac - maps ] ) observations provides a more complete image of the star formation in the area .
specifically , taking into account the loci of the sources classified as candidate ysos , the ism features shown in the image from acs , and the 8 emission shown in the irac image , one may conclude that in the observed area the stellar association ngc 602 ionizes the surrounding material giving birth to the h ii region , n 90 , and producing a cavity in the remaining of its parental molecular cloud .
this event triggers sequential star formation not later than @xmath11 5 myr ago , which seems to propagate mostly to the east and south .
this second star formation event may be characterized by higher a concentration of ysos , most of which may be related to compact pms clusters .
this scenario is also supported by the study of the pms population in ngc 602 and its surroundings by schmalzl et al .
we present our results from _ spitzer_/irac photometry on the star - forming region ngc 602/n 90 in the small magellanic cloud .
we performed psf photometry on irac images taken in four mid - infrared channels ( 3.6 , 4.5 , 5.8 and 8.0 ) centered on the h ii region n 90 .
we identified 57 objects in all four channels and 44 in channels 1 , 2 and 3 .
we classified the sources found in all four irac channels from their loci in the [ 3.6]@xmath2[8.0 ] , [ 8.0 ] color - magnitude diagram , based on templates and models for different red sources in the milky way , and the [ 5.8]@xmath2[8.0 ] , [ 3.6]@xmath2[4.5 ] color - color diagram , according to the classification scheme developed by bolatto et al .
( 2007 ) for smc ysos .
according to our classification there are 22 candidate ysos in the region of ngc 602/n 90 .
we use our results from psf photometry in the filters @xmath15 and @xmath17 of the same region from imaging with the wide - field channel of acs on - board the _ hubble _ space telescope ( schmalzl et al .
2007 ) to identify optical sources , which coincide with the candidate ysos .
all these optical sources are identified as pms stars .
we found that almost half of the candidate ysos match with no more than two pms stars , but examination of their loci on the acs images does not exclude the possibility that there may be more highly embedded sources or not well resolved also in the optical .
the remaining candidate ysos coincide with more than 3 pms stars , and four of them seem to include small compact pms clusters of 7 ( or more ) stars .
we found that these stars are probably formed earlier than 5 myr ago , according to pms evolutionary models .
we could not find any optical counterpart for five ysos , two of which are located outside the field observed with acs . from the positions of the detected candidate ysos
, we conclude that they are probably the products of a sequential star formation process , which presumably propagates mostly to the east and south , on the ridges of the molecular cloud , as they formed from the photo - ionizing process of the stellar association ngc 602 located in the center of the apparent cavity of the cloud .
finally , it should be mentioned that our sample of candidate ysos may be contaminated by other objects like small h ii regions , or it may not include some true ysos . based on our results
we stress the importance of high - resolution near - infrared imaging of the candidate ysos in order 1 ) to resolve more faint red stellar sources and detect a higher number of related pms stars , and 2 ) to fill the wavelength gap between optical and mid - infrared by providing near - infrared measurements of all detected sources , so that their seds will be thoroughly modeled , as well as of high - resolution radio observations , which will provide additional information on the ionizing flux for the identified objects .
dimitrios a. gouliermis kindly acknowledges the support of the german research foundation ( deutsche forschungsgemeinschaft - dfg ) through the individual grant 1659/1 - 1 .
quanz kindly acknowledges financial support from the german friedrich - ebert - stiftung .
this paper 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 , and on observations made with the nasa / esa hubble space telescope , obtained from the data archive at the space telescope science institute .
stsci is operated by the association of universities for research in astronomy , inc . under nasa contract
nas 5 - 26555 .
this research has made use of nasa s astrophysics data system , _ aladin _ ( bonnarel et al . 2000 ) , _ wcstools _ ( mink 2001 ) , and the simbad database , operated at cds , strasbourg , france . allen , l. e. , et al .
2004 , , 154 , 363 bakes , e. , & tielens , a. g. g. m. 1994 , apj , 427 , 822 benjamin , r. a. , et al .
2003 , pasp , 115 , 953 bolatto , a. d. , et al .
2007 , apj , 655 , 212 bonnarel , f. , et al .
2000 , a&as , 143 , 33 clayton , g. c. , wolff , m. j. , sofia , u. j. , gordon , k. d. , & misselt , k. a. 2003 , apj , 588 , 871 cohen , m. 1993 , aj , 105 , 1860 davies , r. d. , elliott , k. h. , & meaburn , j. 1976 , mmras , 81 , 89 desert , f .- x . ,
boulanger , f. , & puget , j. l. 1990 , a&a , 237 , 215 engelbracht , c. w. , et al .
2006 , apj , 642 , l127 fazio , g. , et al .
2004 , apjs , 154 , 10 gouliermis , d. , brandner , w. , & henning , t. 2006 , , 636 , l133 hartmann , l. , et al .
2005 , apj , 629 , 881 henize , k. g. 1956 , apjs , 2 , 315 hester , j. j. , et al .
1996 , , 111 , 2349 jger , c. , krasnokutski , s. , staicu , a. , huisken , f. , mutschke , h. , henning , t. , poppitz , w. , & voicu , i. 2006 , apjs , 166 , 557 jones , t. j. , woodward , c. e. , boyer , m. l. , gehrz , r. d. , & polomski , e. 2005 , , 620 , 731 hodge , p. 1985 , pasp , 97 , 530 li , a. , & draine , b. t. 2002 , apj , 576 , 762 massey , p. 1993 , massive stars : their lives in the interstellar medium , 35 , 168 massey , p. 2002
, , 141 , 81 mink , d. j. 2002 , asp conf .
ser . 281 : astronomical data analysis software and systems xi , 281 , 169 palla , f. , & stahler , s. w. 1999 , apj , 525 , 772 sauvage , m. , vigroux , l. , & thuan , t. x. 1990 , a&a , 237 , 296 schwering , p. b. w. 1989 , a&as , 79 , 105 simon , j. d. , et al .
2007 , submitted to apj schmalzl , m. et al .
2007 , to be submitted to apj stanimirovi , s. , staveley - smith , l. , van der hulst , j. m. , bontekoe , t. r. , kester , d. j. m. , & jones , p. a. 2000 , mnras , 315 , 791 weingartner , j. c. , & draine , b. t. 2001 , apj , 548 , 296 werner , m. , et al .
2004 , apjs , 154 , 1 whitney , b. a. , wood , k. , bjorkman , j. e. , & cohen , m. 2003 , apj , 598 , 1079 whitney , b. , indebetouw , r. , bjorkman , j. e. , & wood , k. 2004 , apj , 617 , 1177 wolfire , m. , hollenbach , d. , mckee , c. f. , tielens , a. g. g. m. , & bakes , e. l. o. 1995 , apj , 443 , 152
|
we present _ spitzer_/irac photometry on the star - forming h ii region
n 90 , related to the young stellar association ngc 602 in the small magellanic cloud .
our photometry revealed bright mid - infrared sources , which we classify with the use of a scheme based on templates and models of red sources in the milky way , and criteria recently developed from the _ spitzer _ survey of the smc ( bolatto et al .
2007 ) for the selection of candidate young stellar objects ( ysos ) .
we detected 57 sources in all four irac channels in a 62 @xmath0 48 field - of - view centered on n 90 ; 22 of these sources are classified as candidate ysos .
we compare the locations of these objects with the position of optical sources recently found in the same region with high - resolution hst / acs imaging of ngc 602 by schmalzl et al .
( 2007 ) , and we find that 17 candidate ysos have one or more optical counterparts .
all of these optical sources are identified as pre - main sequence stars , indicating , thus , ongoing clustered star formation events in the region .
the positions of the detected ysos and their related pms clusters give a clear picture of the current star formation in n 90 , according to which the young stellar association photo - ionizes the surrounding interstellar medium , revealing the h ii nebula , and triggering sequential star formation events mainly along the eastern and southern rims of the formed cavity of the parental molecular cloud . the postscript file is removed due to size limitations . + to see the figure download the corresponding jpg file : `` f1.jpg ''
|
accelerator based x - ray sources produce nowadays very intense radiation in a broad spectral range @xcite .
third generation synchrotron radiation light sources , with reduced emittance and large use of insertion devices provide a high brilliance with partial transverse coherence for users , enabling for example coherent imaging experiments @xcite . furthermore ,
fourth generation light source generally rely on the free electron laser ( fel ) process using relativistic electrons propagating in a periodic magnetic field as a gain medium .
fels provide additional longitudinal coherence and extremely short pulses , enabling to follow dynamics process of dilute species @xcite . +
a planar undulator of period @xmath3 and peak field @xmath4 emits with an observation angle @xmath5 a radiation at wavelength @xmath6 and its harmonics according to : @xmath7 with the deflection parameter given by @xmath8 \times \lambda_u[mm]$ ] , @xmath9 is the electron charge , @xmath10 the electron mass , @xmath11 the speed of light , and @xmath12 the normalized energy .
the radiation produced in an undulator is very intense and concentrated in narrow energy bands in the spectrum .
@xmath6 can be varied by a modification of the undulator magnetic field amplitude ( eq . [ eqn1 ] ) .
undulators with high fields and short period enabling a higher number of periods to be used for a given length can be employed on intermediate energy storage rings @xcite and short wavelength fel , for providing hard x - ray radiation . +
different technologies can be used for generating the undulator periodic magnetic structure @xcite . in the case of a permanent magnet undulator , in the so - called _ halbach
_ configuration @xcite , two parallel arrays separated by an air - gap accommodate magnets with the magnetisation rotating from one block to the other by 90 @xmath13 . in hybrid undulators @xcite , the vertical magnetized magnet blocks are replaced by soft iron poles which further increase the magnetic field strength in the gap of the undulator . + the undulator peak field can be enhanced as the magnetic gap between the upper and lower arrays of magnets is decreased .
the reduction of the gap is limited by the size of the vacuum chamber @xcite and sets some restrictions in terms of physical aperture for the electron beam evolution .
the idea to place the entire undulator in vacuum enabled the user to reach smaller gaps , and thus higher magnetic fields @xcite .
rare earth materials are used for the magnets to generate the magnetic field @xcite .
large coercive materials have better resistance against demagnetisation that might occur due to electron beam losses or vacuum baking @xcite .
@xmath14 @xcite magnets present a high coercivity @xmath15 and a good radiation resistance against demagnetisation @xcite .
@xmath2 @xcite magnets achieve a higher remanent field @xmath16 , with intermediate coercivities @xcite .
unfortunately large coercive @xmath2 magnets show small remanence magnetization .
therefore , the undulators can not take full advantage of the magnetic performance of @xmath2 . in order to shift further
the emitted radiation towards higher energies ; i.e. to the hard x - ray region , the peak magnetic field of the in - vacuum undulators can be increased when operating at cryogenic temperature . cooling down @xmath2 permanent magnets increases the remanent magnetization @xmath17 up to a certain temperature at which the process is limited by the appearance of the spin reorientation transition ( srt ) phenomenon @xcite .
the easy magnetisation axis is tilted from the crystallographic c - axis [ 001 ] by an angle that increases when lowering the temperature @xcite .
the coercivity is not affected by the srt phenomenon and remains increasing at low temperature .
in contrast to the @xmath2 case , when a @xmath1 permanent magnet is cooled down to cryogenic temperature , no srt occurs and its remanent magnetization keeps increasing at least until the liquid helium temperature of 4.2 k @xcite . such a magnet grade is well adapted for the cryogenic undulator application since it enables direct cooling at the liquid nitrogen temperature ( 77 k ) , enabling a high level of thermal stability .
cryogenic permanent magnet undulator ( cpmu ) is one of the evolutions of in - vacuum undulators .
initially proposed at spring-8 @xcite , where a prototype of 0.6 m length with a period of 14 @xmath18 , using high remanence @xmath2 grade cooled down to 140 @xmath19 , has been developed .
following the first demonstration at spring-8 @xcite , several cpmu prototypes were built .
after first prototypes @xcite , a 8 @xmath18 x 14.5 @xmath18 periods cpmu prototype has been built at brookhaven national laboratory ( bnl ) using @xmath2 grade reaching a magnetic gap of 5 @xmath18 .
another 8 periods prototype with 16.8 @xmath18 period and a gap of 5 @xmath18 hybrid undulator composed of high coercive @xmath1 magnets ( cr-47 ) and vanadium permendur poles , have been manufactured at bnl @xcite in order to investigate the resistance of the cr-47 magnets to a baking at 373 @xmath19 .
helmholtz - zentrum berlin ( hzb ) with the collaboration of ucla @xcite , built two cmpu prototypes ( 20 x 9 @xmath18 period and magnetic gap of 2.5 @xmath18 ) , using @xmath1 magnets cooled down to a of 20 - 30 @xmath19 . at soleil ,
three hybrid prototypes cpmu have been built and characterized , a 4 periods 18 @xmath18 length with @xmath2 magnets ( bh50 hitachi - neomax)@xcite , also a 4 @xmath18 x 18 @xmath18 period with @xmath1 ( cr53 hitachi - neomax ) @xcite , and a 4 @xmath18 x 15 @xmath18 period with @xmath1 .
the first full scale cryogenic undulator had been developed at esrf @xcite with a period length of 18 @xmath18 using a relatively low remanence @xmath2 magnet grade ( @xmath16=1.16 @xmath20 ) cooled down to around 150 k , reaching a gap of 6 @xmath18 .
esrf recently developed two more cryogenic undulators , one with the same period length as the first using a high remanence @xmath2 ( @xmath16=1.38 @xmath20 ) cooled down to 135 @xmath19 @xcite , and a second one of period 14.5 @xmath18 using @xmath1 magnet grade cooled down to 100 @xmath19 .
based on the esrf development , danfysik build for diamond @xcite a 17.7 @xmath18 period cryogenic undulator using high remanence @xmath2 magnets cooled down to 150 k and a magnetic gap of 5 @xmath18 .
@xmath1 based cmpus are under construction at diamond .
spring-8 in collaboration with sls @xcite developed a cryogenic undulator using high remanence @xmath2 cooled down around 140 k. the cryogenic undulators developed so far use @xmath2 permanent magnet working around 140 k. they are cooled down to the liquid nitrogen temperature at 77 k and heated to reach the working temperature in order to avoid the srt phenomenon appearance .
hzb is currently building 2 new @xmath1 based cmpu ( 1.6 @xmath10 long with a 17 @xmath18 period with a gap of 5 @xmath18 , and a 2 @xmath10 long with 15 @xmath18 period and a gap of 2 @xmath18 ) .
a 2 @xmath10 long cryogenic undulator with 140 @xmath18x 13.5 @xmath18 period is under construction for the high energy photon source test facility ( heps - tf ) in korea .
@xmath1 based cmpus are under construction at national synchrotron radiation research center @xcite . in this paper , the development of the first full scale ( 2 m long ) @xmath1 cryogenic undulator which has been installed on a storage ring is presented , i.e. soleil in france .
it has 107 periods of length 18 mm .
the magnetic analysis and measurement results indicate that the quality of the magnet grade satisfies the undulator requirements and demonstrate a further increase of magnetic field ( by a few percents ) . we show that the direct operation at 77 k enables a good thermal stability .
we also present the design , construction steps ( with optimization , magnetic measurements and shimming at low temperature ) .
we then report on the commissioning with the electron beam and the current successful operation both from an electron or photon point of view ( with measurements of undulator radiation ) .
indeed , we show that some results of photon beam alignment using the nanoscopium long section beamline , or precise adjustment of the undulator taper with the photon beam itself .
the soleil @xmath1 based u18 cryogenic undulator has been modeled using radia @xcite software , as illustrated in fig .
the characteristics are indicated in table [ table2 ] , considering the longitudinal coordinate s , the transverse horizontal and vertical are x and z respectively .
the model contains two parts ; the central one ( fig . [ fig1](a ) ) constituted by full size permanent magnets and poles producing the main magnetic field of the undulator , and the extremities ( fig . [ fig1](b ) ) located at each end of the undulator constituted by two magnets and one pole .
the extremity parts are optimized to minimize the on - axis field integral variations versus the undulator gap @xcite .
.soleil cryogenic undulator main characteristics . [ cols="<,^,^",options="header " , ] [ table19 ] table [ table19 ] presents the total calculated thermal heat load dissipated by the liquid nitrogen closed loop in the case of the @xmath21 build at soleil .
the heat load comes from heat deposited by the beam ( wakefield and synchrotron radiation ) , the vacuum chamber , and the rods .
the heat load coming from the electron beam represents more than 25@xmath22 of the total heat load , due to the synchrotron radiation coming from the upstream dipole and the wake field power deposited by the passage of the electron beam on the liners and tapers .
half of the dissipated heat load is coming from the rods , which connect the out - vacuum and in - vacuum girders .
the diameters and contact surfaces of the rods have been reduced without any effect on their rigidity to reduce the heat load .
since the vacuum chamber is at room temperature , whereas the in - vacuum girders are at liquid nitrogen temperature , the inside part of the vacuum chamber has been polished to improve the emissivity and to reduce the heat load by a factor of 2 .
the cooling down of the undulator to liquid nitrogen temperature causes contraction on the in - vacuum girders and the rods , calculated with equation @xmath23 , where @xmath24 is the contraction , @xmath25 is the thermal dilatation coefficient and @xmath26 is the temperature variation .
the contraction of the in - vacuum girder by 8 mm is then taken into account in the magnet assembly at room temperature by keeping 40 @xmath27 between magnet modules .
the contraction of the rods by 1 mm causes an opening of the gap by 1 mm , which should be adjusted before conducting magnetic measurements .
the assembly and the magnetic corrections of the cryogenic undulator are performed at room temperature in the same conditions as a standard in - vacuum undulator .
a standard magnetic bench allowing hall probe and flip coil measurements has been used for the assembly and corrections at room temperature .
an optimization software called id - builder @xcite developed at soleil has been used at all steps of the undulator construction : magnets sorting , period assembly , shimming ( vertical displacement of magnets and poles to correct the field integrals and the phase error ) , and magic fingers ( small magnets installed at the extremities of the undulator to correct the field integral ) .
the electron beam should emit the most intense radiation when it crosses the undulator without disturbing the beam dynamics in the storage ring .
the figures of merit during the assembly and corrections are the field integrals , the trajectory straightness and the phase error .
they have to be minimized to reduce the impact of the magnetic errors on the undulator performances in terms of photon spectrum and beam dynamics .
figure [ fig5 ] presents the vertical field integral @xmath28 and the horizontal field integral @xmath29 of the cryogenic undulator versus transverse position at minimum gap of 5.5 mm at the end of the assembly at room temperature and after the complete iterative correction process .
after the corrections , the undulator field integrals present a smoother variation versus horizontal position and the on - axis integral is less than 0.4 @xmath30 and the higher off - axis field integral are reduced from 3 @xmath30 to less than 1 @xmath30 . )
vertical field integral ( iz ) after magic finger corrections .
( @xmath31 ) horizontal field integral ( ix ) after magic finger corrections .
( @xmath32 ) vertical field integral ( iz ) before magic finger corrections ( @xmath33 ) horizontal field integral ( ix ) before magic finger corrections .
field integrals precision is @xmath34 0.05 @xmath30 , and local measurements ( hall probe ) is 0.5 @xmath35 . ]
figure [ fig6 ] presents the electron beam trajectory in the cryogenic undulator calculated from the magnetic field measurements versus the longitudinal position at the end of the assembly at room temperature and after the corrections .
the trajectory position at the exit of the undulator after the magnetic assembly is 8 @xmath27 , it is kept below 2 @xmath27 after magnetic corrections . ) .
( solid line ) horizontal and vertical trajectories after the assembly and corrections .
( dash line ) horizontal and vertical trajectories after the assembly and before shimming .
] the rms phase error ( calculated by b2e from the measured magnetic field @xcite ) at minimum gap after the magnetic assembly of the undulator is found to be @xmath36 , and has been corrected to @xmath37 after shimming the magnet modules .
the vertical position of the magnet modules is modified by using shims ( copper pieces ) to correct locally the magnetic field value .
magnetic measurement benches have been developed over the past years to increase precision and enable to conduct measurements for insertion devices at low temperature .
a cryogenic undulator requires a specific bench with certain characteristics .
esrf @xcite had developed a magnetic bench to perform magnetic field measurements for cryogenic undulators .
this bench is mounted inside a vacuum chamber , where the hall probe ( with a ball bearing ) is fixed on a mobile carriage guided along a rail .
the longitudinal position of the hall probe is measured with a laser tracker .
the stretched wire motion is performed by two translation motors located outside the vacuum chamber and transmitted inside by rods through bellows .
spring-8 developed a bench called safali @xcite for hall probe magnetic measurements and is mounted inside the vacuum chamber of the undulator .
the hall probe is displaced longitudinally by a vacuum compatible stepper motor installed inside the vacuum chamber .
the mechanical default position of the hall probe on the rail is measured with lasers and corrected with small displacements on the rail .
helmholtz - zentrum berlin ( hzb ) built a 2 @xmath10 long in - vacuum hall - probe measuring bench @xcite for the characterization of several in - vacuum cryogenic undulators under development .
the bench employs a system of laser interferometers and position sensitive detectors , which are used in a feed - back loop for the hall probe position and orientation .
figure [ fig7 ] presents the cryogenic magnetic measurement bench developed at soleil .
it is used to measure the field created by the undulator at both room and cryogenic temperature .
the local magnetic field and field integrals are measured by a hall probe ( model bell gh701 ) and a stretched wire respectively .
the cryogenic magnetic measurement bench is installed inside the undulator vacuum chamber and is removed after the measurements .
the hall probe is fixed on a trolley which moves on a rail by a stepper motor , mounted outside the vacuum chamber and whose movement is transmitted inside the vacuum chamber via a magnetic coupling system .
the rail is mechanically independent of the vacuum chamber and is fixed from the outside by seven rods .
the deformations of the rail are measured with an optical system and the longitudinal position of the hall probe is measured with a thermalized optical rule ( heidenhain lida 405 ) .
the deformations of the rail are measured with an optical system constituted by an angular reflector for the hall probe angular deformation measurements , and a linear reflector ( renishaw ) for the hall probe horizontal and vertical deformation measurements .
the cryogenic magnetic bench is aligned inside the vacuum chamber .
the angular deformations of the bench are corrected by using shims with different thicknesses installed at the deformation location , however the vertical deformation of the bench is not corrected mechanically by shims , and it is corrected directly on the magnetic measurement results . ) @xmath38 after corrections , ( @xmath39 ) @xmath40 after corrections , ( @xmath41 ) @xmath42 after corrections , ( @xmath33 ) @xmath38 before corrections , ( @xmath32 ) @xmath40 before corrections ( @xmath43 ) @xmath42 before corrections . ] figure [ fig8 ] presents the rotation angle of the hall probe at different longitudinal positions before and after correction .
the angles are measured with a laser interferometer , they are reduced from @xmath340.8 @xmath44 to @xmath340.1 @xmath44 using small mechanical shims located along the longitudinal position .
figure [ fig9 ] presents the measured horizontal and vertical hall probe angles at different longitudinal positions of the cryogenic magnetic bench .
the horizontal hall probe position variation is 81 @xmath45 m along the bench .
this variation is negligible for the magnetic field measurement because of the very low field gradient ( calculated with radia ) over few millimeters in this direction ( @xmath46 @xmath47 ) .
however the vertical variation of 287 @xmath45 m along the bench has a non - negligible effect on the magnetic field measurement as the field gradient ( calculated with radia ) is high in this direction ( 45 @xmath47 ) . the measured magnetic field
is then corrected using the equation [ eqn5 ] .
@xmath48 where @xmath49 is the corrected magnetic field , @xmath4 the measured magnetic field , @xmath50 the vertical default position , @xmath51 the longitudinal position and @xmath52 is a factor which depends on the undulator characteristics , a radia calculation identifies a value of @xmath52 = 1.12 . ) translation in the vertical z axis direction,(@xmath53 ) translation in the horizontal x axis direction . ]
the hall probe is calibrated with temperature variation .
the resistance of the hall probe ( r ) is measured at different temperatures ( t ) and a linear variation with temperature is measured , leading to @xmath54 with @xmath55 @xmath56 , @xmath57 @xmath58 .
the hall probe temperature varies by 4 @xmath19 during the magnetic measurements .
this variation and the thermal sensitivity coefficient of the hall probe -0.0043 @xmath22/k are taken into account to correct the measured magnetic field values .
figure [ fig10 ] presents the magnet temperature and vacuum pressure variation during the cooling down of the cryogenic undulator using a cryocooler system ( cryotherm bruker ) .
the magnets reach liquid nitrogen temperature of 77 @xmath19 after 6 hours . despite the absence of backing of the undulator to prevent the permanent magnet from demagnetization , the undulator vacuum pressure drops quickly , due to the cold mass which acts as a cryo - pump .
the warming of the undulator from cryogenic to room temperature is relatively long , about 72 hours , but it could be accelerated ( if needed ) by injecting nitrogen gas in the cooling circuit .
the warming time is then reduced to 36 hours . fig .
[ fig : warmingup ] shows the warming of the undulator by injecting nitrogen gas at 60 @xmath59c under 2.5 @xmath60 in the undulator to reduce again this long delay to approximately 24 hours .
figure [ fig11 ] presents the hall probe magnetic measurement results at room temperature and cryogenic temperature of 77 @xmath19 .
the effect of cooling down the permanent magnets results in an increase of the magnetic field by 11@xmath22 as predicted ( from 1.041 @xmath20 to 1.157 @xmath20 ) . .
( red curve ) at room temperature 293 @xmath19 , ( blue curve ) at cryogenic temperature 77 @xmath19 , with a precision of 0.5 @xmath35 . ]
figure [ fig12 ] presents the electron trajectory at room temperature and cryogenic temperature calculated from the measured magnetic field . despite the cooling down , the electrons trajectory position is kept below 4 @xmath27 along the undulator .
, ( blue curve ) at cryogenic temperature 77 @xmath19 . ]
the cryogenic undulator rms phase error has been corrected ( using shims ) at room temperature to 2.8@xmath59 .
after the cooling down , the rms phase error is increased to 9.1@xmath59 because of mechanical contractions .
the rods are contracted vertically by 1 @xmath18 , the variation of this contraction between rods causes in - vacuum girder deformation and then phase error degradation .
mechanical shims have been used to modify the vertical position of the 24 rods in order to correct the rms phase error and bring it down to 3 @xmath59 .
the thickness of the shim is calculated using equation [ eqn7 ] . @xmath61 where @xmath62 is the magnetic field variation due to a gap variation of @xmath63 , and @xmath4 is the measured magnetic field before the gap modification .
figure [ fig13 ]
( a ) presents the improvement of the phase error calculated from the magnetic field measurement at 77 k during rods shimming iterations .
the 3@xmath59 rms phase error is obtained from the @xmath64 iteration ; the @xmath65 and @xmath66 iterations do not improve further the phase error .
+ figure [ fig13 ] ( b ) presents the phase error variation versus gap .
the value at minimum gap of 5.5 mm is 3@xmath59 rms , the maximum value is a 3.2@xmath59 rms at gap 8 mm . at minimum gap of 5.5 @xmath18 .
( dash line ) at room temperature of 293 @xmath19 , ( solid line ) at cryogenic temperature of 77 @xmath19 .
beam characteristics shown in table [ table12 ] . ]
figure [ fig14 ] presents the spectrum of the cryogenic undulator at room temperature and cryogenic temperature of 77 @xmath19 calculated from the measured magnetic field at minimum gap . at 77 @xmath19 the photon flux increases by 24@xmath22 for the seventh harmonic and 34@xmath22 for the ninth harmonic
figure [ fig15 ] presents the cryogenic undulator installed in the long straight section sdl13 in soleil storage ring .
this straight section has been modified to allow for the installation of two canted in - vacuum undulators in order to extract two different long beam lines nanoscopium and anatomix @xcite .
after the installation and alignment of the undulator in the tunnel , the magnetic axis has to match with the electron beam path in the straight section .
a vertical misalignment of the undulator has a very strong effect on the magnetic field and leads to a non - optimized phase error and trajectories . to perform this optimization with the electron beam
, the whole carriage is moved in the vertical direction ( offset movement ) between + /-2 @xmath18 at a fixed gap of 5.5 @xmath18 and for each offset value , the vertical tune values are recorded ( see fig .
[ fig : aliguwe ] .
the magnetic axis corresponds to the offset value where the magnetic field value is miniumum so as the vertical tune ( @xmath67 scaled ) .
the reproducibility of this measurement is 50 @xmath27 .
the magnetic axis can also be determined by measuring the decay of the electron beam current versus offset .
figure [ fig16 ] presents the evolution of the undulator vacuum pressure during the first closing of the gap at high beam current of 400 @xmath68 .
the gap has been closed in only 40 min and the maximum pressure reached 3.10@xmath69 @xmath70 . . ]
figure [ fig17 ] displays the evolution of the undulator permanent magnet temperature during the first closing at the minimum gap at high beam current of 400 @xmath68 .
the average permanent magnet temperature without the electron beam is 82 @xmath19 .
it increases only by 2 @xmath19 at high electron beam of 400 @xmath68 .
the temperature dispersion between the permanent magnets is less than 3 @xmath19 , and has a negligible effect on the magnetic field variation value . .
] the focusing effect of the cryogenic undulator on the electron beam has been measured as a function of gap .
the vertical focusing is generated on one hand by the vertical field ( @xmath67 scaling ) and on the other hand by the normal integrated gradient measured on the magnetic bench .
the horizontal focusing is generated only by the normal integrated gradient measured on the magnetic bench .
figure [ fig19 ] presents the variation of both horizontal and vertical tunes versus gap .
as expected from magnetic measurements , a significant horizontal focusing is observed due to the non - zero normal integrated gradient measured on the bench ( + 200 @xmath35 ) .
actually the magnetic correction strategy was to perfectly cancel the skew integrated gradient term to the detriment of the normal one .
the vertical focusing is then reduced compared to the one expected from the measured vertical field scaling .
the u18 cryogenic undulator is in use by the nanoscopium long beamline .
this 155 @xmath10 long multimodal beamline @xcite is dedicated to scanning hard x - ray nano - probe experiments in the 30 @xmath71 @xmath72 @xmath27 spatial resolution range by combining x - ray fluorescence ( xrf ) , absorption spectroscopy ( xas ) , and phase - contrast imaging ) .
nanoscopium aims at reaching 30 - 200 @xmath71 resolution in the 5 - 20 @xmath73 energy range for routine user experiments .
the beamline design tackles the tight stability requirements of such a scanning nano - probe by creating an overfilled secondary source , implementing all horizontally reflecting main beamline optics , and constructing a dedicated high stability building envelope .
this beamline provides high sensitivity elemental and sample morphology mapping with down to 30 @xmath71 spatial resolution by fast scanning spectro - microscopy combined with absorption , differential phase contrast and dark field imaging and electron density mapping by coherent imaging techniques .
the typical scientific fields cover biology and life sciences , earth- and environmental sciences , geo - biology and bio - nanotechnology .
the beamline is especially well suited for studies seeking information about highly heterogeneous systems at multiple length scales also in natural or in operando conditions .
the spectrum emitted by the cryogenic undulator u18 has been measured on the nanoscopium long beamline .
figure [ fig18 ] shows the photon flux on the harmonic h9 at 11.873 @xmath73 of the spectrum measured on the beamline and compared to the one calculated from the magnetic measurements using the parameters of table [ table12 ] .
gap through a 0.1 @xmath18 x 0.1 @xmath18 aperture at a distance of 77 @xmath10 from the undulator .
electron beam parameters are shown in table [ table12 ] , with horizontal emittance = 4.6 @xmath71 , vertical emittance = 0.0473 @xmath71 , @xmath74= 7.579 @xmath10 , @xmath75= 2.34 @xmath10 , @xmath76= -0.851 @xmath77 , @xmath78= @xmath79 .
( a ) : @xmath80 harmonic , ( b ) : @xmath81 , and ( c ) : @xmath82 harmonic.,title="fig : " ] gap through a 0.1 @xmath18 x 0.1 @xmath18 aperture at a distance of 77 @xmath10 from the undulator .
electron beam parameters are shown in table [ table12 ] , with horizontal emittance = 4.6 @xmath71 , vertical emittance = 0.0473 @xmath71 , @xmath74= 7.579 @xmath10 , @xmath75= 2.34 @xmath10 , @xmath76= -0.851 @xmath77 , @xmath78= @xmath79 .
( a ) : @xmath80 harmonic , ( b ) : @xmath81 , and ( c ) : @xmath82 harmonic.,title="fig : " ] gap through a 0.1 @xmath18 x 0.1 @xmath18 aperture at a distance of 77 @xmath10 from the undulator .
electron beam parameters are shown in table [ table12 ] , with horizontal emittance = 4.6 @xmath71 , vertical emittance = 0.0473 @xmath71 , @xmath74= 7.579 @xmath10 , @xmath75= 2.34 @xmath10 , @xmath76= -0.851 @xmath77 , @xmath78= @xmath79 .
( a ) : @xmath80 harmonic , ( b ) : @xmath81 , and ( c ) : @xmath82 harmonic.,title="fig : " ] a very good agreement has been found in terms of bandwidth between the measured spectrum on the beamline and the calculated one from the magnetic measurements .
this result confirms the small optimized phase error of 3 @xmath59 rms obtained thanks to the in - situ corrections without dismounting of the vacuum chamber .
after the installation of an undulator in a storage ring , the magnetic axis is well aligned with the electron beam , to ensure the best performance possible . when the user beamline starts to be commissioned , the undulator alignment can be performed in a more precise manner in looking at the undulator harmonic shapes for different electron translation and angle bumps .
after some time , misalignment might occur and it is worth checking with the photon beam alignment .
thus experiments have been done on the spectrum of the undulator radiation with the nanoscopium beamline at soleil , to ensure good alignment .
one illustrates here an offset optimization while monitoring the spectrum of the undulator radiation on the nanoscopium long section beamline with a window aperture ( 0.2 @xmath18 @xmath83 0.8 @xmath18 ) placed 77 @xmath10 away from the undulator , and a photodiode placed at a distance of 83 @xmath10 .
the adjustment has been done by monitoring the @xmath81 harmonic , since high harmonics are very sensitive to any change in the beam parameters or the undulator characteristics .
the offset was varied , as in moving the girders up or down , while keeping the magnetic gap constant ( 5.5 @xmath18 ) .
placed at a distance of 77 @xmath10 from the undulator with electron beam parameters shown in table [ table12 ] , but with current i= 17 @xmath68 , @xmath74= 7.507 @xmath10 , @xmath75= 2.343 @xmath10 , @xmath76= -0.846 @xmath77 , @xmath78= 0.03 @xmath77 . ]
fig [ fig : offsetm ] shows the evolution of the intensity and bandwidth versus the offset variation .
the best alignment is found when the highest intensity with the lowest half band width are observed .
originally the offset was at 400 @xmath27 and is considered as the reference case for the new calculations shown in fig [ fig : offsetm ] , and the new one is found to be at @xmath34500 @xmath27 .
therefore u18 offset had been adjusted to this value to achieve the best alignment possible as in highest intensity .
these photon beam based alignment techniques are also carried out with electon ( translation , angle ) in the undulator , for a precise knowledge of the magnetic axis .
harmonic as a function of the offset that varied from 0.3 @xmath18 up to 0.8 @xmath18 . ]
undulator tapering technique ( i.e. variation of the peak field along the undulator axis ) is vastly used in the fel to improve its output power , by keeping the resonant radiation wavelength constant , despite the deviation in the energy of the electron beam . as for synchrotron radiation ,
it is one solution to perform an energy scan with a more constant intensity .
one way to modify the field is to slightly vary the gap at the entrance or exit of the undulator , either by closing or opening girders at the extremities .
fig [ fig : taper ] shows the change of intensity of the @xmath81 harmonic as we change the taper value . harmonic as a function of the taper offset . ]
the highest intensity with the lowest band width is at -10 @xmath27 ; i.e. the girders at the end of the undulator are closed by 10 @xmath27 .
this optimization was useful for the beamline at a precision of @xmath345 @xmath84 and increased the flux by @xmath85 .
the undulator has been installed in the storage ring since august 2011 .
the cryosystem , which is installed out of the tunnel , is connected to the undulator thanks to 2 flexible lines of 15 @xmath10 each via a `` valve panel '' .
this panel was developed by soleil enabling to change the cryosystem while keeping the undulator at cryogenic temperature .
this operation has been done only once in july 2014 where an ice plug blocked the flow in a line between the panel and the cryosystem .
after some tests , it appeared that this ice plug was in the panel which was bypassed until the end of the run . during the tests and the by - pass of the panel , the undulator started to warm - up leading to a beam loss of 4 hours due to a vacuum interlock .
this failure was the only problem on this cryogenic device in 5 years which led to a beam loss .
parameters of the cryosystems have been adjusted with the beam so that the thermal longitudinal gradient remains below 3 @xmath19 , while not decreasing the lifetime of the cryogenic pump .
no disturbance have been observed on the beam due to the liquid flow , whatever the parameters used for the cryosystem .
the undulator is warmed up one or twice a year , during electrical tests .
a 2 m @xmath1 cryogenic undulator with 18 @xmath18 period has been designed so that it suits for the nanoscopium long section beamline of soleil . the use of @xmath1 permits to operate at the liquid nitrogen temperature without undergoing srt effects resulting in an increase of the magnetic field peak , and attain a good thermal stability along the undulator .
in contrast , @xmath2 based cpmu requires thermal resistances distributed along the girders for an operation at 140 @xmath19 .
the cmpu has been built , optimized at room temperature .
it has been further corrected at cryogenic temperature using an additional in - vacuum measurement bench leading to a rms phase error of @xmath86 .
after the installation at soleil storage ring , it has been commissioned for different electron beam filling modes and different gaps .
the thermal behavior of the undulator is stable because the maximum operation temperature variation for all electron beam filling modes and gap variations is less than 3 @xmath19 ( instead of 26 @xmath19 for @xmath2 ) .
effects on the electron beam are in agreement with the magnetic measurements .
the spectrum has been measured on the beamline and shows a very good agreement with the one calculated from the magnetic measurements .
photon beam based alignment for the nanoscopium beamline provides a very fine tuning of the undulator axis .
the spectra are also very nice tools for the taper optimization allowing for a precision of 5 @xmath84 .
we would like to thank joel chavanne from esrf for his kind support , and members of the accelerator and engineering division of soleil led by a. nadji .
the authors are also very grateful for the support of j. m. filhol , the former head of the accelerator division of soleil and the european research council for the advance grant coxinel ( 340015 ) .
p. cloetens , r. barrett , j. baruchel , j .-
guigay , and m. schlenker , `` phase objects in synchrotron radiation hard x - ray imaging , '' _ journal of physics d : applied physics _ , vol .
29 , no . 1 , p. 133
, 1996 .
p. elleaume , j. chavanne , and b. faatz , `` design considerations for a 1 sase undulator , '' _ nuclear instruments and methods in physics research section a : accelerators , spectrometers , detectors and associated equipment _ ,
455 , no . 3 , pp .
503523 , 2000 .
m. e. couprie and m. vallau , `` radiation , polarisation , devices , new sources , vith international school on magnetism and synchrotron radiation : towards the fourth generation light sources , '' _ proc . of the 6th int .
school synchrotron radiation and magnetism(mittelwihr , france , ) _ , pp . 5194 , 2012 .
w. gudat , j. pflger , j. chatzipetros , and w. peatman , `` an undulator / multipole wiggler for the bessy storage ring , '' _ nuclear instruments and methods in physics research section a : accelerators , spectrometers , detectors and associated equipment _ , vol .
246 , no . 1 ,
pp . 5053 , 1986 .
s. yamamoto , t. shioya , m. hara , h. kitamura , x. w. zhang , t. mochizuki , h. sugiyama , and m. ando , `` construction of an in - vacuum type undulator for production of undulator x rays in the 525 kev region , '' _ review of scientific instruments _
63 , no . 1 ,
pp . 400403 , 1992 .
f. h. oshea , g. marcus , j. rosenzweig , m. scheer , j. bahrdt , r. weingartner , a. gaupp , and f. grner , `` short period , high field cryogenic undulator for extreme performance x - ray free electron lasers , '' _ physical review special topics - accelerators and beams _ , vol .
13 , no .
7 , p. 070702
, 2010 .
t. bizen , y. asano , t. hara , x. marechal , t. seike , t. tanaka , h. lee , d. kim , c. chung , and h. kitamura , `` baking effect for ndfeb magnets against demagnetization induced by high - energy electrons , '' _ nuclear instruments and methods in physics research section a : accelerators , spectrometers , detectors and associated equipment _ , vol .
515 , no . 3 , pp .
850852 , 2003 .
l. garca , j. chaboy , f. bartolom , and j. goedkoop , `` orbital magnetic moment instability at the spin reorientation transition of nd 2 fe 14 b , '' _ physical review letters _ ,
85 , no . 2 , pp .
429432 , 2000 .
s. hirosawa , y. matsuura , h. yamamoto , s. fujimura , m. sagawa , and h. yamauchi , `` magnetization and magnetic anisotropy of r2fe14b measured on single crystals , '' _ journal of applied physics _
59 , no . 3 , pp .
873879 , 1986 .
h. hiroyoshi , m. yamada , n. saito , h. kato , y. nakagawa , s. hirosawa , and m. sagawa , `` high - field magnetization and crystalline field of r2fe14b and r2co14b , '' _ journal of magnetism and magnetic materials _
70 , no . 1 - 3 ,
pp . 337339 , 1987 . d. goll , m. seeger , and h. kronmller , `` magnetic and microstructural properties of nanocrystalline exchange coupled prfeb permanent magnets , '' _ journal of magnetism and magnetic materials _ , vol . 185 , no . 1 ,
4960 , 1998 .
t. hara , t. tanaka , h. kitamura , t. bizen , x. marchal , t. seike , t. kohda , and y. matsuura , `` cryogenic permanent magnet undulators , '' _ physical review special topics - accelerators and beams _ , vol . 7 , no . 5 , p. 050702
, 2004 .
t. tanaka , r. tsuru , t. nakajima , and h. kitamura , `` magnetic characterization for cryogenic permanent - magnet undulators : a first result , '' _ journal of synchrotron radiation _
14 , no . 5 , pp .
416420 , 2007 .
t. tanabe , j. ablett , l. berman , d. a. harder , s. hulbert , m. lehecka , g. rakowsky , j. skaritka , a. deyhim , e. johnson , _ et al .
_ , `` x-25 cryo - ready in - vacuum undulator at the nsls , '' in _ synchrotron radiation instrumentation ; part one(aip conference proceedings volume 879 ) _ , vol . 879 , pp . 283286 , 2007 .
t. tanabe , o. chubar , d. a. harder , m. lehecka , j. rank , g. rakowsky , c. spataro , r. garrett , i. gentle , k. nugent , _ et al .
_ , `` cryogenic field measurement of pr2fe14b undulator and performance enhancement options at the nsls - ii , '' in _ aip conference proceedings _ , vol . 1234 , p. 29
, 2010 . c. kitegi , p. cappadoro , o. chubar , t. corwin , d. harder , p. he , h. fernendez , g. rakowsky , c. rhein , j. rank , _
et al . _ ,
`` development of a @xmath89 cryogenic undulator at nsls - ii , '' _ ipac _ , vol .
12 , p. 762
, 2012 . c. benabderrahmane , p. berteaud , m. vallau , c. kitegi , k. tavakoli , n. bchu , a. mary , j. filhol , and m. couprie , `` nd 2 fe 14 b and pr 2 fe 14 b magnets characterisation and modelling for cryogenic permanent magnet undulator applications , '' _ nuclear instruments and methods in physics research section a : accelerators , spectrometers , detectors and associated equipment _ , vol .
669 , pp . 16 , 2012 .
m. couprie , c. benabderrahmane , p. berteaud , f. briquez , c. bourassin - bouchet , f. bouvet , l. cassinari , l. chapuis , m. diop , j. daillant , _ et al . _ , `` the status of the lunex5 project , '' in _ proceedings fel _ , vol . 14 , 2014 . c. kitegi , j. chavanne , d. cognie , p. elleaume , f. revol , c. penel , b. plan , and m. rossat , `` development of a cryogenic permanent magnet in - vacuum undulator at the esrf , '' vol . 6 , pp .
35593561 , 2006 .
t. tanaka , t. seike , a. kagamihata , h. kitamura , t. schmidt , a. anghel , m. brgger , w. bulgheroni , and b. jakob , `` in - situ magnetic correction for cryogenic undulators , '' _ proceedings of ipac _ , vol .
10 , pp . 31473149 , 2010 .
huang , c .- h .
chang , c .- h .
chang , t .- y .
chung , c .- s .
hwang , h. kitamura , c .- k .
yang , and y. t. yu , `` challenge of in - vacuum and cryogenic undulator technologies , '' in _
7th international particle accelerator conference ( ipac16 ) , busan , korea , may 8 - 13 , 2016 _ , pp . 10801085 , 2016 .
r. walker , `` interference effects in undulator and wiggler radiation sources , '' _ nuclear instruments and methods in physics research section a : accelerators , spectrometers , detectors and associated equipment _ , vol .
335 , no . 1 ,
pp . 328337 , 1993 .
o. chubar , o. rudenko , c. benabderrahmane , o. marcouille , j. filhol , and m. couprie , `` application of genetic algorithms to sorting , swapping and shimming of the soleil undulator magnets , '' vol .
879 , no . 1
, pp . 359362 , 2007 .
t. tanaka , t. seike , a. kagamihata , t. schmidt , a. anghel , m. brgger , w. bulgheroni , b. jakob , and h. kitamura , `` in situ correction of field errors induced by temperature gradient in cryogenic undulators , '' _ physical review special topics - accelerators and beams _ , vol .
12 , no . 12 , p. 120702
, 2009 .
a. loulergue , c. benabderrahmane , f. bouvet , p. brunelle , m. couprie , j. denard , t. moreno , a. nadji , l. nadolski , f. polack , _ et al .
_ , `` double low beta straight section for dual canted undulators at soleil , '' _
wepea011 , ipac2010 , kyoto _ , pp . 24962499 , 2010 .
a. somogyi , c. kewish , f. polack , and t. moreno , `` the scanning nanoprobe beamline nanoscopium at synchrotron soleil , '' in _
the 10th international conference on x - ray microscopy _ , vol . 1365 , pp .
5760 , 2011 .
a. somogyi , k. medjoubi , c. kewish , v. le roux , m. ribbens , g. baranton , f. polack , and j. samama , `` status of the nanoscopium scanning nanoprobe beamline of synchrotron soleil , '' in _ spie optical engineering applications _ , pp .
885104885104 , 2013 .
|
short period , high field undulators are used to produce hard x - rays on synchrotron radiation based storage ring facilities of intermediate energy and enable short wavelength free electron laser .
cryogenic permanent magnet undulators take benefit from improved magnetic properties of @xmath0 ( rare earth based magnets ) at low temperatures for achieving short period , high magnetic field and high coercivity .
using @xmath1 instead of @xmath2 , which is generally employed for undulators , avoids the limitation caused by the spin reorientation transition phenomenon , and simplifies the cooling system by allowing the working temperature of the undulator to be directly at the liquid nitrogen one ( 77 k ) .
we describe here the development of a full scale ( 2 m ) , 18 mm period @xmath1 cryogenic permanent magnet undulator ( u18 ) . the design , construction and optimization , as well as magnetic measurements and shimming at low temperature are presented .
the commissioning and operation of the undulator with the electron beam and spectrum measurement using the nanoscopmium beamline at soleil are also reported .
|
high resistive ( @xmath0 @xmath1 10@xmath2 - 10@xmath3 @xmath4 cm ) bakelite rpcs @xcite , prepared with a silicone compound coating on the inner electrode surfaces , were studied in a cosmic ray test bench to characterize its long term stability of operation @xcite .
need for the silicone compound coating had been illustrated through the measurement of variation of efficiency with applied high voltage ( hv ) in our earlier work @xcite .
study of variation of time resolution and cross - talk between neighbouring pick - up strips with applied hv were also reported earlier @xcite .
such high resistive electrode - based rpcs are being explored as active elements of a large scale iron calorimeter for the proposed india - based neutrino observatory ( ino ) @xcite .
one of the limitations of the streamer mode operation of high resistive bakelite rpcs is that the rate handling capability is relatively worse than that in the avalanche mode .
for the same energetic particle falling on the rpc , the charge content contributing to the resulting pulses in the two modes differ by orders of magnitude .
this , as well as the physical extent of the streamer and the electron avalanche , may result in variation of the pulse timing properties .
these aspects need to be studied both through experiment and through simulation .
stability of operation of the rpcs and its dependence on the detector parameters are the other important aspects which need to be studied in detail .
it is established that a rough inner electrode surface of an rpc is prone to cause field emission , which is a source of high dark current @xcite . in this context , it is relevant to investigate the effect of surface roughness on the detector dynamics .
it is conceivable that imperfections on the bakelite surface may be responsible to cause local non - uniformity of the electric field . depending upon their shape and size
, they may often lead to discharge severely degrading the performance of the device . in this article ,
some experimental results on operation of our bakelite rpc in the avalanche mode , systematic studies and comparison of charge contents of the resulting pulses in the avalanche mode and the streamer mode operation of the same rpcs are reported . in our endeavor of corroboration of experimental results with simulation , preliminary results of numerical simulation of the effect of roughness of the resistive bakelites on the field configuration
are also reported . using garfield - nebem @xcite code ,
the field calculation has been done for several models designed on the basis of measurements of the bakelite surface profiles @xcite .
all the rpcs were tested in a cosmic ray test bench described in ref .
the cosmic ray telescope was constructed using three scintillators , two placed above the rpc and one below .
triple coincidence of the signals , obtained from these three scintillators was used to select a cosmic ray event(master trigger ) .
the signals obtained from the pick - up strips were put in coincidence with the master trigger obtained as above .
this is referred to as the coincidence trigger .
finally , the efficiency was calculated as the ratio of the coincidence trigger to the master trigger .
premixed gas of argon , isobutane and tetrafluroethane(r-134a ) in the volume ratio of 55/7.5/37.5 ( equivalent to a mass mixing ratio of 34/7/59 ) was used in the streamer mode , while in the avalanche mode , r-134a / isobutane in 95/5 , r-134a / isobutane / sf@xmath5 in 95/4.5/0.5 and 95/2.5/2.5 volume mixing ratio were used . a typical flow rate of 0.4 ml per minute was maintained by the gas delivery system @xcite , resulting in @xmath1 3 changes of gas gap volume per day .
charge content of the pulses were measured using a charge to digital converter , referred to as qdc . the set - up
is shown in fig .
[ fig:1 ] , which is very similar to that used in ref .
the master trigger gate of width 1.5 @xmath6s was also used as gate to the qdc .
it was observed that the charge spectrum and the measured mean charge were independent of gate width over a range of 150 ns to 1.5 @xmath6s .
the pulse height in the streamer mode was noted to be a few hundred mv , while that in the avalanche mode was found to be @xmath7 10 mv .
a @xmath8 voltage pre - amplifier was used for the avalanche mode operation of the rpcs .
schematic representation of the charge spectrum measurement setup .
phillips scientific 7186 charge to digital converter ( qdc ) was used.,title="fig : " ] +
in the avalanche mode , the rpcs were tested using different gas mixtures mentioned as above . the efficiency and time resolution of the rpcs
were studied by varying the applied hv .
the efficiency plateau was obtained at > 90% for all the gas mixtures , and it was found to be marginally higher for gas mixtures containing less amount of sf@xmath5 . at the plateau region , the time resolution was found to be @xmath1 2.5 ns(fwhm ) and the average signal arrival time decreased with the increase of hv , which is a common feature of any gas filled detector . the charge spectrum of the rpc operated in the avalanche mode using the gas mixture of r-134a / isobutane / sf@xmath5 in 95/4.5/0.5 ratio , at two different high voltages.,title="fig : " ] + [ ] the charge distribution spectrum in the streamer mode with a gas mixture of argon / isobutane / r-134a in 55/7.5/37.5 ratio.,title="fig : " ] + in fig .
[ fig:2 ] ( a ) , a typical charge spectrum of the rpc operated in the avalanche mode ( hv @xmath9 kv ) using the gas mixture of r-134a / isobutane / sf@xmath5 in 95/4.5/0.5 ratio , together with the polya distribution fitting curve is shown . at higher hv values
, a streamer peak appears after the avalanche as shown in fig .
[ fig:2 ] ( b ) . the charge distribution spectrum in the streamer mode with a gas mixture of ar , isobutane and r-134a in 55/7.5/37.5 ratio
is shown in fig .
[ fig:3 ] . corresponding best - fit polya distribution curve to the streamer charge spectrum
is also shown in the same plot .
it is clear that the fitted curve underpredicts the pulse charge content at the higher charge tail of the spectrum .
this possibly indicates the onset of a breakdown mechanism responsible for the excess streamer charge beyond that can be predicted by the clustering model . the induced charge as a function of the applied hv in the avalanche mode is shown in fig . [ fig:4 ] , whereas that in the streamer mode is shown in fig .
[ fig:5 ] . for the rpc ,
when operated in the avalanche mode with the gas mixtures of r-134a / isobutane in 95/5 and r-134a / isobutane / sf@xmath5 in 95/4.5/0.5 ratio it was observed that at higher voltages a substantial streamer peak appears . but the charge shown in fig .
[ fig:4 ] is the mean charge of the avalanche peak only .
it is clear from the figures that the charge saturates at @xmath1 1 - 2 pc in the avalanche mode for all the gas compositions , while in the streamer mode , the charge is @xmath1 100 pc . the induced charge as a function of hv for a silicone coated rpc in the avalanche mode using three different gas mixtures.,title="fig : " ] + the induced charge as a function of hv for a silicone coated rpc in the streamer mode with a gas mixture of argon / isobutane / r-134a in 55/7.5/37.5 ratio.,title="fig : " ] +
a preliminary calculation of the electric field to simulate the effect of surface roughness in bakelite rpc has been done as a proof of concept .
the surface profile of the bakelite sheet ( grade p120 ) has been studied using a dektak 117 profilometer which sampled 2500 times over a small span of 5 cm at three arbitrary locations .
[ fig : linfit ] shows the measurement taken at one of these locations where a least square fit has been done in order to define the baseline of the surface . in fig .
[ fig : surfacedata ] the residuals of the data with respect to the baseline are depicted as well to indicate the amplitudes of the roughness . a typical surface profile of bakelite p120 measured with dektak 117 profilometer .
the least square fit to the data and the residuals w.r.t the fit are shown.,title="fig : " ] + the amplitudes of one such measurement is represented in histogram as shown in fig .
[ fig : amplhist ] . it can be found that the maximum amplitudes of the roughness in positive ( upward ) and negative ( downward ) directions can be of the order of 4 @xmath6 m and 2 @xmath6 m , respectively .
[ set3c]m.,title="fig : " ] for studying the effect of roughness , a device has been modeled following the geometry of an rpc .
the device comprises of two bakelite slabs , each with thickness 2 mm ( in z - direction )
. a gas layer of the same height is interposed between . a fine layer of graphite with thickness 20 @xmath6 m has been considered on other sides of the bakelites .
the overall dimension of the rpc has been taken to be 5 mm ( in x - direction ) @xmath10 50 mm ( in y - direction ) .
[ fig : rpcmodel ] depicts the model described above .
the model rpc considered in the calculation.,title="fig : " ] + the roughness has been added to one of the bakelite surface by building up a tiny structure with dimensions corroborated by the measurements .
two such structures have been introduced to represent the surface roughness with positive and negative amplitudes . for the preliminary calculation ,
the shape of the roughness has been considered as a rectangular ridge stretched along the whole length of the device in y - direction , i.e. 50 mm , with height either 4 @xmath6 m ( in positive z - direction ) or 2 @xmath6 m ( in negative z - direction ) .
the cross - section of the base of the ridge has been taken to be 50 @xmath6 m @xmath10 50 mm .
the electric field ( z - component ) has been evaluated at several positions with respect to the top edge of the structure for the positive roughness as shown in fig .
[ fig : fieldposrough ] . it can be noticed that a change of about 12% in the field value is possible within 1 @xmath6 m from the structure .
the distortion in the field due to the structure diminishes with the distance and it takes more than 100 @xmath6 m to get nullified .
variation of electric field with distance away from the positive roughness.,title="fig : " ] + the same calculation has been carried out for the model with negative roughness .
the results are illustrated in fig .
[ fig : fieldnegrough ] . here
the maximum change is about 5% at the closest vicinity of 1 @xmath6 m from the bakelite surface which is 3 @xmath6 m away from the base of the roughness ridge .
it may inflate upto approximately 7 - 8% when studied at just 1 @xmath11 m above the base . here
, obviously the effect of the roughness dies down within 100 @xmath11 m from the bakelite surface .
variation of electric field with distance near the negative roughness.,title="fig : " ] +
systematic studies on the performances of the silicone coated bakelite rpcs were performed both in the streamer and the avalanche mode .
an efficiency > 90% and time resolution @xmath1 2 ns ( fwhm ) were obtained in both the modes of operation .
the measured induced charges on the rpc pick - up strips were found to be 1 - 2 pc in the avalanche mode , and @xmath1 100 pc in case of the streamer mode of operation .
the preliminary calculation proves that the minute structures with straightforward shape on the bakelite surface do affect the field configuration by 5 - 12 % depending on its height or depth in its close proximity .
it indicates that the structures with different shapes and sizes may be significant in distorting the nearby field configuration .
a detailed study on the roughness structures is therefore required , corroborated by precise measurements to advance the investigation on the effect of roughness on the performance of the rpc .
we acknowledge the service rendered by mr .
ganesh das of vecc for meticulously fabricating the detectors and mr .
chandranath marick of sinp for making some of the related electronic circuits .
50 r. santonico , r. cardarelli , nucl .
inst . and meth . 187 ( 1981 ) 377 .
s.biswas , et al . , nucl . instr . and meth . a 602 ( 2009 ) 749 .
s.biswas , et al . , nucl . instr . and meth . a 617 ( 2010 ) 138 .
s.biswas , et al . , nucl . instr . and meth . a 604 ( 2009 ) 310
ino project report , ino/2006/01 , june 2006 , @xmath12http://www.imsc.res.in/@xmath1ino/@xmath13 .
changguo lu , nucl .
instr . and meth . a 602 ( 2009 ) 761 .
s.bose , et al . ,
instr . and meth . a 602 ( 2009 ) 839 .
c. lu , snic symposium , stanford , california -3 - 6 april 2006 .
|
performances of several single gap ( gas gap 2 mm ) prototype resistive plate chambers ( rpc ) made of high resistive ( @xmath0 @xmath1 10@xmath2 - 10@xmath3 @xmath4 cm ) bakelite , commercially available in india have been studied in recent times . to make the inner electrode surfaces smooth ,
a thin coating of silicone has been applied .
an efficiency > 90% and time resolution @xmath1 2 ns ( fwhm ) have been obtained for both the streamer and the avalanche mode .
the induced charge distributions of those silicone coated rpc are studied and the results are presented .
a numerical study on the effect of surface roughness of the resistive electrodes on the electric field of the device has been carried out using garfield - nebem code .
a few results for a simplified model representing surface roughness , measured using a surface profilometer for the bakelite surfaces , have also been presented . , , , , , , , , rpc ; streamer mode ; avalanche mode ; bakelite ; cosmic rays ; silicone ; charge distribution ; garfield - nebem ; roughness 29.40.cs
|
many galaxies at @xmath5 have morphologies that are very different from the galaxy population at @xmath7 ( lotz et al . 2004 ;
lotz et al . 2006 ; law et al . 2007 ; scarlata et al . 2007 ; elmegreen et al . 2007
; pannella et al .
however , the morphologies of these objects only tell one part of the story . at @xmath8 ,
rest - frame uv wavelengths , which are most impacted by dust obscuration and star formation , can be traced by high resolution optical imaging ( toft et al . 2007 ;
overzier et al .
2010 ; cameron et al .
the location and distribution of stellar populations within these galaxies , particularly sites of active star formation , is likely to be affected by various physical and dynamical mechanisms which are present in earlier epochs in the galaxy s evolution .
the pixel-@xmath0 technique ( conti et al . 2003 ; welikala et al .
2008 ; welikala et al .
2009 ) can measure spatially resolved stellar populations for large samples of @xmath5 galaxies using their multi - band hst images . with this approach , one can study asymmetries that exist in the star formation , age and dust distribution across these galaxies . in parallel with the detailed kinematical studies of these objects from spatially resolved spectroscopy ,
this will enable insights into the formation mechanisms of these galaxies .
one such mechanism that could be responsible for the formation of these clumps on kiloparsec scales is disk fragmentation in toomre - unstable gas - rich disks ( elmegreen et al .
2007 , elmegreen et al . 2009 ) .
there is now direct kinematic evidence that high - z disks are much more turbulent than their counterparts at @xmath9 ( bournaud et al . 2007 ;
bournaud et al .
2008 ; genzel et al . 2008 ; genzel et al .
for example , high - redshift disks have been found to have local intrinsic gas velocity dispersions of @xmath10kms@xmath11 as well as high gas - to - total baryonic mass fractions ( daddi et al .
2010 ; tacconi et al .
numerical simulations of gas - rich turbulent disks indicate that massive kpc - sized clumps can form in - situ through gravitational instabilities , a phenomenon known as a ` clump - cluster ' phase .
( noguchi et al . 1999 ; immeli et al .
2004a , b ; bournaud et al . 2007 ; elmegreen et al .
2008 ; dekel , sari , & ceverino 2009 ; agertz et al . 2009 ;
ceverino et al .
2010 ; aumer et al .
2010 ; genzel et al .
according to these simulations , the clumps can migrate towards the gravitational center as a result of both their mutual interactions and dynamical friction against the host disk , so that they can eventually coalesce into a young bulge on the order of a few dynamical times .
alternatively , these high - redshift clumpy disk and irregular galaxies may form via mergers .
in particular , numerical simulations of gas - rich mergers result in remnant disks that show a low velocity field asymmetry that also satisfies the criteria necessary to be classified as a high - redshift disk galaxy observationally .
for example , robertson & bullock ( 2008 ) compared one of the merger remnants to the bulk properties of the well - studied @xmath12 galaxy bzk-15504 and showed that it has a star formation rate ( sfr ) , gas surface density , and a circular velocity - to - velocity dispersion ratio that is in excellent agreement with bzk 15504 .
such numerical simulations suggest that gas - rich mergers can play a prominent role in the formation of disk galaxies at high redshift .
recent and upcoming larger ifu programs such as the sins survey ( frster - schreiber et al . 2006 ,
genzel et al . 2006 ) , the sauron project ( bacon et al .
2001 ) and the massiv survey ( contini et al . 2011 ; epinat et al . 2011
; queyrel et al .
2011 ) are expected to shed further light on these formation mechanisms by performing detailed studies of the gas kinematics of @xmath13 galaxies .
other large ifu programs include the califa project ( sanchez et al .
2011 ) , which will target approximately 600 objects with pmas ifus on the calar alto telescope , and the sami program ( croom et al .
2011 ) which is a planned low redshift survey of serveral thousand galaxies .
in addition to kinematic studies , a complementary study of the spatial distribution of the stellar populations within these galaxies can provide additional insights into whether these objects are formed through mergers or are , in fact , disks formed in - situ .
frster - schreiber et al .
( 2011a , b ) used a combination of near - infrared integral field spectroscopy from the sinfoni instrument on the vlt , combined with deep high resolution hst nic2/f160w imaging of six @xmath13 star - forming galaxies to characterize the properties of kpc - scale clumps and their contribution to the rest - frame optical emission . in our next papers
, we aim to extend these previous studies and make detailed spatially - resolved measurements of the properties of large samples of disk galaxies .
these studies will also make use of new fully reduced nir data from the hst wide field camera 3 ( wfc3 ) which will become available for the goods - south field in early 2012 .
the relation between galaxy environment and galaxy properties have been extensively studied in the local universe ( gomez et al .
2003 , blanton et al .
2005 , hogg et al .
a number of studies have observed that the high-@xmath0 ( @xmath14 ) sfr - density relation is either reversed or weaker at @xmath15 than that seen locally ( cucciati et al . 2006 ; cooper et al . 2007 ; elbaz et al .
2007 ; ilbert et al . 2006 ; poggianti et al . 2006 ; ideue et al . 2009 ;
salimbeni et al . 2009 ; scodeggio et al . 2009 ; tran et al .
2010 ; grtzbauch et al .
these examples are consistent with the scenario that galaxies in dense environments form stars rapidly at early times , quickly building up mass and becoming quiescent , while galaxies in less dense environments form stars at a more sedate pace but over longer timescales , a phenomenon we refer to as _ in - situ _ evolution ( wijeshinge et al .
this is distinct from the `` infall and quench '' scenario where the properties of galaxies are impacted directly by their local environment through physical mechanisms such as ram - pressure stripping and galaxy harassment .
in addition , peng at al .
( 2010 ) makes a distinction between this type of quenching which is directly related to the environment of galaxies and _ mass quenching _ which dominates at high masses and early cosmic times . a few recent studies have , however , continued to observe the same sfr - density relation upto @xmath16 as seen locally ( patel et al .
a few studies have investigated the relation beyond @xmath16 .
quadri et al .
( 2011 ) showed that galaxies with quenched sf tend to reside in dense environments out to at least @xmath17 and that the the sfr - density relation holds even at fixed stellar mass .
their density estimates , however , have relatively high uncertainties as they are derived from photometric redshifts .
while the sfr - density relation at high redshifts is a matter of some debate , it is clear that an _ in - situ _ evolution makes a distinct prediction compared to the `` infall and quench '' models .
the latter suggests that galaxies in dense environments should show a sfr distribution that is progressively suppressed from the outside in , as the outer regions are those which will be affected first by their rapidly changing environment .
the former , on the contrary , suggests that the suppression should either happen uniformly as a galaxy ages , or that the inner regions should be suppressed first . by studying the spatial distribution of sfr in star - forming galaxies as a function of environment
, we should be able to distinguish clearly between these two scenarios .
studies by welikala et al .
( 2008 ; 2009 ) at @xmath18 confirm the global sfr - density relation observed locally but also show that the sfr at @xmath19 0.1 is centrally concentrated relative to the outskirts , and in more dense environments it is this central sfr that is supressed , favouring the _ in - situ _ evolution scenario .
park et al . ( 2007 ) also studied the color gradients of galaxies as a function of the local galaxy density and galaxy morphology in the sdss .
they found no environmental dependence of the color gradient for early - type galaxies at a given luminosity , and only a weak dependence on environment for faint late - type galaxies which are observed to become bluer in their outskirts ( relative to the galaxy center ) in low density environments .
it thus becomes extremely interesting to determine whether the radial variation of star formation with environment seen locally is also in place at high redshifts .
the development of mos ifu instruments and surveys with large photometric and spectroscopic samples , such as the galaxy and mass assembly ( gama ) survey ( driver et al .
2011 ) , will allow such a statistical analysis for a large sample of galaxies .
there are two complementary approaches to this .
while spatially resolved spectroscopy provides detailed measures of the current spatial distribution of physical properties within galaxies , pixel-@xmath0 adds the ability for the star formation history to be probed .
while pixel-@xmath0 also provides the current sfr and other properties of the stellar populations in subcomponents of galaxies , these measurements will have higher uncertainties than spectroscopic measures , and can not provide detailed kinematic information that spectroscopic observations can .
however , until the advent of highly - multiplexed mos ifus , even ifu surveys will be limited to several thousands of targets , while pixel-@xmath0 can be applied to tens of thousands and up to millions of objects in order to determine the radial variation of stellar population parameters within them and the evolution of this radial variation with redshift .
we therefore aim to extend studies performed by welikala et al .
( 2008 ; 2009 ) which applied the pixel-@xmath0 method to @xmath20 galaxies at @xmath21 in the sdss and determined the radial variation of star formation for this sample , the scatter in this relation , and its dependence on galaxy environment and galaxy morphology .
there are , however , inherent systematic effects that may limit the utlity of pixel-@xmath0 in probing spatially resolved stellar populations in galaxies .
the current analysis explores these in order to quantify the impact of such systematics on the measurements made by pixel-@xmath0 .
a number of studies have sought to extract stellar populations for spatially resolved colors .
these compare the colors in individual pixels to a library of seds generated by stellar population synthesis models . the majority of these utilised small samples of galaxies at @xmath9 and @xmath15 ( abraham et al . 1999
; johnston et al . 2005 ; kassin et al . 2003 ; lanyon - foster et al . 2007 ; zibetti et al .
2011 ) . given a multi - band image of a galaxy , the pixel-@xmath0 method ( conti et al . 2003
; welikala et al . 2008 , 2009 )
computes the stellar population properties such as stellar population age , sfr e - folding timescale @xmath22 , dust obscuration through @xmath23 and metallicity @xmath24 in each pixel in the galaxy , as well as the associated uncertainties in these parameters .
it does this by taking a library of seds generated from stellar population synthesis ( sps ) codes , redshifting each of these seds and convolving them with the available passbands .
the simulated fluxes are then compared to the observed fluxes in each pixel and a best - fitting sed is computed for that pixel .
the statistical errors in each parameter are calculated by marginalizing the likelihood function from the fit over the remaining parameters , and are conservative .
the pixel-@xmath0 approach makes two fundamental assumptions : * * independently evolving pixels : * the pixels in the galaxy image are assumed to be evolving independently of each other .
this ignores any mixing of stellar populations between neighboring pixels which is expected to occur in real galaxies .
neighbouring pixels will also not truly be independent , due to a finite point spread function ( psf ) , and this effect is also neglected . * * single stellar populations with exponentially declinining sfh : * the stellar population underlying each pixel is assumed to be one whose spectrum is derived from integrating over single stellar populations ( ssps ) of a given age , weighted by a sfr which is characterized by an initial burst followed by an exponential decrease in sfr .
the decrease is parameterized by a range of allowed values of e - folding times @xmath22 . in reality
, galaxies are expected to have more complicated sfhs , including multiple bursts .
in addition to these two fundamental assumptions , there is a limitation in the form of degeneracies among the sed models which could impact the accuracy with which pixel-@xmath0 determines stellar population parameters in the pixels .
sps models show that age , metallicity and dust all tend to affect spectra in similar ways ( bruzual & charlot 2003 ) .
multiple seds each generated for different populations ( different ages , metallicites and sfhs ) can , as a result , still predict very similar colors of a galaxy or its subcomponents .
some of these degeneracies arise from well known physical correlations , such as those between age and metallicity of the stellar population .
worthey ( 1994 ) constructed detailed models of old and intermediate stellar populations and their absorption indices to show that if @xmath25/@xmath26 for two populations , they would appear identicial in most absorption indices .
the age - metallicity degeneracy can also affect the photometric evolution of the ssps since optical and near - infrared ( nir ) colors reflect the relative contribution of hot and cool stars to the integrated light .
bruzual & charlot ( 2003 ) followed the evolution of optical and nir colors and the stellar mass - to - light ratio @xmath27 in their model , and found that at fixed age , increasing metallicity tends to redden the colors and increase the @xmath27 ratio .
this is explained by the fact that increasing the metallicity ( at fixed stellar mass ) causes ( a ) stars to evolve at lower effective temperatures and lower luminosities ( girardi et al .
2000 ) and ( b ) changes the relative number of red and blue supergiants , which strongly impacts the color evolution of the ssp
. they also found that increasing metallicity at fixed age had a similar effect to increasing age at fixed metallicity .
the age - metallicity degeneracy is therefore inherent in the ssp model used .
there could also be other degeneracies , including those between dust obscuration and metallicity as well as between age and dust obscuration .
the redshift - evolution of the obscuration - metallicity degeneracy in the pixels was characterized by conti et al .
this work does not test the two fundamental assumptions above . as for degeneracies , exploring the different correlations between the parameters on a pixel - to - pixel basis and propagating
the resulting errors are beyond the scope of this paper .
welikala et al .
2008 characterized some of these correlations using the likelihood function from the fit to the pixels in sdss galaxies .
since the severity of the degeneracies depend on the number and type of available broadband colors , the sps model and the chosen grid of stellar population parameters , we focus on these systematics in this paper . the validity of both the two assumptions and the impact of degeneracies will be explored in a future paper that compares the spatial distribution of stellar populations inferred from spatially resolved spectroscopy obtained from large ifu programs ( croom et al .
2011 ) with those inferred from applying the pixel-@xmath0 technique to the same galaxies .
such a detailed comparison with ifu programs will enable a robust calibration and testing of pixel-@xmath0 , leading eventually to a possible refinement of these assumptions . in this paper , we focus on the following systematics that could potentially bias our results : * * the sps model : * an inevitable bias is the choice of the stellar population synthesis model .
different stellar population synthesis models predict different colors .
in particular , there are significant differences in the prescriptions for thermally pulsating asymptotic giant ( tp - agb ) stars among the following models : maraston 2005 ( m05 ) , bruzual & charlot 2003 ( bc03 ) and charlot & bruzual 2007 ( cb07 ) . in this paper ,
we compare the outcomes of using these different models on a pixel - by - pixel basis , for all galaxies in the sample . similar comparisons of population synthesis models have been performed previously on the integrated colors of galaxies but these have focused primarily on the effect on their stellar mass ( maraston et al .
2006 ) . in this work ,
we focus on the effect of these differences on the age , sfr e - folding time and obscuration estimated in the pixels . *
* the available passbands : * the accuracy with which the pixel-@xmath0 parameters are determined in the pixels depends on the the available passbands .
this paper explores the systematic effects arising from passband limitations , using both optical and near - ir colors . * * the parameter space searched by pixel-@xmath0 and the number of seds per model : * pixel-@xmath0 searches a grid of ages , e - folding times , dust attenuation values and metallicities . in general , a large as possible range of values is chosen while physically unmeaningful values will be associated with high statistical uncertainties . despite this
, a large parameter space can result in degeneracies among the various seds searched by pixel - z which in turn may bias the pixel-@xmath0 estimates .
some priors are already built in to stop pixel-@xmath0 searching unphysical parameters e.g. , the age of a stellar population in a pixel can not exceed the age of the universe at a given redshift . here , we explore the systematic error arising from increasing the range of parameter values allowed on the pixel-@xmath0 grid .
in particular , we focus on the impact of increasing the metallicity range .
this has two predicted effects : ( a ) it increases the severity of the age - metallicity degeneracy and would bias the pixel-@xmath0 estimates in quantities such as age , @xmath22 and dust and ( b ) it introduces an uncertainty associated with the models : while spectral synthesis models have been well established for solar metallicity stellar populations in optical photometry , the models are less well established for very sub - solar or super - solar populations . increasing
the metallicity range will test the impact of both these effects on the derived pixel-@xmath0 parameters . *
* global sed - fitting versus pixel-@xmath0 : * we examine any biases resulting from measuring stellar population quantities , particularly the sfr , in individual pixels compared to the same quantities derived from aperture photometry and sed - fitting .
these biases arise because individual pixels can have different colors to the integrated color of the galaxy
. this bias could give rise to a difference between ( 1 ) the mean properties of the galaxy obtained by integrating the fluxes in an aperture and sed - fitting to the total fluxes and ( 2 ) the mean properties derived from fitting to the individual pixels within the same aperture and summing the resulting fits .
we use a magnitude - limited ( @xmath28 ) sample of galaxies in the goods - south survey ( giavalisco et al . 2004 ) with imaging in four acs bandpasses , @xmath29 , @xmath30 , @xmath31 and @xmath32 .
we refer to the full set of these four acs filters as @xmath1 throughout this paper .
we make use of the version 2.0 of the publicly available acs source catalog as well as a photometric redshift galaxy catalog for cdfs ( cardamone et al .
cutout images of each objects are obtained from the reduced , calibrated , stacked and mosaiced acs images .
the pixel scale in these final drizzled images is @xmath33 . in the pixel-@xmath0 analysis ,
the redshift in each pixel is fixed to that of the host galaxy and we fit only to the sed type . using spectroscopic redshifts
provides a more accurate determination of the stellar population properties in each pixel , compared to using photometric redshifts .
we thus select only galaxies which have been spectroscopically measured in the goods - vimos campaign ( popesso et al .
2009 ; balestra et al .
goods - vimos consists of two surveys which target galaxies in different redshfit ranges .
the vimos low resolution blue ( lr - blue ) is aimed at observing galaxies mainly at @xmath34 while the medium resolution ( mr ) orange grism is aimed mostly at galaxies at @xmath35 and lyman break galaxies ( lbgs ) at @xmath36 .
our sample has measured spectroscopic redshifts from both grisms .
a final selection is done on galaxy size to ensure that there are a sufficient number of pixels for the pixel-@xmath0 analysis .
galaxies are chosen to have a half light radius @xmath37 according to simulations in welikala et al .
( 2011 ) that estimate the minimum galaxy sizes needed to recover color gradients within galaxies with a fractional error smaller than 0.1 .
this implies that the half light diameter should be at least 20 pixels across , corresponding to 5.5 times the psf fwhm .
this leaves us with a sample of 467 galaxies with four - band imaging ( @xmath1 ) . out of this sample
, we select a subsample of 46 objects with nir colors which have been detected by the goods - nicmos survey ( bouwens et al .
2011 ; magee et al 2007 ; conselice et al .
this provides deep ( @xmath38 mag at @xmath39 ) nicmos data with the nic3 camera over the goods north and south fields .
these studies have found several ` first - light ' ( @xmath40 ) galaxy candidates but the survey has also detected several @xmath5 galaxies that we use in this work .
the redshift distribution of both the optical and the nir sample is shown in figure [ fig : redshift_distribution ] .
the median redshift of the optical sample is @xmath41 , while that of the nir sample is @xmath42 . for the optical - only sample ,
there are 145 galaxies in @xmath43 and 77 galaxies in @xmath44 . for the nir sample
, there are 22 galaxies in @xmath45 , 12 galaxies in @xmath43 and one object found at @xmath46 . using the sersic indices ( @xmath47 ) for these objects obtained from a publicly available morphology catalog for the goods - south field , we find that the optical - only sample consists of approximately 4 times more late - type galaxies compared to early - types : 299 late - type galaxies ( @xmath48 ) and 81 early - type galaxies ( @xmath49 ) .
this study is not concerned with tracing a particular population of galaxies , which would require a volume - limited sample of objects .
rather , we investigate systematics of a particular technique which can be applied to galaxies in a range of redshifts , luminosities , morphological types and inclinations as long as they are well - enough resolved for the pixel-@xmath0 analysis . nevertheless , we repeat our analysis on subsets of the full galaxy sample in order to test the robustness of the results against selection effects . finally , we use preliminary nir data from the first 6 epochs of the candels survey ( grogin et al . 2011 ; koekemoer et al .
2011 ) using the wfc3 @xmath50 and @xmath51 filters in the goods - south field , in order to illustrate the power of combining this high resolution optical and nir data from acs and wfc3 with the pixel-@xmath0 method in order to probe substructure in @xmath5 galaxies and to determine the properties of the stellar populations in the pixels .
a more thorough investigation of substructure in galaxies using the final candels data product will be performed in a future paper .
acs and nicmos science images are in counts / sec , while the weight maps are inverse variance maps ( i.e. , @xmath52 ) .
flux calibration of the acs and nicmos science and weight postage stamp images of each source is peformed using the acs and nicmos zeropoints .
these contain the background variance from the sky noise , read - out noise and dark currents .
further , because the pixels are correlated , the actual noise is higher than the theoretical noise in the weight files ( casertano et al .
we account for this by computing the noise in empty patches ( whose size is larger than the correlation length scale ) within the science image and we calculate the ratio of this measured noise value to the theoretical noise value .
the noise in the weight images is then rescaled by this ratio to give a more accurate estimate of the noise in the pixels due the background and instrument . the poisson uncertainty from pixels in the source itself is then added in quadrature .
these final acs flux and variance images are resampled to the coarser 0.1 arcsecond pixel scale of the nic3 images using a lanczos4 kernel .
we then perform psf - matching of the acs images to the nic3 images using the
_ _ ip_diffim _ _ image mapping software currently in the pipeline of the large synoptic survey telescope ( lsst ) .
the _ ip_diffim _
code is an implementation of the higher order tranform of psf and template subtraction code ( _ _ hotpants _ _ ) , and uses the algorithm of alard ( 1999 ) and alard & lupton ( 1998 ) to find the psf - matching kernel that maps the input ( resampled ) acs images to the nic3 ones .
the kernel is decomposed into a linear set of basis functions , which are gaussians of varying fwhm .
the process is linear and ends up matching the psfs of the two input images .
a similar process is used to psf - match the variance images .
finally , for the analysis on the candels data , the wfc3 images from the first 6 hst epochs are stacked by performing a weighted sum of the images in the different epochs and dividing by the sum of the weight maps in order to increase the signal - to - noise in the final flux images .
the existing acs images are resampled to the pixel scale of the wfc3 ones and psf - matching is performed using the same method . the pixel-@xmath0 method ( conti et al .
2003 ; welikala et al .
2008 ; welikala et al .
2009 ) fits seds from a library of stellar population synthesis models to colors of individual pixels .
the redshift is fixed to the spectroscopic redshift of the galaxy , and the fitting is done for the sed only .
a best - fitting sed is identified for each pixel using a maximum likelihood method and an age , @xmath22 , dust obscuration and metallicity is then inferred for each pixel .
rms errors are found for each of these parameters by marginalizing the likelihood function in each pixel over the remaining parameters .
a signal - to - noise threshold @xmath53 in the @xmath54 band is set to ensure that the fitting is done to pixels in the galaxy and not to the sky pixels .
low snr pixels in the outskirts of a galaxy and in any residual sky pixels that have been artificially fitted generally have the largest uncertainties in their pixel-@xmath0 parameters .
we examine three distinct stellar population synthesis models which are currently widely used : bruzual & charlot 2003 ( bc03 ) , charlot & bruzual ( cb07 ) and maraston 2005 ( m05 ) . for each model , we generate spectra derived from single stellar populations with an exponentially declining star formation history . in order to test differences between the codes , the parameter space has to be made identical ( or if not , as similar as possible ) . the grid of stellar population parameters ecompasses 2178 seds : * 11 values of stellar population age : this is the time elapsed since the intial burst of stars
. the stellar populations range from extremely young ( 0.001 , 0.01 , 0.1 , 0.5 gyr ) , to middle age ( 1.0 , 3.0 , 5.0 gyr ) to very old ( 7.0 , 9.0 , 11.0 , 12.0 gyr ) . in addition
, the pixel-@xmath0 algorithm rejects ages which are greater than the age of the universe at the redshift of each galaxy . * 11 values of @xmath22 : 0.1 , 0.25 , 0.50 , 0.75 , 1.0 , 1.5 , 2.0 , 3.0 , 4.0 , 5.0 , 10.0 gyr . * 6 values of @xmath23 from the calzetti al .
1998 attenuation curves : 0.0 , 0.1 , 0.2 , 0.3 , 0.5 , 0.9 .
* 3 values of metallicity z : 0.01 ( 0.5 @xmath55 ) , 0.02 ( @xmath55 ) and 0.04 ( 2 @xmath55 ) . in the case of metallicity , stellar populations generated from the m05 models allow only 3 values .
we have adopted 3 values of z across all the models compared here ( in bc03 and cb07 , these are 0.008 , 0.02 and 0.05 ) . to test the effect of this restricted metallicity range
, we perform a control test over the entire population of galaxies and all the associated pixels , using the bc03 models .
the allowed number of metallicities is increased to include significantly sub - solar values ( @xmath56 ) .
the resulting systematic error , i.e. , the difference in each pixel - z parameter ( age , @xmath22 , @xmath23 ) between using 6 values and 3 values of @xmath24 , is computed pixel - by - pixel , and we examine the distribution of this systematic acrosss the entire galaxy sample .
the effect of this relatively restricted range of metallicity is that the systematic difference in the @xmath24 values recovered from pixel-@xmath0 between the models is usually no more than 1 @xmath57 , indicating that the majority of pixels in the galaxy sample are described by metallicites which are close to solar .
permitting only metallicities that are close to solar is therefore a reasonable approximation .
the effect on the other parameters ( age , @xmath22 and dust obscuration ) is examined below .
this control test is also a test of the impact of degeneracies in the models used .
increasing the input metallicity range would be expected to increase the severity of these degeneracies when fitting to the pixel colors , and therefore increase the systematic error in the pixel-@xmath0 parameters .
the results of this test are presented in section [ subsec : paramspace ] . in the tests below
, we compute the ratio of the systematic error in a particular pixel-@xmath0 quantity ( age , @xmath22 , @xmath23 ) to the statistical error in that quantity , for each pixel in the galaxy : * when testing the effect of the stellar population synthesis models , we measure for each pixel and for each parameter @xmath58 the quantity @xmath59 , where : @xmath60 where @xmath61 and @xmath62 are the 2 models being compared and @xmath63 is the rms uncertainty in the parameter @xmath58 for that pixel . * when testing the effect of limiting the available passbands , we measure for each pixel and for each parameter @xmath58 the quantity @xmath64 + where @xmath65 and @xmath66 are the filter set used . *
the systematic error resulting from expanding the metallicity range from 3 to 6 values is computed similarly . for each galaxy in the sample ,
we measure @xmath59 within @xmath67 , considering only pixels with @xmath68 to remove sky pixels .
since the effect of each systematic could vary from galaxy to galaxy , we are interested in the distribution of a particular systematic error across the whole galaxy sample . for each test of a particular systematic , we plot the distribution of @xmath59 for all the pixels inside @xmath67 for all the galaxies in the sample . with such statistics , we are able to determine conclusively the effect of the particular systematic being tested .
we check how robust the results are against selection effects by repeating the analysis on several independent subsets of the galaxy sample . in particular , we find that , for the optical - only sample , the results are robust against the effect of redshift .
we thus present our results for the full sample for the optical - only case . when considering only the 46 galaxies of the nir sample and comparing the results of using only the optical data for this sample with those obtained using the optical and nir data combined
, we find a small difference in the results for galaxies with @xmath35 and @xmath8 .
for the nic3/nir sample , therefore , we present our results for @xmath35 and @xmath8 separately .
we estimate the bias in the inferred stellar population parameters from using pixel-@xmath0 relative to the global quantities inferred from aperture photometry .
in particular , we investigate whether the sum of the sed fits to individual pixels in a galaxy is consistent with the fit to the total flux from the same pixels .
these differences can arise because individual pixels can have colors that differ from the mean color of the galaxy . as a result ,
individual pixels can have stellar population quantities that can differ significantly from what is derived from the integrated colors of galaxies .
the result of summing the fits to the individual pixels in a galaxy may , in principle , be significantly different from the global quantity measured from sed - fitting to the total fluxes of the same pixels .
our aim is to investigate these differences and quantify their magnitude .
we focus on the sfr since it is a direct observable in spectroscopic ( including spatially resolved spectroscopic ) measurements . in both approaches , the current sfr is derived using the best - fit star formation history , normalized using the same scale factor required to normalize the sed to match the observed broad - band luminosity , in either the pixel ( for the first approach ) or the entire galaxy ( for the second ) .
we perform the comparison of the two approaches for the same pixels by considering only pixels with @xmath69 and within a fixed aperture of radius @xmath70 .
we use the same library of sed templates throughout the comparison : 2178 seds generated by the bc03 models . the residual difference between the sum of the sed fits to the invidual pixels and the sfr derived from sed - fitting to the total flux in the same pixels
is then computed in terms of the statistical error in the sfr measurement .
as described in section [ sec : intro ] , we aim to use pixel-@xmath0 to study ( a ) the spatial distribution of sfr and dust and its dependence on the galaxy environment and ( b ) clumpy star - forming disk galaxies at @xmath71 .
we use selected face - on disk galaxies to demonstrate how pixel-@xmath0 can be used to measure large - scale features such as radial stellar population gradients and to compute the mean properties of spiral arm structures and clumps .
we also wish to know how the different systematic errors computed for the individual pixels , as described in section [ subsubsec : methods_pixels ] , translate to systematics in these large - scale features in the galaxy . * * radial gradients of stellar populations : * we compute the mean luminosity - weighted age , @xmath22 and e(b - v ) of successive radial annuli in the galaxy .
the luminosity @xmath72 in each pixel is computed directly from the normalization of the best - fitting sed template to the observed fluxes .
the mean pixel-@xmath0 quantity @xmath73 computed in annulus @xmath74 is thus : + @xmath75 + where @xmath76 is the number of pixels in the annulus .
+ from the age , @xmath22 and @xmath77 in each pixel , two futher quanties , the @xmath78 and the mean specific star formation rate ( @xmath79 ) are also derived for each annulus .
the gradients of these quantities are then computed for ( a ) varying stellar population synthesis models and ( b ) varying passbands .
the size of the smallest annulus is chosen such that it is larger than the full width half maximum ( fwhm ) of the psf in the @xmath54 band .
the radius in @xmath80 is computed directly from the spectroscopic redshift of each galaxy and the acs angular scale of @xmath81pixel , and assuming @xmath82 , @xmath83 , and @xmath84 . *
* mean properties of spiral arms and clumps : * we use the galfit code ( peng et al . 2002 ; 2010 ) to isolate components such as spiral arms and clumps for selected face - on disk galaxies in our sample .
a disk galaxy model is subtracted from the galaxy image in the @xmath54 band and the residual image is then used as a mask to effectively isolate both the spiral arms and clumps in the galaxies .
these structures of interest correspond to the tail of the flux distribution in the residual image , and by adjusting the threshold , we are able to isolate the components individually
. the pixel-@xmath0 code is then run on the total fluxes of the pixels identified as being part of these substructures .
the distribution of the age , @xmath22 , @xmath23 , sfr and ssfr of all the pixels belonging to each clump / spiral arm is then calculated for the different cases of stellar population synthesis models .
we perform these analyses on selected disk galaxies , an example of which is shown in this work , in order to identify preliminary trends , and to refine our approach for a more thorough investigation .
a statistical study of radial gradients , spiral arms and clumps , which will also require a more careful selection of galaxies , will be explored in a future paper .
here , we illustrate the effect of using different stellar population synthesis models on the values of the parameters that pixel-@xmath0 determines using the _ bviz _ passbands .
this is first shown qualititatively by displaying the resulting maps , and second , quantitatively , by measuring the systematic error in each parameter and following its distribution . in the top panel of figure [ fig : models_optical_age ]
, we illustrate the effect of the chosen stellar populations model on the age maps .
there is a clear correlation between the pixel-@xmath0 maps of this galaxy and the @xmath54 band image .
the qualitative differences from pixel - to - pixel between the various models(m05,bc03,cb07 ) are small but there is a difference along the right spiral arm between bc03 and cb07 on the one hand and m05 on the other . bc03 and cb07 both predict that the majority of pixels along the arm have ages @xmath85gyr , while m05 shows a larger proportion of pixels showing younger ages ( @xmath86gyr )
. otherwise , both age values and their associated uncertainties in the pixels are qualitatively similar for all 3 models .
the red pixels in the inter - arm regions indicate much older populations but since these are also typically lower signal - to - noise pixels , the rms error is also significantly worse for these pixels .
pixels belonging to the sky which are artificially fit are also assigned large uncertainties .
the dust obscuration , shown in figure [ fig : models_optical_ebv ] is also correlated with the spatial distribution of the stellar age . unlike the disagreement seen in the age values of the spiral arm between the m05 models on one hand and the bc03 and cb07 models on the other hand ,
the obscuration maps demonstrate a high level of agreement among the various models . in all the models ,
the arm regions , which were predicted to have the youngest stellar populations , also have a moderately high level of dust obscuration , with @xmath87 .
the uncertainties in the dust obscuration in the pixels are similar among the various models as well .
these uncertainties correlate strongly with morphological features in the image as in the case of the stellar population age .
the knots on the spiral arms , corresponding to star - forming hii regions , are clearly displayed in the obscuration uncertainty map with the lowest errors of the entire galaxy .
these statistical uncertainties are similar among all the models considered .
the inner arms and the outermost edges of the galaxy , which are the regions with the lowest signal - to - noise in the entire galaxy , have , predictably , the highest rms uncertainties in their pixel-@xmath0 quantities . in order to quantify these effects , we use pixels from all 467 galaxies in our sample .
the systematic error due to different population synthesis models is measured according to equation 1 , with the resulting distribution shown in figure [ fig : models_test_optical ] .
the distributions are generally non - gaussian , peaking around zero systematic error , and showing a relatively small fraction of outliers .
the shape of the distribution is a result of the quantized nature of the problem , since the pixel-@xmath0 parameters lie on a discrete grid of values .
the level of agreement and the fraction of outliers in this distribution for every pixel-@xmath0 parameter is summarized in table 1 for each of the pairs of models being compared .
the two models that have the highest level of agreement in all the parameters concerned ( age , @xmath22 and e(b - v ) ) are bc03 and cb07 . in the @xmath22 parameter ,
94 percent of pixels within the galaxies are within @xmath88 and the fraction of outliers ( @xmath89 ) is 0.04 .
the high level of agreement between bc03 and cb07 is due to the fact that the continuum in the rest - frame uv is very similar between these two models .
there is a larger systematic difference in the pixel-@xmath0 parameters between bc03 and m05 , and between cb07 and m05 . in particular
, the @xmath22 parameter is most sensitive to differences between these models .
the fraction of pixels with a systematic difference in @xmath22 less than @xmath88 , is 0.71 and 0.80 for the [ bc03,m05 ] and [ cb07,m05 ] model pairs respectively .
the fraction of outliers in @xmath22 ( @xmath89 ) is 0.15 and 0.17 respectively for the same model pairs .
the impact of model differences between bc03/cb07 and m05 is much smaller for the age and obscuration .
dust obscuration is least sensitive to differences between any of the 3 models , with only @xmath90 percent of pixels having systematic differences in @xmath23 which are larger than @xmath2 .
for the inferred stellar age , the fraction of pixels having systematic differences in their stellar population age of @xmath91 is 0.86 and 0.87 for [ bc03,m05 ] and [ cb07,m05 ] respectively , while the fraction of outlier pixels in the age distribution for the same pairs of models is 0.11 . for [ bc03,cb07 ]
, 0.94 of pixels agree in their inferred stellar population age to within @xmath88 , while the fraction of outlier pixels for the same models is 0.04 .
the comparison of the stellar population synthesis models for the metallicity are not shown in table 1 because the small number of values of metallicity being sampled means that all fits are within @xmath2 . out of the three pixel-@xmath0 parameters tested ,
it is thus in the sfr e - folding time where the largest discrepancy lies between m05 on one hand and bc03/cb07 on the other .
this discrepancy is unlikely to be due only to the differing treatment of tp - agbs stars in these models , since if that were the case , we would expect better agreement between cb07 and m05 as contrasted against bc03
. the discrepancy observed in @xmath22 is likely related to fundamental differences between m05 on one hand and bc03/cb07 on the other when the sed fitting is performed using optical colors only . in section 3.5.2 , we use some example galaxies to investigate where the discrepancy in the predicted values of @xmath22 ( between m05 and bc03/cb07 ) occurs spatially in the galaxies .
finally , we test the effect of redshift on these results by repeating the above analysis for subsamples of galaxies in redshift intervals of 0.5 from @xmath92 and find that our results for the optical - only passbands are not significantly impacted by the redshift of the galaxy . in this test , we examine the effect of changing the parameter space on the pixel-@xmath0 estimates using the bc03 models .
the number of metallicity values used in the pixel-@xmath0 grid of stellar population parameters is increased from 3 to 6 , to include sub - solar metallicities , as detailed in section [ sec : sspsfromcolors ] .
the increase in the metallicity range means we are comparing the original 2178 seds with an expanded set of 4356 seds fitted to each pixel .
the distribution of the systematic error ( as a function of the statistical error ) introduced by extending the allowed metallicity range into the sub - solar regime is shown in figure [ fig : metallicity_test ] for all pixels in all the galaxies in the sample .
the fraction of pixels with small systematic errors ( @xmath91 ) and the fraction of outliers in each parameter is summarized in table 2 .
the largest effect of a broadening the metallicity range is , predictably , on the stellar population age but the effect is minimal : the fraction of outliers in the age systematic is only 0.06 .
around 6 percent of pixels have a bias in their @xmath22 that is larger than the @xmath2 uncertainties .
the obscuration is the parameter which is least affected ( @xmath93 of pixels ) by the introduction of sub - solar metallicities .
this is in agreement with the findings of conti et al .
( 2003 ) who examined the evolution of the obscuration - metallicity degeneracy in pixels in galaxies in the hdf - north , and found that there is a strong degeneracy in galaxies at low redshifts but that the relation flattens out at @xmath8 .
the conclusion of this test is that while the stellar population age and @xmath22 are marginally affected by the increased metallicity range , as expected from the age - metallicity degeneracy , the resulting systematic differences in all the pixel-@xmath0 quantities are still well within the uncertainties between different stellar population synthesis models .
here we test the effect of the passband limitations on the estimates of the pixel-@xmath0 parameters .
the stellar population synthesis model is fixed to bc03 in this test and passbands are added and removed and the pixel-@xmath0 output is compared to that produced using the full complement of 4 optical bands .
this test is initially performed for selected individual galaxies to identify the likely trends , before a full investigation on all pixels in the full galaxy sample is performed .
figure [ fig : passbands_optical_age ] illustrates the test for one example galaxy , which is the same as shown in figures [ fig : models_optical_age ] and [ fig : models_optical_ebv ] .
it shows the impact of passband variations on the age values in the pixels and their associated uncertainties .
starting with only 2 optical bands ( @xmath94 ) , the effect of different permutations of bands are tested . from the stellar population age maps ,
it is clear that 2 passbands are insufficient to resolve many of the features seen in the light distribution of the galaxy . for the case of 3 bands , the use of the redder @xmath95 combination of filters comes closer to the result obtained with the full @xmath1 filters than does the @xmath96 set .
the @xmath95 combination produces an age map that comes closer to mirroring the galaxy morphology than does the @xmath96 set , separating the arm and inter - arm regions more clearly . for this example galaxy at @xmath16 ,
most of the relevant color information for the pixel-@xmath0 therefore comes from the optically redder passbands .
a similar conclusion can be drawn from the map of rms values in the age parameter . with only 2 optical bands ,
high uncertainties throughout the disk of the galaxy reflect the inability of pixel-@xmath0 to find the correct sed with only 1 color , thus resulting in higher statistical errors in the ages inferred for the pixels . with 3 passbands ,
again it is the @xmath95 combination that seems to be closest to the @xmath1 values .
the maps of dust obscuration in figure [ fig : passbands_optical_ebv ] show that the @xmath96 combination coming closest to the result predicted from the full complement of bands .
nevertheless , the @xmath96 , @xmath95 and @xmath1 combinations all agree that the arms have intermediate levels of obscuration ( @xmath97 ) and low associated rms errors . moving to the full sample of all pixels in all galaxies
, we use equation 2 to compute for each parameter and for each pixel , the systematic error ( again in terms of the statisitical error in the parameter ) due to the difference between a given filter permutation and the full optical filter set ( @xmath1 ) .
the stellar population model is fixed to bc03 as before .
figure [ fig : passbands_optical_test ] shows the distribution of this systematic error for all the pixels in the full galaxy sample .
it is evident that the systematic error distribution has larger tails and a lower peak ( @xmath91 ) than the systematic errors resulting from stellar population model differences . as for the example galaxy in figures [ fig : passbands_optical_age ] , it is clear also that 2 filters ( @xmath94 ) result in a large fraction of outliers ( @xmath98 ) . for the 3-filter permutations ( @xmath99 , @xmath96 and @xmath95 ) , across the entire sample of galaxies
, we find that there is only a small difference in the distribution of the systematic error among the filter permutations .
we show the fraction of outliers in each case in table 3 .
it is clear that the sfr e - folding time ( @xmath22 ) and the stellar population age are the most impacted of the 3 parameters by passband limitations . in @xmath22 and in the stellar population age , over 20 percent of the pixels become outliers ( @xmath89 ) when one band of the full filter set @xmath1 is dropped .
there is not a very significant difference between the 3 sets of filters , but omitting the @xmath0 band results in an outlier fraction for the @xmath22 parameter of 0.26 compared to 0.20 for the other bands .
for the stellar population age , the outlier fraction varies from @xmath100 relative to the full filter set ( @xmath1 ) .
the dust obscuration is least impacted by passband changes , as long as a minimum of 3 are used . for each of the 3 filter sets , about 80 percent of pixels in the full galaxy sample
have an @xmath23 which agrees with the @xmath1 result to within @xmath88 . finally , as in section [ subsec : model_tests ]
, we test the effect of redshift on these results by repeating the above analysis for subsamples of galaxies in different redshift intervals from @xmath92 and find that our results for the optical - only passbands are not significantly impacted by the redshift of the galaxy . in this section
, we explore the effect of adding nir passbands from nic3 to the existing optical data for a subsample of 46 galaxies .
adding the nir data allows us to explore systematic differences between various population synthesis models more fully , since they can predict different nir colors .
in particular , it is known that bc03 and m05 differ significantly in their treatment of tp - agb stars , consequently resulting in different predicted nir colors .
this is illustrated in figure [ fig : spectra ] which shows an example sed from all three models considered in this work , for a galaxy at @xmath101 containing a stellar population that is 3 gyr old , with @xmath102 gyr , @xmath103 and @xmath104 .
there is very little difference between the shape of the continuum between the the models in the optical .
however , the same galaxy in the nir shows significant differences in the shape of the continuum among the three models . these will give rise to different predicted nir colors in the pixels .
figures [ fig : age_nir_example ] and [ fig : ebv_nir_example ] show examples of applying pixel-@xmath0 to a galaxy at @xmath105 with @xmath1 and @xmath106 and @xmath107 band imaging from nic3 .
there is a good agreement among the three models compared here in both their stellar population age maps and in the age rms map .
all three models predict a very young population of stars throughout ( @xmath108 gyr ) .
all models considered predict the core of the galaxy to have an older population of stars ( @xmath85 gyr ) than the outskirts ( @xmath109gyr ) .
the one notable difference is that the m05 models predict a slightly larger proportion of very young stellar populations ( @xmath110gyr ) surrounding the nucleus .
the models also agree very well on the age distribution within the companions of this galaxy ( to the right ) .
the models also agree well in the uncertainty maps in the stellar population age .
figure [ fig : ebv_nir_example ] shows the obscuration maps of the same galaxy as predicted by the three models .
again , there is excellent agreement among the three models in the distribution of dust and its associated uncertainty . in general
, all three models predict that the galaxy has moderate levels ( e(b - v)@xmath111 magnitudes ) of dust , but they also show pockets of high obscuration ( e(b - v)@xmath112 magnitudes ) within the main galaxy and also in its two companions .
the location of these pockets are the same in three models .
this example therefore implies that model differences do not introduce large systematic biases in the pixel-@xmath0 parameters for @xmath113 galaxies with such irregular morphologies .
we investigate this hypothesis fully when we measure systematic differences across all pixels in the combined optical - nir galaxy sample for all 3 models considered in this paper . for each parameter
, we compare this distribution of the systematic differences with that obtained with using only the optical data .
we account for the effect of redshift , particularly because for @xmath8 , the 4000 break shifts into the @xmath106 and @xmath107 passbands .
we thus split the nir sample into two : @xmath35 and @xmath8 . in figure
[ fig : nir_test_zle1 ] , we show the distribution of the systematic difference between models for the low redshift sample . the fraction of pixels showing @xmath91 uncertainties in their pixel-@xmath0 parameters for each pair of models is given in table 4 for both the optical - only colors and the optical and nir colors .
for the low redshift sample , there is a very high level of agreement between the bc03 and cb07 models in all the pixel-@xmath0 parameters , with over @xmath114 percent of pixels showing less than @xmath88 deviation between the two models . with the addition of the nir bands however
, the systematic differences between bc03 and cb07 models are exposed .
this is a result of relatively small differences among the models in their predicted rest - frame uv colors but much larger differences among them in their predicted rest - frame optical ( red ) and rest - frame nir colors .
this affects in particular the stellar population age ( with only 78 percent of pixels showing less than @xmath88 difference between the two models ) and @xmath22 ( 88 percent of pixels showing agreement between the two models ) .
the obscuration is less impacted by the model differences .
the addition of the nir colors also exposes differences between bc03 and m05 , and this impacts the stellar population age most strongly , with the fraction of pixels showing @xmath91 difference in the age decreasing from 0.89 with optical - only colors to 0.80 with nir colors .
the differences between bc03 and m05 have less impact on @xmath22 and the obscuration when the nir colors are added . when comparing cb07 with m05 , it is again the stellar population age in the pixels that is most affected by the addition of the nir passbands , with only 80 percent of pixels showing an agreement in the age to within @xmath88 between the two models , compared to 89 percent for the optical - only case . for the high - redshift ( @xmath8 ) sample ,
the story is somewhat different , as illustrated in figure [ fig : nir_test_zgt1 ] and table 5 .
there is now a much higher level of agreement between the results for the optical - only colors and the optical and nir colors combined , for all the parameters concerned .
this is due to the fact that the rest - frame uv and blue wavelength optical spectra ( which are sampled by the optical and the nir filters for @xmath8 ) are similar among the models .
this similarity in the predicted rest - frame uv / optical ( blue ) colors among the models thus manifests itself as follows . in the @xmath35 sample
discussed above , the stellar population age showed a large discrepancy between the @xmath1 and @xmath115 samples in terms of the fraction of pixels that agree on their age for the different models being compared .
for the high redshift sample , the pixels in both the @xmath1 and @xmath115 samples show similar levels of agreement in the age parameter between the different models compared .
for example , for the the bc03 and m05 models , the fraction of pixels with systematic differences in their age that are smaller than @xmath88 is 0.95 in the optical - only case and 0.93 in the nir case .
the conclusion of these tests is that , for @xmath8 galaxies , differences among the stellar population synthesis models in their predicted nir colors do not significantly bias the pixel-@xmath0 estimates of the stellar population parameters in the pixels , compared to using optical colors only .
pixel-@xmath0 can measure global properties of structures composed of many pixels within the same galaxy .
these will be used to determine the radial variation of stellar population parameters in galaxies as a function of the galaxy environment , and also to allow a detailed study of high redshift clumpy , disk galaxies , as outlined in section 1.1 and 1.2 . here
, we investigate the effect of systematic biases that can impact these global measurements , in particular of the sfr since this will be a direct observable in spectroscopic measurements . in figure
[ fig : sfr_pixelz_global ] , we show the level of consistency between the sum of sed fits to individual pixels in a galaxy and the global fit to the total flux in those same pixels , as described in section 2.3.2 .
we see a high level of consistency between the two approaches in their estimates of the integrated sfr .
we also quantify the magnitude of any deviation between the two approaches in terms of the statistical uncertainty in the sfr estimate for each galaxy from sed - fitting .
we see that for the majority of galaxies in the sample , the difference between the two sfr estimates for the galaxies are within the statistical uncertainties of the sfr measurement .
however , there is a small population of outliers ( @xmath89 ) which are galaxies with low sfr ( @xmath116 ) and which show a negative residual i.e. , the sfr estimates from pixel-@xmath0 , obtained by summing the results of the individual fits in each pixel , are higher than the sfr estimates from global fits to the total flux in the pixels .
in addition , there is a second small population of outliers which are galaxies with high sfrs ( @xmath117 ) . in these ,
the sfr estimate from the global fit is significantly higher than that estimated from the sum of the individual sed fits . in figure
[ fig : radialgradient_models ] , we present the radial variation of the stellar population age , sfr e - folding time , obscuration , sfr and ssfr using the @xmath1 passbands for the same galaxy at @xmath16 in figures [ fig : models_optical_age ] and [ fig : models_optical_ebv ] .
the radial trends in each quantity are measured according to the method described in section 2.3.3 .
all stellar population synthesis models agree on the qualitative trends of all the parameters with radius but there are some quantitative differences that arise , particularly between the different population synthesis models all three models agree that the stellar populations in the galaxy at all radii are young ( @xmath118gyr ) and that the stellar population age decreases rapidly from the center to the outskirts .
m05 predicts a mean stellar population age that is somewhat lower in the innermost part of the galaxy compared to bc03 and cb07 , and this systematic difference is of the order of the statistical uncertainty in the age measurement . bc03 and cb07 also predict a somewhat sharper drop in the stellar population age between 2 and 3 kpc , but again , the difference is well within the statistical uncertainties .
the models agree on a range of @xmath119 gyr across the galaxy . between 3 and 5 kpc
, however , the m05 model prediction diverges by @xmath88 from both bc03 and cb07 models .
this region corresponds to the spiral arm and the clumps of star - forming hii regions within it .
this implies that the different treatment of tp - agb stars in m05 compared to the other models has its biggest effect in the spiral arms of the galaxy .
the dust obscuration trends are quite similar among the models which show that the center of the galaxy is moderately obscured ( @xmath120 ) .
the obscuration then decreases up to 2 kpc and then increases to a maximum ( @xmath121 ) in the spiral arms .
the different models all find that the mean sfr in each annulus peaks in the center of the galaxy .
to explain this , we note that the current sfr is determined not only by the mean age and @xmath22 of the stellar populations , but also by the total stellar mass formed to date , which normalizes the sfr .
m05 predicts a lower stellar mass throughout the galaxy than does bc03 or cb07 . when the stellar mass is normalized out to give the specific sfr ( ssfr )
, we see that the discrepancy between m05 and bc03/cb07 narrows considerably : @xmath88 at most in all radial bins . comparing the trend in ssfr with the morphology of the galaxy ,
we conclude that the bar structure in the center of the galaxy is actively star - forming .
the models disagree , however , on how much more star - forming the bar is relative to the spiral arms at a radius of @xmath122kpc .
bc03 and cb07 predict a slightly higher ssfr ( by around @xmath123 ) in the center of the galaxy compared to the spiral arms , whereas m05 predicts a somewhat lower ssfr in the center relative to the spiral arms .
nonetheless , a general picture emerges for this galaxy , which is that star formation is not simply confined to the center but takes place throughout the galactic disk .
the results , albeit for a single galaxy , imply that pixel-@xmath0 is robust to differences between various stellar population synthesis codes when it is used to measure radial variations across galaxies .
figure [ fig : radialgradient_passbands ] shows the effect of passband limitations on measuring radial variations of stellar population parameters .
the population synthesis model was fixed to bc03 . using only the @xmath124 and @xmath125 bands
results in large systematic uncertainties which affects all the radial quantities but particularly the mean age and the obscuration determinations .
the result of using only the @xmath124 and @xmath125 bands is that the mean age is overestimated throughout the galaxy relative to the @xmath1 filter set , by as much as @xmath126 .
this can be explained by the fact at @xmath16 , the 4000 break passes out of the bluest optical bands and , as a result , the colors predicted by the model for the pixels are redder , leading to an older stellar population for the pixels .
the obscuration is underestimated throughout , and at a radius of 4 kpc the difference with respect to the @xmath1 value is as much as @xmath126 .
the @xmath22 and the ssfr estimates are less impacted by the loss of 2 passbands . the 3-passband combinations @xmath96 and @xmath95 both come closest to reproducing the @xmath1 radial trends in all the parameters .
the systematic difference between the @xmath96 and @xmath95 combinations is relatively small , but some differences do emerge nevertheless . in the center of the galaxy , the @xmath95 filter combination predicts a larger ( by @xmath88 ) mean stellar population age than the bluer @xmath96 bands . elsewhere in the galaxy ,
the three - filter combinations give results which are consistent ( within their statistical uncertainties ) with each other and also with the full @xmath1 filter set .
the radial variation of the dust obscuration is consistent among the @xmath96 , @xmath95 and @xmath1 filter sets except in the innermost annulus where the @xmath95 filters underpredict the level of obscuration relative to the other two combinations .
the differences in @xmath22 are small , although there is one important difference at @xmath127kpc where the @xmath96 set predicts a higher @xmath22 compared to the @xmath95 and @xmath1 sets .
the effect of filter cominbations on the ssfr is also relatively small throughout the galaxy except in its outermost part ( @xmath128kpc ) where the three - filter combinations overestimate the ssfr relative to the full optical set .
if we were to generalize this test , a minimum of 3 optical passbands are therefore needed for robustly computing radial trends of stellar populations in these high redshift galaxies .
in figures [ fig : spiralarms ] , [ fig : clumpa ] and [ fig : clumpb ] , we illustrate the way pixel-@xmath0 can be used to study the stellar populations of subcomponents of galaxies such as spiral arms and clumps , and we explore how this can be biased by the choice of the sps model .
the optical filter set @xmath1 is used throughout this analysis . in the spiral arm of the disk galaxy which has been isolated ,
the bc03 and cb07 models predict almost identical distributions for the parameters among the pixels that make up the spiral arm . in contrast
, some differences start to emerge between m05 and the other two models . the age distribution of the pixels peaks at 0.1 gyr for the bc03 and cb07 models , and at a lower age ( 0.01 gyr ) for the m05 model .
all three models show good agreement for the dust distribution among the pixels in the arm , which peaks at @xmath130 gyr .
there is however a systematic offset in the sfr distributions which peak at 0.003 @xmath131 with m05 , whereas it peaks at 0.03 @xmath131 with bc03/cb07 .
this difference is largely due to the lower stellar mass predicted for the pixels in the arm by m05 compared to the other models .
when stellar mass is normalized , we see a much closer agreement between m05 and bc03/cb07 .
all the models predict a ssfr distribution that is bimodal in the arm , one peaking at @xmath132 and the other at @xmath133 .
we also investigate clumps in the disk and the impact of model differences on the inferred properties of clumps in the disk of the galaxy .
clump a in figure [ fig : clumpa ] is part of the outer arm of the galaxy . as in the case of the spiral arm ,
the distribution of the stellar population age and dust obscuration among the pixels in clump a , is quite similar among the different population synthesis models . as with the full spiral arm ,
a substantial difference exists in the sfr distributions .
the m05 models predict the clump to have a lower sfr than bc03/cb07 .
the distribution peaks at @xmath134 with m05 and @xmath135 with bc03/cb07 , due to a lower stellar mass predicted for the clump by the m05 model .
it is also worth nothing that the peak of the sfr distribution within the clump occurs at a higher sfr than in the spiral arm .
the ssfr distributions in the clump are far more consistent among all the models , again showing a bimodal star formation in the clump , with one population peaking at @xmath132 and the other at @xmath133 as in the case of the spiral arm .
the sfr and ssfr distributions imply that the star formation process in the spiral arm might be driven by the star formation in the clump itself .
finally , in figure [ fig : clumpb ] , we examine clump b on the left side of the galaxy . again
, model differences affect primarily the sfr , with m05 predicting the sfr distribution to peak at @xmath136 while bc03/cb07 predict that it peaks around 0.03 @xmath131 .
irrespective of the model , however , it is evident that the mean sfr of clump b is significantly lower than clump a , indicating that clump a is the primary driver of star formation in the outskirts of the galaxy . despite some differences between m05 and the other models ,
therefore , pixel-@xmath0 can be used to effectively and accurately probe mean properties of stellar populations within components of galaxies such as spiral arms and clumps . as discussed in section 1.1 ,
this can be used to probe substructure in @xmath71 galaxies in order to determine if these objects are largely driven by mergers or are disks formed in - situ .
figure [ fig : candels_maps ] shows the results of adding the nir wfc3 imaging in the @xmath50 and @xmath51 passbands from the candels survey to the existing @xmath1 bands for an example disk galaxy at @xmath137 , with the m05 models .
the most notable difference between using the optical - only passbands and the combined optical and wfc3/nir passbands is in the spatial distribution of the oldest ( @xmath138 gyr ) stellar populations in the galaxy . in the acs - only case ,
the oldest stellar populations in the galaxy are confined in highly localized regions in the galaxy , with a concentration in the center and and in pockets in the south - west and sout - east of the image .
the addition of the wfc3/nir data , however , suggests that the oldest ( and thus redder ) populations are not simply confined in these pockets but are present more extensively in the disk itself . in the map of the specific star formation rate in the galaxy ,
both the optical - only and the combined optical and nir cases suggest a qualitatively similar trend in the spatial distribution of the ssfr i.e. , the ssfr is lowest in the center of the galaxy ( @xmath139 ) and increases towards the outskirts . away from the center , however , the addition of the wfc3/nir passbands , suggests that the disk contains a higher proportion of stellar populations with a ssfr ( @xmath140 ) that is lower than what is predicted by the optical - only colors .
the differences observed in the maps of the stellar population age and ssfr are confirmed by the radial variation of these quantities , shown in figure [ fig : candels_radialplots ] .
the wfc3/nir colors reveal an older population throughout the disk of the galaxy relative to the optical - only colors , with more than a @xmath141 difference in the luminosity - weighted mean age between the @xmath1 and @xmath115 colors and up to a radius of 6 kpc from the galaxy center , .
the radial variation of the ssfr also reflects the fact that the wfc3/nir colors are more sensitive to the older stellar population in the disk , which were otherwise not revealed by the acs - only passbands .
the wfc3/nir colors reveal a lower ssfr at all radii compared to the acs - only colors , although the difference is most significant between 3 and 5 kpc from the galaxy center .
the pixel-@xmath0 method performs sed - fitting to individual pixels in multicolor images of galaxies and obtains the values of stellar population properties , such as age , sfr , obscuration and metallicity , in the pixels .
we use it to make detailed measurements of the properties of components in galaxies such as spiral arms and clumps as well as to measure the radial variation of star formation and other properties across galaxies . in this work
, we present the application of the pixel-@xmath0 technique to a sample of 467 resolved galaxies at @xmath5 in the goods - south field in order to extract spatially resolved properties of these objects from their pixel colors .
we have focused on the impact of systematic biases on how accurately this technique can probe substructure and stellar populations in these high redshift objects .
we investigate various biases that affect the determination of stellar population properties within pixels .
for each potential source of systematic error , we examine the distribution of the systematic error across all pixels in the galaxy sample .
we also study the effect on the derived properties of components of galaxies .
we summarize our findings below . * using optical - only passbands , we find that differences between population synthesis models can bias the derived stellar population properties , although the impact is minimal for the majority of pixels .
the impact of differences between the bc03 and cb07 models are small , with over 94 percent of pixels in the galaxies having resulting systematic errors in their age , dust and @xmath22 properties that are below the statistical errors .
however , a larger systematic uncertainty arises from differences between the m05 models and bc03/cb07 ones .
this is likely related not to the differing prescriptions for tp - agb stars but to other fundamental differences between the m05 models and the bc03/cb07 models .
this causes systematic uncertainties in the pixel properties that impact a higher proportion of the pixels , resulting in an outlier fraction of pixels ( with more than twice the statistical error ) of approximately 0.15 , 0.11 and 0.03 for the @xmath22 , stellar population age and color excess ( dust obscuration ) respectively .
* we explore systematic uncertainties in the stellar population properties within pixels resulting from a changing parameter space used by pixel-@xmath0 . increasing the range of parameter values that pixel-@xmath0 searches
is expected to exacerbate sed degeneracies .
we find that when using the 4 optical bands , this systematic can be well controlled .
the fractions of pixels whose stellar population parameters are affected in this manner are only 0.03 , 0.06 and 0.01 for the @xmath22 , stellar population age and color excess respectively .
* we explore the effect of limiting the number of passbands available for the pixel-@xmath0 analysis .
we find that this results in the largest systematic errors in deriving stellar population properties in the pixels .
using 2 passbands alone results in large ( @xmath142 ) systematic uncertainties in the parameters in the majority of the pixels .
as expected , the more passbands that are available to sample the spectral range of the pixels , the more accurately stellar population parameters can be constrained .
we find , however , that moderately robust results can be obtained with just three of the optical filters within the set of @xmath1 , provided they are selected to span the 4000 break .
we find that omitting one of the 4 optical passbands can bias ( by more than twice the statistical error ) the estimates in the stellar population age and @xmath22 in more than 20 percent of pixels in the sample .
the sfr and the stellar population age are most impacted of all the stellar population properties by such limitations .
passband limitations are thus the largest source of systematic error in the pixel-@xmath0 method . * using a subsample of 46 galaxies with near - infrared data from nicmos / nic3 camera , we study how systematic biases in the pixel-@xmath0 parameters due to differences among stellar population synthesis models , are affected by the addition of the nir passbands relative to the optical - only colors .
the most important factor here is that the different models predict generally similar rest - frame uv colors but show significant differences in their predicted rest - frame nir or optical ( red ) colors .
thus the impact of adding the nicmos data has a redshift dependence .
for the low redshift ( @xmath35 ) sample , the addition of the nir passbands exposes differences in the rest - frame nir colors among the three models being compared , and this impacts the stellar population age most strongly .
for example , for bc03 and cb07 , the fraction of pixels showing @xmath91 differences in age decreases from 0.89 with optical - only colors to 0.81 with nir colors .
however , for @xmath8 , the bias in the pixel-@xmath0 parameters due to these model differences that arise from using nir colors is not significantly different to those arising from using optical - only colors .
* we investigate any systematic difference between summing the sed fits to individual pixels in a galaxy ( as done by pixel-@xmath0 ) and the global sed fit to the total flux in the same pixels within a fixed aperture .
we find in general a good agreement between the sfr estimated by pixel-@xmath0 and the one inferred from sed - fitting to the global photometry in the galaxy . while the majority of galaxies show differences in the sfr determined by the two methods which are smaller than the statistical uncertainties of the sfr measurements , we detect two small populations of galaxies with significant offsets .
the first is in galaxies which show low sfr ( @xmath116 ) , where pixel-@xmath0 estimates a higher sfr ( by more than @xmath2 ) than that determined by the global sed fit .
a second small outlier population is found for galaxies showing sfr@xmath143 , where the global sed fit to the total flux produces a higher sfr than that estimated by the sum of the individual sed fits .
* we show in some individual examples , that pixel-@xmath0 can be robustly applied to substructure and stellar populations within components in galaxies at @xmath5 .
we illustrate this with measurements of the radial variations of the age , obscuration , sfr and ssfr in these objects .
we show that systematic uncertainties in these measurements that result from using different stellar population synthesis models can generally be well constrained . in these examples , we find that the largest systematic differences arise between the m05 and bc03/cb07 models , when using optical - only passbands , as suggested by our statistical study on the full pixel and galaxy sample .
* we also show how pixel-@xmath0 can be used to measure robustly the mean properties of components within disk galaxies such as spiral arms and clumps .
our results for individual galaxies suggests that pixel-@xmath0 is generally robust to systematic errors from stellar population synthesis model differences in determining the distribution of age , dust obscuration and ssfr within these clumpy structures in high redshift disk galaxies .
it is important to note that , despite testing the above systematics , there are still several underlying implicit assumptions within pixel-@xmath0 which may contribute systematic effects to inferred results on the stellar population parameters .
these assumptions include independent pixels , independent stellar populations in each pixel and star formation histories that are described by a single burst and exponential decline .
these are likely to have less impact when the statistical properties of large populations of galaxies ( and pixels ) are investigated . in order to address the impact of these assumptions rigorously , a detailed comparison of this technique against spatially resolved spectroscopic observations for a large sample of galaxies ( as proposed by croom et al .
2011 ) will be required .
comparison with such ifu observations for a large sample of galaxies will enable a calibration of the pixel-@xmath0 method and an eventual refinement of these assumptions .
nw thanks a. becker and r. owen at the university of washington for help and advice on using code from the lsst trac , s. charlot and g. bruzual for providing the cb07 models , and r. griffith and a. beelen for useful discussions .
nw acknowledges support from the centre national dtudes spatiales ( cnes ) and the centre national de la recherche scientifique ( cnrs ) , as well as support from nsf ast 0806367 during his time as a graduate student at the university of pittsburgh .
ajc acknowledges support from nsf award ast-0709394 .
arz is funded by the pittsburgh particle physics , astrophysics , and cosmology center ( pitt pacc ) at the university of pittsburgh and by nsf grants ast 0806367 and ast 1108802 .
.effect of stellar population synthesis models .
shown for each pixel-@xmath0 inferred parameter ( row ) is the fraction of pixels with systematic differences of ( a ) @xmath91 and ( b ) @xmath89 for each of the pairs of models that were compared .
[ cols=">,>,>,>,^,>,>,>,^ , > , > , > " , ]
|
we perform a pixel - by - pixel analysis of 467 galaxies in the goods - vimos survey to study systematic effects in extracting substructure and properties of stellar populations ( age , dust , metallicity and star formation history ) from the pixel colors using the pixel-@xmath0 method .
several systematics are examined in this paper , including the effect of the input stellar population synthesis models whose seds are fitted to each pixel s colors , the effect of passband limitations and differences between the individual sed fits to pixels and global sed fitting to a galaxy s colors .
we find that with optical - only colors ( @xmath1 ) , the systematic uncertainties due to differences among stellar population synthesis codes are well constrained .
the largest impact on the stellar population age and sfr e - folding time estimates in the pixels arises from differences between the maraston ( 2005 ) models on one hand and the bruzual & charlot ( 2003 ) and charlot & bruzual ( 2007 ) models on the other , when optical - only ( @xmath1 ) colors are used .
this results in systematic differences larger than the @xmath2 uncertainties in over 10 percent of all pixels in the galaxy sample .
the effect of varying the number and choice of available passbands is more severe . in 26 percent of the pixels in the full sample ,
these limitations result in systematic biases in the age determination which are larger than the @xmath2 uncertainties in the measurements .
robust results can , however , still be obtained with a minimum of 3 optical filters provided they span the 4000 break .
near - ir data is also added to a subsample of 46 galaxies from the goods - nicmos survey and systematics arising from model differences are again investigated .
differences among the models in their predicted rest - frame red / nir colors manifest themselves as follows . for @xmath3 galaxies
the observed optical / nir colors span the rest frame uv - optical sed , and the use of different population synthesis models does not significantly bias the estimates of the stellar population parameters compared to using optical - only colors .
however , for @xmath4 , where the rest - frame nir is still probed , there is a larger discrepancy between models when using optical - only colors compared to optical / nir colors .
this affects in particular the age determination in the pixels . with this characterization of the systematic errors
, we illustrate how pixel-@xmath0 can be applied robustly to make detailed studies of substructure in high redshift objects such as ( a ) radial gradients of properties such as age , sfr and dust and ( b ) the distribution of these properties within subcomponents such as spiral arms and clumps .
finally , we show preliminary results from applying pixel-@xmath0 to galaxies in the candels survey , illustrating how the new hst / wfc3 data can be exploited to probe substructure and stellar populations in @xmath5 galaxies .
|
the femtoscopic correlation function , @xmath3 , is defined as a ratio of the probability to observe a correlated pair of particles at a given relative momentum in the same event ( numerator ) , @xmath4 , to the probability to observe such a pair in an uncorrelated state ( denominator ) , @xmath5 : @xmath6 the uncorrelated distribution is usually obtained by mixing particles from different events . here the relative momentum is given in the pair center of mass frame as @xmath7 for particles 1 and 2 . here
@xmath8 is the notation commonly used for femtoscopic correlations for pairs of two different types of particles , and will be used interchangeably with @xmath9 ( traditionally used in femtoscopic correlations of pairs of identical partilces ) later in the paper .
we expand the numerator , denominator and correlation function in spherical harmonics , e.g. : @xmath10 .conditions on the distribution imply relations between the different terms in the @xmath11 expansion .
condition 1 is always valid for the true and mixed pair distributions and the correlation function .
condition 5 is always valid for like pair correlation and the corresponding pairs distributions . by exploiting symmetries in 2 - 5
, we can reduce the number of components in our data vectors .
[ table1 ] [ cols="<,<,<",options="header " , ] we can compute the pairs distributions directly in spherical harmonics by observing that @xmath12 can be built up by summing over the pairs , which is essentially a monte - carlo integration process : @xmath13 here , @xmath14 is the number of pairs in the spectrum . in our approach
a pair is added to @xmath15 ( @xmath16 ) in the following way .
first we calculate the relative momentum @xmath17 and decompose it into @xmath18 ( which determines the 1d @xmath19-bin number ) , @xmath20 and @xmath21 . having the angles , we can calculate the spherical harmonics functions , usually up to some limiting value of @xmath22 .
then the pair is added to all histograms of the corresponding function , with weights equal to the respective @xmath11 s .
we could eliminate many of the components of @xmath15 and @xmath16 by taking advantage of the symmetries in table i , but we have chosen to keep all components so that we can perform cross - checks of our work .
this does introduce complications when performing some matrix manipulations , as we discuss in the following section .
the covariance can be built in a similiar way by noting ( @xmath23 only ) : @xmath24 taking the diagonal elements ( i.e. @xmath25 ) , we find ( uncorrelated ) uncertainties of @xmath26 in contrast with our approach , in the traditional representation both the numerator and denominator were stored as 3d histograms , using either the bertsch - pratt coordinates @xcite in cartesian form @xmath27 , @xmath28 and @xmath29 , or in spherical : @xmath18 , @xmath30 , @xmath31 . in the traditional representation
the numerator ( denominator ) is a 3d histogram , and each signal ( background ) pair is added to exactly one bin of this histogram with weight 1.0 .
this representation has several disadvantages : one needs significant statistics to have a meaningful number of pairs in each bin , single - particle momentum acceptance can result in `` holes '' or empty bins in two - particle , or relative momentum space , and the binning corrections need to be applied . also going to higher moments in the decomposition
requires larger number of bins on the @xmath31 and @xmath32 direction . in practice , we store both the real and imaginary parts of the numerator and the denominator as an array of one - dimensional histograms in @xmath33 , as seen in eq . , which is represented as the vector @xmath34 : @xmath35 there is also a corresponding covariance matrix @xmath36 , which is a two - dimensional matrix for each @xmath19-bin and is packed in an analogous fashion .
since we have expanded the pair distributions and the correlation in spherical harmonics , we have @xmath37 with the packing in eq .
, equation can be written very compactly : @xmath38 .
gives us a way to compute @xmath39 directly from the pair distributions expanded in spherical harmonics . here
the @xmath40 matrix is written in terms of wigner 3-@xmath41 symbols as : @xmath42 in order to calculate the correlation function , we view the problem as an inverse problem . to solve it
, one needs to minimize the @xmath43 : @xmath44 the formula uses the full covariance matrix in the true distribution , but not in the mixed pair distribution .
because the mixed pair distribution is constructed by pairs from different events , it is not limited by statistics and can be computed to arbitrarily high precision , making the uncertainties negligible for our purposes .
the problem of minimizing the @xmath43 is identical to the one posed by the imaging procedure in ref . @xcite and
the solution is well known : @xmath45 where the covariance is also calculated : @xmath46 the uncorrelated uncertainties of the correlation are just the square root of the trace of the covariance @xmath47 : @xmath48 . written out
, we are simply taking the diagonal elements ( i.e. @xmath25 ) and finding ( uncorrelated ) uncertainties of @xmath49 with the error propagation done in this way , one gets the cross-@xmath50 correlations in the covariance matrix by virtue of the cross-@xmath50 correlations built into the @xmath51 and @xmath36 matrices .
we compute the correlation function according to eq . .
as this involves several matrix inversions , it is important to make sure that the matrix determinant is not zero .
that means that one can not use @xmath50 combinations that are a linear combination of other @xmath50 s .
as all of the pair distributions are real , we can not keep @xmath52 and @xmath53 components at the same time ( see table [ table1 ] ) .
therefore we adopt the convention to only use positive components ( including @xmath54 component ) .
therefore we remove all functions with negative @xmath55 from both @xmath34 and @xmath56 when solving eq .
( [ cformula ] ) .
one can add the missing negative @xmath55 components to the correlation function by multiplying the positive @xmath55 values by the appropriate factor of @xmath57 .
for consistency one may repeat the procedure , this time removing the positive @xmath55 components from @xmath34 and @xmath56 and the obtained results should be identical .
we remind the reader that each @xmath19 bin is independent in @xmath34 , @xmath56 and @xmath58 . therefore solving eq .
can be done for each @xmath19 bin independently .
the starting points then are not vectors of functions , but simply vectors of real numbers @xmath59 and @xmath60 and the result is also a vector of real numbers @xmath61 .
a covariance matrix @xmath62 is in this case a 2d matrix of real numbers , as is the resulting correlation function covariance matrix @xmath63 .
solution of eq . then reduces to a problem of solving a set of linear equations , for which many standard numerical algorithms exist .
the procedure is repeated for each @xmath19 bin and the @xmath58 vector is filled in steps .
we have tested and applied the formalism to the construction of two - particle correlation functions for identical and non - identical particles in relativistic heavy - ion collisions .
our tests include a study of the robustness of our approach in the presence of @xmath64 acceptance holes in relative momentum , a common occurance in experiments including the star experiment of which one of the authors ( kisiel ) is a collaboration member .
below we present some tests of the method with a realistic model .
two particles are femtoscopically correlated if they have a small relative momentum @xmath65 in the pairs rest frame ( for identical particles we use @xmath66 ) .
if they are not identical and have different masses , they must have different momenta in the laboratory frame .
in such case it can happen that due to specific momentum acceptance of the experiment , pairs with specific values of @xmath67 and certain combinations of polar and azimuthal components of @xmath68 can not be measured . in terms of the spherical harmonic representation ,
this results in a hole in the pair acceptance for certain regions in @xmath69 , @xmath70 and @xmath71 .
this is observed e.g. for pion - kaon pairs in the star experiment .
such a hole presents a methodological problem for traditional methods of decomposing the correlation function in spherical harmonics as they rely on the existence of certain symmetries in pair distributions .
in particular , they assume that the multiplicity of pairs with a given @xmath72 is equal to the multiplicity of pairs with @xmath73 . for non - identical particles
there is no such symmetry . while it is certainly possible to improve the existing methods and to remove this dependence , we propose to move to the more advanced decomposition method presented in this paper and bypass the problem altogether . in our method ,
this hole is reflected in both the numerator and denominator by a lower number of pairs contributing at some bins of @xmath69 .
examples of the numerator of the correlation function , binned directly in spherical harmonics , are shown in fig .
[ fig : numlm ] . in this case , the distributions result from a simulation of the @xmath74 correlation function in the therminator model using the star detector acceptance @xcite .
the acceptance is symmetric with respect to @xmath75 so the @xmath76 component vanishes @xcite .
all the imaginary components vanish .
also the @xmath77 as well as @xmath78 and @xmath79 vanish due to polar angle symmetry . apart from that
, the numerator shows non - trivial structure both as a function of @xmath69 and @xmath70 .
the synergy between spherical harmonic decomposition and femtoscopic correlation function is nicely illustrated in this plot .
a full 3d information , which in traditional 3d implementation would require tens of thousands of bins to store , is reduced to a few 1d histograms . out of these only a select few carry important information , while others conveniently vanish due to the intrinsic symmetries of the pair distribution .
the significance of the components diminishes with growing @xmath22 , ensuring that cutting the decomposition at some @xmath80 should not distort the function .
having in mind that the underlying numerator has a non - trivial structure both in @xmath69 and @xmath70 it is interesting to see how the correlation function itself , calculated with the method above , behaves .
it is shown in fig .
[ fig : cfnlm ] .
again , the imaginary components all vanish , as they should .
the @xmath81 component shows the expected behavior coming from a coulomb repulsion of same charge pion and kaon .
the @xmath82 and @xmath83 components show small deviations from zero , which signals the fact that the size of the underlying system is not the same in the @xmath84 , @xmath85 and @xmath86 directions .
also the @xmath87 component deviates from zero significantly , a signature of the average emission point asymmetry between pions and kaons . in summary
the example confirms several important points : ( a ) the correlation function can be calculated via the direct @xmath11 method .
( b ) the important physics signals in @xmath81 , @xmath87 , @xmath82 and @xmath83 , are preserved .
( c ) other components of the correlation function vanish , as they should - an important cross - check of the method . to attempt to determine the practical limits of the technique , we have performed a test .
first , we calculate the correlation function for identical pions using the therminator model .
we chose to use identical neutral pion pairs for the calculation simply because final state interactions do not distort the correlation appreciably , meaning that the correlation shape can be characterized simply by the correlation radii @xmath88 , @xmath89 and @xmath90 .
the source size has been set to reasonable values in the longitudinally co - moving system ( of @xmath91 fm ) .
the first calculation does not have any acceptance holes .
we then repeat the calculation , introducing an artificial hole in the acceptance by removing both from the numerator and denominator all pairs within the hole .
the hole is at midrapidity ( small @xmath92 ) , small @xmath19 ( from @xmath93 to @xmath94 gev / c ) and with varying width in @xmath70 ( from @xmath95 to @xmath96 ) .
the results are shown in fig .
[ fig : cfhole ] .
one can see that introducing the hole had no influence on the extracted correlation function within statistical errors .
indeed , the dominant effect of the acceptance hole has been to decrease statistics , increasing statistical scatter and the corresponding uncertainty . to further make this point
, we show the analytical prediction for how the spherical harmonics should look like for these sizes in the black dashed lines .
as one can see , all points follow the lines perfectly .
our results are in contrast to what would happen in the traditional approach of expanding the correlation in spherical harmonics after making the ratio of 3d histograms .
if there is poor statistics due to a gap in acceptance , then one will need a large number of spherical moments to capture the purely statistical fluctations present in the poorly populated high-@xmath19 bins .
what is more insidious , because the correlation is a ratio , the structure in poorly determined 3d bins appear to `` cancel out '' even when the poorly resolved data should not cancel out .
rather , the poor statistics should give rise to large uncertainties and not contribute to the spherical harmonic expansion ( which is what happens in our method ) . in fig .
[ fig : cfrad ] , we show the values of the analytical fit to the spherical harmonics vs. the hole size .
the lines are the `` input radii '' from the therminator model . as one sees ,
the fit results are very stable and moreover in reasonable agreement with the input values .
note : exact agreement between the input and extracted radii can not be expected because the collective motion in the therminator model shrinks the effective homogeneity length seen by the pairs and hence the correlation radii .
even the very large hole of @xmath96 ( only a quarter of acceptance remaining ! ) our method seems to preserve all the relevant components , provided that enough statistics remains outside the hole region . in order to be useful , the procedure for calculating the correlation function directly in spherical harmonics should allow for the application of the standard experimental corrections .
here we briefly describe how this can be done
. experimental resolution for two - particle reconstruction and identification is usually dominated by two issues : track merging ( where two tracks in the detector are reconstructed as one ) and track splitting ( where a single track is mistakenly reconstructed as two ) .
these have non - trivial dependence on both the particle momenta as well as their trajectory in the detector .
this is usually corrected for by assigning a weight to each pair , based on the detailed detector simulation .
such weighting can be incorporated in the procedure in a straightforward way . when filling the numerator and the denominator with pairs , one simply fills it with the appropriate weight .
mathematically it amounts to modifying eq . by multiplying the @xmath97 by an additional weight @xmath98 , coming from the above mentioned correction .
another common issue is the particle purity , namely the fraction of pairs in the sample that should be treated as correlated .
a pair may be not correlated if one of the particles is misidentified or if at least one of the particles comes from a weak decay .
the experiment should be able to estimate the the average purity of pairs @xmath99 , which can be ( and usually is ) a function of particles momenta , and therefore also of the pair relative momentum @xmath9 .
we use the traditional formula : @xmath100 from the correlation function @xmath101 , we can obtain the correlation effect @xmath102 . in
spherical harmonic representation this only modifies the @xmath103 component : @xmath104 , while others remain the same : @xmath105 .
. simplifies to : @xmath106 we immediately note that it is equivalent to calculating the correlation function from the numerator and denominator .
therefore it is enough to express purity @xmath99 directly in spherical harmonics and treat it as denominator , take the measured correlation function and treat it as numerator and finally apply the mathematical formalism described in this work to obtain the correlation function corrected for purity .
the method presented in this paper has been successful applied to the femtoscopic correlation functions in heavy - ion collisions .
it should be possible to apply it to other fields as well , however one has to take into account limits of the method applicability .
the basic formula is strictly correct mathematically only if one uses an infinite number of @xmath2 components for all the functions @xmath34 , @xmath56 and @xmath58 . in practical application
one needs to limit oneself to a specific value of @xmath22 .
this is only allowed if the higher @xmath22-moments are negligible .
the femtoscopic correlation function is very well suited to the method because the intrinsic symmetries of the pair distributions limit the relevant @xmath22 components to a practical maximum of 6 .
most important information is contained in @xmath107 , @xmath108 and @xmath109 components .
we have shown that the method remains stable for any reasonable acceptance hole in @xmath70 region .
it is also clear that the method will start breaking down only for really small values of @xmath110 acceptance and in the extreme case of the `` hole '' taking up the whole @xmath110 acceptance the method will simply break down due to lack of data .
the method in effect interpolates the correlation function in the region where there is no data by assuming certain symmetries in the underlying pair distribution .
using the method described here for femtoscopy , acceptance holes are `` irrelevant '' - any reasonable femtoscopic measurement will have a large enough acceptance to be insensitive to the holes .
the authors wish to thank andrew glenn for his careful reading of the manuscript
. 70 tapley , b. , j. ries , s. bettadpur , d. chambers , m. cheng , f. condi , b. gunter , z. kang , p.nagel , r. pastor , t. pekker , s.poole , f. wang , j. geodesy ( 2005 ) , doi 10.1007/s00190 - 005 - 0480-z .
p. magain , f. courbin , and s. sohy , astro .
j. , * 494 * , 472 - 477 ( 1998 ) ; f. courbin , p. magain , m. kirkove , and s. sohy , astro .
j. * 529 * , 1136 - 1144 ( 2000 ) ; j.r .
bond , a.h .
jaffe , l. knox , phys .
d 15 , pp .
2117 - 2137 ( 1998 ) .
g. hinshaw , _
et al . _ data processing , sky maps , and basic results .
astrophysical journal supplement ( submitted ) , , ( 2008 ) .
d. brown , _ et al .
c * 72 * : 054902 ( 2005 ) .
z. chajecki and m. lisa , arxiv:0803.0022 [ nucl - th ] ( 2008 ) .
l. tenorio , j.a .
scales , and r. snieder , _ noise , discretization and truncation in inverse problems _ , http://landau.mines.edu/~jscales/papers/noise_disc.ps.gz .
blais , d.a .
provins , j. geodesy 76 : 29 - 35 ( 2002 ) , doi 10.1007/s001900100217 . s. pratt , t. csrg , t. zimayi , phys . rev .
c * 42 * , 2646 ( 1990 ) ; g. bertsch , m. gong , m. tohyama , phys .
c * 37 * , 1896 ( 1988 ) .
a. kisiel , braz .
j. phys . ,
* 37 * 3a , 917 - 924 ( 2007 ) .
|
we present the formalism for calculating the femtoscopic correlation function directly in spherical harmonics . the numerator and denominator
are stored as a set of one - dimensional histograms representing the spherical harmonic decompositions of each .
we present the formalism to calculate the correlation function from them directly , without going to any three - dimensional histogram .
we discuss the practical implementation of the method and we provide an example of its use .
we also discuss the stability of the method in the presence of angular holes in the underlying data ( e.g. from experimental acceptance ) .
spherical harmonics are one of the most commonly used mathematical tools for the analysis of experimental data .
for example , geopotential models of the earth s gravitational field are matched to experimental data up to harmonic order @xmath0 @xcite .
measurements of the cosmic microwave background is expanded to by the cosmic background explorer ( cobe ) @xcite and successor experiments extend further to an impressive @xmath1 @xcite . in the femtoscopic measurements of particle emitting sources in heavy - ion collisions , one also finds it useful to expand the measured correlations and extracted sources in spherical harmonics @xcite . in nearly all applications of spherical harmonics
to data analysis and reduction , one is faced with the problem of `` holes in the data '' i.e. sampling bias @xcite . in the cobe analysis mentioned above , this bias occurs because the milky way masks a sizable solid angle of the sky .
when constructing potential maps of the earth , the problem is even more severe as one uses strips of data obtained from various satellite and balloon - borne experiments to derive the map @xcite . in heavy - ion collisions ,
the sampling bias most often arises because the detector acceptance does not span all of phase - space . unlike other applications ,
the femtoscopic correlation functions are actually ratios of two single particle distributions : the true pair distribution ( the numerator ) and the mixed pair distribution ( the denominator ) .
thus , expanding the correlation function in spherical harmonics is a more involved than expanding the underlying single particle distributions .
it is often impractical to simply bin the two singles distributions in 3d histograms and make a ratio since one does not have a meaningful number of pairs in each bin , either due to the statistics of particle production or due to detector acceptances .
furthermore , since the end result is a spherical harmonic decomposition of the correlation , one must have absurdly high harmonics in one s decomposition to resolve high - momentum bins which have small angular extent and marginal statistics . rather than pursue this
, we adopt an alternate approach : we construct the raw pair distributions directly in spherical harmonics , then we extract the correlation function , also in spherical harmonics , by viewing the ratio as an inverse problem .
a feature of this approach is that one preserves the full cross-@xmath2 data covariance that is currently ignored when imaging 3d correlations @xcite .
we now outline this paper . in the first section ,
we detail how we expand the numerator and denominator distributions in spherical harmonics and how we pack them into the vectors and matrices for further manipulation . in the second section ,
we describe how to compute the correlation in spherical harmonics using these harmonic expansions . in the last section ,
we demonstrate the technique in realistic examples .
further , we show how this technique is insensitive to the sampling bias imposed by sizable holes in pair acceptance .
|
our quest for understanding the origin and evolution of galaxies and larger
scale structures is based on the exploration of large volumes of space through redshift surveys of luminous objects .
this short review concentrates on using clusters of galaxies as the basic bricks by which to study large - scale structure ( lss hereafter ) on the largest accessible scales .
in fact , with mean separations @xmath1 , clusters of galaxies are ideal objects for sampling efficiently long wavelength ( @xmath2 ) density fluctuations over large volumes , i.e. well in the linear regime where inhomogeneities are expected to fully reflect initial conditions as they emerged at recombination .
this means that , for example , we do not necessarily need to go through n
body simulations to produce reliable model predictions for the clustering of clusters , but can use analytic approximations with sufficient accuracy @xcite .
the second important advantage , which will be discussed at some length later in the text , is that clusters can be selected directly through their x - ray luminosity , a quantity that correlates well with their mass @xcite .
this is not what happens with galaxies , where luminosity is usually a measure of star formation efficiency , rather than of total mass itself .
at the same time , with respect to the loose optical definition of a cluster , x - ray selection provides the ability to select samples with a well defined selection function , essentially that of a flux
limited sample as in the case of galaxy surveys . on the other hand ,
clusters provide a _ biased _ view of clustering : they only represent the very high peaks of the global density field , such that their variance is amplified with respect to galaxies ( and of course , mass ) as first described by nick kaiser @xcite .
therefore , they miss the small scale details of the lss ( that can be complementarily better studied through galaxy surveys ) .
nevertheless , for the same reason the cluster distribution might provide an enhanced view of possible low
amplitude features in the clustering spectrum on very large scales .
this review inevitably reflects a personal perspective and covers a limited amount of work .
however , an effort was made to provide references to complementary topics , as e.g. galaxy surveys and optically selected clusters . using these
, the interested reader can surely enlarge the restricted view of large - scale structure presented here .
i also apologize to those colleagues whose work is not adequately represented in this paper .
even when studying the distribution of galaxies through redshift surveys ( see the review by da costa in this same volume ) , one should always be aware that what is being described is the clustering of a specific class of objects , not of the matter .
we know for example that infrared selected galaxies show a weaker clustering with respect to optically selected galaxies @xcite , or that elliptical galaxies are more clustered than spirals @xcite , or that high - redshift ly - break selected galaxies look ( at @xmath3 ) as clustered as nowadays normal galaxies @xcite . in other words , with redshift surveys of visible objects we always study the distribution of _ tracers _ of the lss .
the relation to the mass distribution depends on the kind of tracer we are using , such that the clustering in the light will be in some way proportional to the clustering in the mass .
the proportionality between these two distributions is expressed through the _ bias _ function , in general a function of scale and cosmic epoch .
one of the simplest possibilities , which is shown to work reasonably well in the local universe , is that of a statistical _ bias parameter b _ independent of scale relating the variances in galaxy counts and in the mass : @xmath4 a fair amount of work has been dedicated to the problem of galaxy biasing during the last fifteen years , both theoretically ( e.g. @xcite ) and observationally ( e.g. establishing that for `` normal '' optically selected galaxies we have @xmath5 over a fair range of scales @xcite ) .
however , understanding the physical origin of the bias is not an easy task , as it involves comprehending the details of how the mass of a galaxy translates into the visible stellar light we use to select it for our surveys . despite the great advances of the last few years in our knowledge of the early phases of galaxy formation @xcite , we are still very ignorant about the details of star formation and evolution within galaxies , and as a consequence @xmath6 remains a free parameter when comparing galaxy clustering to the models .
clusters of galaxies have a honourable history as an alternative , complementary tracer of lss .
although one could spend endless discussions on the semantic of the word `` structure '' and argue that only galaxies do describe it properly , it is in fact a mere question of `` spatial resolution '' and of the scales on which we want to focus our attention .
galaxies are best for the fine details , but ( even in the 2df and sdss era )
. the latter , in particular , is measuring redshifts for 1 million galaxies using a dedicated telescope , with additional photometric data in five bands for almost @xmath7 galaxies .
] ) it is difficult to fully cover large volumes of space using galaxy redshift surveys , because measuring galaxy redshifts costs time .
groups and clusters are much more efficient to cover large volumes , as they are sparser objects , and are therefore excellent for studying the gross structure and its statistical properties in the weak clustering regime .
since clustering extends to scales of the order of 1 gpc , there is a good deal of structure to be `` seen '' and measured using such sparse tracers ( note e.g. the chains of clusters visible both in the data and simulations of figs.[fig : reflex_cone ] and [ fig : sim_clus ] ) .
also , we are used to the rarity of very rich , massive clusters , but in fact this is just a limitation of current x - ray surveys .
an all sky x - ray survey to a sufficiently faint flux limit ( @xmath8 erg s@xmath9 @xmath10 , i.e. two orders of magnitude fainter than the current rosat all sky survey ) would provide a finer description of structures connecting rich clusters out to @xmath11 , detecting chains of faint x - ray groups as pearls on a necklace .
but , are clusters and groups so well defined as a class to be used to trace lss in a homogeneous and statistically reliable way ? if defined in the optical as galaxy overdensities , clusters are in fact just a `` collection of pieces '' , thus intrinsically prone to subjective interpretations and biases @xcite . overcoming at least the human intervention in this process prompted the construction of automatic cluster catalogues , based on the first digitised galaxy photographic surveys produced in the uk , notably the edcc @xcite and apm @xcite catalogues ( see also contributions by gal @xcite and kim @xcite in this volume ) .
however , it is when we come to estimating its mass that we are faced with the intrinsic ambiguity of the optical definition of a cluster .
richness ( i.e. the number of galaxies observed in projection within a fiducial radius , corrected for the expected background contamination ) has a poor correlation with mass @xcite . a much better way would be to measure the velocity dispersion of the cluster galaxies , which at equilibrium ( no galaxy infall ) is a measure of the cluster potential well .
however , this requires a very large number of galaxy velocities to be reliably measured @xcite and still one will not be sure that the sample contains all clusters within a given volume and above a given mass .
this simply because in the first place the cluster selection function was only loosely related to this fundamental parameter , which is what model predictions are based upon .
this is a serious problem if one wants to do cosmology with clusters of galaxies , which means e.g. measuring their mean density or two point statistics , because questions as simple as `` what is the volume explored by my survey ? ''
do not have an obvious answer when clusters are selected just as overdensities in the galaxy distribution .
x - ray selection represents currently the most physical way by which to identify and homogeneously select large numbers of clusters of galaxies surface brightness dimming is compensated by a @xmath12 increase in the cmb energy density , which would make an sz survey for clusters effectively flux limited also at very large redshifts ( see e.g. @xcite for a review ) .
large - scale applications of this effect have been so far limited by technology issues , such that only about 20 clusters with observed effect are currently known , but one can foresee a rapid development in the coming years . ] .
clusters shine in the x - ray sky thanks to the _ bremsstrahlung _ emission produced by the hot plasma ( @xmath13 kev ) trapped within their potential wells . for this mechanism the bolometric emissivity ( i.e. , the energy released per unit time and volume ) at temperature @xmath14 scales as @xmath15 , where @xmath16 and @xmath17 are the number densities of electrons and ions , respectively .
the dependence on the density squared is one reason why the identification of clusters in the x - ray band is less affected by false projection effects . in fact , traditional optical selection based on galaxy overdensities depends only linearly on the density of galaxies .
fig.[fig : xray_field ] explicitly shows how in the x - ray band clusters emerge as single , mostly isolated , extended sources ( virtually at any @xmath18 ) , which can be unambiguously identified by an x - ray telescope with sufficient resolution and sensitivity . at the same time
, the selection function of an x - ray cluster survey can be determined to high accuracy , knowing the properties of the x - ray telescope used , in a similar way to what is usually done with magnitude
limited samples of galaxies @xcite .
this is a crucial point if our cluster surveys are to be used as cosmological probes to address global properties and not only for studying single , yet interesting , objects .
the intracluster gas temperature , as measured through the observed x - ray spectrum , is the most direct probe of the potential well : @xmath19 ( where @xmath20 is the proton mass , @xmath21 the gas mean molecular weight , @xmath22 the galaxy 1d velocity dispersion and @xmath23 the cluster virial mass ) .
current large surveys of x - ray clusters are based on instruments with limited spectral resolution ( essentially the rosat satellite ) and this quantity is therefore not available in general .
however , at least on a phenomenological basis , x - ray luminosity , a more directly observable quantity , shows a good correlation with temperature , @xmath24 with @xmath25 and a small scatter , @xmath26 .
although one can debate at length about the physical processes that shape up such a relatively tight relationship ( and there were a few discussions at the meeting on this subject , see e.g moscardini and white contributions in this volume ) , this does not affect its practical usefulness : we can select clusters by x - ray luminosity and be sure we are selecting them by mass with an error estimated to be @xmath27 % when taking into account all uncertainties in the various steps @xcite . even if pre
heating processes as e.g. supernovae explosions are important in the overall cluster thermodynamics ( and necessary , given the observed slope of the relation ) , it has been shown that these can not perturb the temperature mass relation at more than the 1015% level @xcite .
therefore , once we establish that the l t relation is tight , the l
m will be as well and we can safely use @xmath28 as possibly the cheapest reliable way to select clusters by mass . what is surprising is that this relation seems to hold rather well also at redshifts close to unity and above @xcite , which makes pre heating even more necessary . still , this means that we can safely make cosmological predictions for clusters of a given mass out to these redshifts and sensibly compare them with observed clusters of a given luminosity ( @xcite ) .
this is by far superior to any attempt to estimate the mass of clusters using an optical estimator , as e.g. richness @xcite ( although see @xcite for a possible improvement ) .
the study of clustering of optically selected clusters has a long tradition , since the early pioneering work on the abell catalogue @xcite , and subsequent studies on automatic cluster catalogues as the edcc @xcite and apm @xcite . only in recent years
similar clustering works have become possible on x - ray selected samples of clusters @xcite , mostly thanks to the completion of the rosat all sky survey ( rass , see @xcite for details ) .
the largest , statistically homogeneous sample of clusters with measured distances constructed so far from the rass for clustering studies is the reflex ( rosat - eso flux limited x - ray ) survey @xcite .
the survey contains 452 clusters over the southern celestial hemisphere ( @xmath29 ) , at galactic latitudes @xmath30 and is more than 90% complete to a flux limit of @xmath31 erg s@xmath9 @xmath10 ( in the rosat band , 0.12.4 kev ) .
redshifts for all reflex clusters have been measured during a long observing campaign ( 1992 - 2000 ) using eso telescopes .
[ fig : reflex_cone ] plots the 3d distribution of reflex clusters within @xmath32 , showing a number of superstructures with sizes @xmath33 . despite the fading with distance due to the flux
limited selection function , it is visually evident how clusters are still strongly clustered among themselves .
one of the main motivations for constructing this survey was to measure the power spectrum on scales approaching 1 h@xmath9 gpc .
[ fig : pk ] shows an estimate of the power spectrum of reflex clusters ( from @xcite ) , compared to that of galaxies and to the best fitting cdm model ( dashed line ) .
this comparison is particularly significant because here , contrary to when comparing models to galaxy power spectra , the normalisation ( i.e. the bias factor of the specific clusters used ) is not a free parameter , but is computed given the well understood mass selection function of reflex and the appropriate theory @xcite . in this way
, it has been shown that an @xmath34 model ( open or @xmath35dominated ) , best matches * both * the shape and amplitude of the observed reflex correlation function and power spectrum @xcite .
the sensitivity of these quantities to variations in the value of @xmath36 is shown by the plots in fig .
[ fig : pkxi_bg2001 ] . from these plots
, one can also appreciate the remarkable agreement in shape between galaxies and clusters on all scales , with the correlation function breaking down to zero around @xmath37 for both classes of objects .
such simple proportionality was never seen so clearly with previous cluster samples and is a confirmation of the bias scenario where clusters form at the high , rare peaks of the mass density distribution @xcite .
it also confirms the reassuring view that at least above @xmath38 the galaxy and _ mass _ distributions are as well linked by a simple constant bias . one important point from fig .
[ fig : pk ] is how significant is the evidence for a turnover in the power spectrum at low @xmath39 s .
this issue has been addressed using the so called karhunen - loeve transform to the reflex data , exploring volumes up to @xmath40 size @xcite .
such specific likelihood analysis which is based on finding the optimal basis of eigenvectors given the survey geometry @xcite , confirms the evidence for a turnover in the reflex @xmath41 at @xmath42 , providing a best fitting value for the cdm shape parameter @xmath43 .
similar power on these scales is seen in the 2df galaxy redshift survey results ( although published ) could not be obtained in electronic form at the time of writing , and a direct comparison plot similar to fig .
[ fig : pk ] could not be done . ]
@xcite .
finally , fig .
[ fig : reflex_csipz ] plots the two point correlation map for the reflex survey data .
this function is used to evidence distortions produced by peculiar velocities on the observed galaxy and cluster 3d maps , which by definition are constructed in _
redshift space_. in the reflex case , it shows for the first time a clear indication of infall of clusters towards superstructures @xcite .
this result indicates that in most previous cluster samples this effect was masked by selection effects and redshift errors .
contrary to these , in this sample there is basically no elongation of the contours along the @xmath44 direction ( the line of sight component of the separation @xmath45 between pairs , @xmath46 ) , which tells us that both projection effects and redshift errors are negligible in this survey .
on the other hand , the contours are significantly compressed at large @xmath47 s : this is the fingerprint of streaming motions ( see @xcite for a discussion of these effects based on the hubble volume simulations ) , and is a function of the parameter @xmath48 ( see @xcite for details and @xcite for an application to the 2df survey ) .
the reflex power spectra from the largest volumes available , tend to show significant power around @xmath49 , but still consistent with a cdm model with very low value of @xmath50 , as stated above .
significant power in this range of scales is also expected if the baryonic contribution to the mean density is non
negligible with respect to the dark matter . in this regard
, there has been significant interest in the last few years about the possibility of detecting other `` baryon features '' , i.e. wiggles in the observed power spectrum amplitude produced by acoustic oscillations within the last scattering surface @xcite .
the observational results so far have been contradictory : miller et al .
@xcite claim to detect oscillations through a joint statistical analysis of the power spectra from three different cluster@xmath51galaxy data sets .
interestingly , their finding is very consistent to what one would expect by just taking the best fitting set of cosmological parameters from the very last cmb observations @xcite and computing the cdm+baryons transfer function . on the other hand , the wiggles visible in the latest 2df power spectrum
are currently not considered as particularly significant by the authors @xcite . as an aside , it remains still a fascinating issue to understand whether one of these features , if confirmed , could be related to the `` periodic '' pattern detected more than 10 years ago by broadhurst and collaborators @xcite in their pencil beam surveys at the galactic poles ( see @xcite for more discussion of past results , and @xcite for similar tentative indications from the 2df qso survey ) .
[ fig : sim_clus ] ( from @xcite ) shows in a clear way how sensitive the evolution of structure is to the cosmological model , and in particular to the value of @xmath36 .
clusters of galaxies basically describe the growth of structures within comoving volumes of @xmath52 size , and one can notice how their abundance at even moderate redshifts is strikingly different in the two scenarios : massive systems are already quite rare at @xmath53 and substantially absent at @xmath54 in the einstein de sitter model .
this figure provides one of the main motivations for the deep surveys of clusters to @xmath55 and beyond , that have been performed in recent years based mainly on the rosat pspc archive .
such surveys constructed samples of serendipitously - observed clusters which have been used to measure directly the cosmic evolution of the mean density of x - ray luminous systems ( see @xcite and @xcite for reviews ) .
the lack of evolution in the x - ray luminosity function ( xlf ) for @xmath56 erg s@xmath9 out to @xmath57 resembles the situation of the top panel of fig .
[ fig : sim_clus ] ( once we translate temperatures into luminosities ) , thus favouring low values for @xmath36 ( see @xcite for details ) . at the same time , these results confirm the early findings from the emss @xcite of mild evolution at the bright end of the xlf @xcite .
[ fig : xlf ] plots a comparison of the latest xlf from the reflex survey , which represents the state of the art in the local ( @xmath58 ) universe @xcite , to measurements at higher redshift @xcite . despite these strong observational efforts ,
however , the number of high redshift x - ray selected clusters is still disappointingly small ( @xmath59 known above @xmath60 ) .
also , essentially all current large samples are based on the same rosat
pspc instrument , including the most recent attempt to exploit the large area of the rass to find massive clusters beyond the reflex limit ( the macs survey at @xmath61 @xcite ) .
a different approach has been taken by a new survey , recently started after a thorough re
analysis of the higher resolution rosat
hri archive , a database so far neglected for cluster searches .
this is the bmw ( brera multiscale wavelet ) cluster survey @xcite . although still in the early phases of its optical follow up , the bmw survey , which features a very interesting sky coverage to farly faint fluxes ( @xmath62 sq.deg around @xmath63 erg @xmath10 s@xmath9 , 1 sq.deg at @xmath64 erg @xmath10 s@xmath9 ) , is producing very promising first results ( two clusters above @xmath60 and a few @xmath55 candidates , see also http://www.merate.mi.astro.it/@xmath65guzzo/bmw/gallery.html ) and is expected to provide an important complementary `` bridge '' between the pspc surveys and future ( @xmath66 ) surveys based on chandra and xmm data ( see e.g. pierre , this volume ) .
abell , g.o . , 1958 , 3 , 211 bahcall , n.a . ,
soneira , r.m . , 1983 , 270 , 20 benoist , c. , et al . , 1999 , 514 , 563 birkinshaw , m. , 1999 , phys .
310 , 97 bhringer , h. , et al .
( the reflex team ) , 2001 , 369 , 826 bhringer , h. , et al .
( the reflex team ) , 2001 , , in press ( astro - ph/0106243 ) borgani s. , guzzo l. , 2001 , _ nature _ 409 , 39 borgani , s. , plionis , m. , kolokotronis , e. , 1999 , 305 , 866 borgani , s. , et al . , 2001 , in press ( nov 2001 issue , also astro - ph/0106428 ) broadhurst , t.j , et al . , 1990 , _
nature _ 343 , 726 collins , c.a . , et al .
( the reflex team ) , 2000 , 319 , 939 dalton , g.b . , et al . , 1994 , 271 , 47 dalton , g.b . , et al . , 1997 , 289 , 263 de bernardis , p. , et al . , 2000 , _
nature _ 404 , 955 della ceca , r. , et al . ,
2000 , 353 , 98 donohue , m. , et al .
, 2001 , 552 , l93 ebeling , h. , et al . , 2001 , 553 , 668 eisenstein , d.j . , hu , p. , 1998
, 496 , 605 ellis , r. , 2001 , in `` lectures of the xith canary island winter school of astrophysics '' , ( astro - ph/0102056 ) gal , r.r , 2001 , phd thesis , california inst . of technology giavalisco , m. , et al . , 1998 , 503 , 543 gioia , i. , et al . , 1990 , 72 , 567 gioia , i. , et al . , 2001 , 553 , l105 giovanelli , r. et al
, 1986 , 300 , 77 girardi , m. , et al .
, 1998 , 505 , 74 governato , f. , et al . , 1998 , _
nature _ 392 , 359 guzzo , l. , et al .
, 1997 , 489 , 37 guzzo , l. , 2000 , in _ xixth texas symposium on relativistic astrophysics & cosmology _ , nucl .
b ( proc . suppl . ) vol.80 , p.233 ( 09/06 ) , e. aubourg et al .
( astro - ph/9911115 ) guzzo , l. , 2001 , in `` clustering in the universe : from highly nonlinear structures to homogeneity '' , _ graduate school in contemporary relativity and gravitational physics _ , v. gorini & u. moschella eds .
, in press .
( astro - ph/0102062 ) guzzo , l. et al .
( the reflex team ) , 1999 , _ the messenger _ 95 , 27 guzzo , l. et al .
( the reflex team ) , 2001 , in preparation hawkins , e. , et al . , 2001 , 325 , 589 henry , j.p . , et al . , 1992 , 386 , 408 henry , j.p . , 2001 , in `` amiba 2001 : high - z clusters , missing baryons , and cmb polarization '' , in press ( astro - ph/0109498 ) hoyle , f. , et al .
, 1999 , 309 , 659 kaiser , n. , 1984 , 284 , l9 kim , s.j . , 2001 , phd thesis , princeton university klypin , a.a . , kopylov , a.i . , 1983 , _ sov
_ 9 , 41 lahav , o. , et al . , 1989 , 238 , 881 lazzati , d. , et al . , 1999 , 386 , 408 lin h. , et al .
, 1996 , 471 , 617 lumsden , s.l . , et al . , 1992 , 258 , 1 miller , c.j . , melott , a.l .
, nichol , r.c . ,
preprint ( astro - ph/9912362 ) miller , c.j . ,
nichol , r.c . ,
batuski , d.j . , 2001 , 555 , 68 mo , h.j .
& white s.d.m . , 1996 , 282 , 347 moretti , a. , et al .
( bmw team ) , 2001 , in `` x - ray astronomy 2000 '' , r. giacconi et al .
, in press ( astro - ph/0103348 ) moscardini , l. , et al . , 2000 , , 314 , 647 moscardini , l. , et al . , 2000 , 316 , 283 mushotzky , r.f . ,
scharf , c.a .
, 1997 , 482 , l13 nichol , r.c . , et al . , 1992 , 255 , p21 nichol , r.c . , et al . , 1994 , 267 , 771 nichol , r.c
. et al . , 1999 , 521 , l21 padilla , n.d . ,
baugh , c.m .
, 2001 , submitted ( astro - ph/0104313 ) percival , w. , et al .
( the 2df team ) , 2001 , 327 , 1297 reiprich , t.h . ,
bhringer , h. , 1999 , _ astr .
_ 320 , 296 romer , et al .
, 1994 , _ nature _ 372 , 75 rosati , p. , et al . , 1995 , 445 , l11 rosati , p. , et al .
, 1998 , 492 , l21 rosati , p. , et al .
, 2000 , in `` large scale structure in the x - ray universe '' , plionis , m. & georgantopulos , i. eds .
( paris : atlantisciences ) , p. 13
roukema , b.f .
, et al . , 2001 , submitted ( astro - ph/0106135 ) schuecker , p. , et al .
( the reflex team ) , 2001 , 368 , 86 schuecker , p. , et al .
( the reflex team ) , 2001 , in preparation sheth , r.k .
, mo , h.j . , tormen , g. , 2001 , 323 , 1 sutherland , w. , 1988 , 234 , 159 suto , y. , et al . , 2000 , 534 , 551 tozzi , p. , norman , c. , 2001 , 546 , 63 tucker , d.l .
et al . , 1997 , 285 , l5 peacock , j.a . , et al .
( the 2df team ) , 2001 , _ nature _ 410 , 169 york , d.g . , et al .
( the sdss collaboration ) , 2000 , 120 , 1579 vikhlinin , a. , et al .
, 1998 , 498 , l21 vogeley , m.s . ,
szalay , a.s .
, 1996 , 465 , 34 voges , w. , et al . , 1996 , in _ rntgenstrahlung from the universe _ , ( h.u .
zimmermann et al . ) , mpe report no .
263 , p. 637
yair , s. , branchini , e. , dekel , a. , 2000 , 540 , 62 zwicky , f. , et al .
, 1961 - 68 , _ catalogue of galaxies and of clusters of galaxies _ ( pasadena : california institute of technology )
|
i review recent progress in the study of the large - scale structure of the universe through the distribution of clusters of galaxies , concentrating on new results using x - ray selected samples . after discussing the importance of understanding the properties of the tracers used to map structure and their relation to the underlying mass ,
i elaborate on the advantages and disadvantages of clusters of galaxies to this end .
i then present the most recent estimates of the power spectrum and correlation function of x - ray clusters in the local ( @xmath0 ) universe , and their implications for cosmological models .
finally , i briefly summarize most recent results from deep x - ray surveys as probes of the evolution of structure and highlight current ongoing observational efforts in this field . *s * _ med 3j_3
|
distance estimates to celestial objects are one of the most fundamental measurements in astronomy and astrophysics .
knowledge of the distance to astronomical sources in the sky are needed in order to estimate their physical properties , such as luminosities , masses , kinematics , and dynamics . in recent years
, the technique of very long baseline interferometry ( vlbi ) has been used to make precise astrometric measurements ( to a precision of of a few tens of microarcseconds for a single observation ) and thus holds the promise to extend the range of the direct distance measurements of annual parallax up to at least 10 kpc with 10% accuracy using radio telescopes such as the very long baseline array ( vlba ) of the national radio astronomy obsevatory ( nrao ) .
compact , bright radio - emitting objects such as pulsars and masers are choice beacons for vlbi distance measurements across the galaxy ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
such distance measurements enable enormous improvements in our understanding of galactic structure and also the physics of individual objects .
@xcite reviewed the importance of measuring parallaxes and proper motions to help delineate galactic structure and showed , in a survey for parallaxes and proper motions of young , high - mass stars , that kinematic distances are systematically too large . in this paper
, we report on using the vlba to measure the annual parallax of the star - forming region iras 00420@xmath15530 , using the bright water masers associated with the iras source .
iras 00420@xmath15530 @xmath6 , @xmath7 is a star formation region exhibiting a molecular outflow @xcite , 3.6 cm and 3 mm continuum emission @xcite , and h@xmath0o masers ( e.g. , * ? ? ?
it is also coincident with dense gas traced by hco@xmath8(1 - 0 ) , indicating that the cluster of young stars remains deeply embedded in its natal cloud of molecular gas @xcite .
the systemic velocity of the gas around iras 00420@xmath15530 is @xmath951 km s@xmath10 with respect to the local standard of rest ( lsr ) based on the measurements of ammonia radial velocities @xcite and the hco@xmath8(1 - 0 ) @xcite .
the water masers occur at similar line - of - sight velocities as the surrounding gas , but with a somewhat broader range ( lsr velocities @xmath552 to @xmath540 km s@xmath10 ) . distances to iras 00420@xmath15530 of 4.3 to 7.7 kpc ( presumably kinematic , in most cases ) have been used in recent papers , despite a much closer photometric distance of 1.7 kpc @xcite . reflecting this broad range of distance estimates , luminosities of the iras source ( presumably the exciting star ) between @xmath11 and @xmath12
have been reported ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . clearly , a better distance estimate is desirable for this object .
in addition , iras 00420@xmath15530 lies in the direction of the perseus arm , which has been the topic of a recent distance study by @xcite .
@xcite addressed the large discrepancy between the luminosity and kinematic distance estimates in the perseus arm by accurately measuring the distance ( using vlbi techniques ) of the methanol masers in the compact hii region w3(oh ) .
it is important that distances to other objects in the direction of the perseus arm be precisely measured to compare to the work presented by @xcite , and to determine small - scale deviations from the distances and kinematics in this region which could be due to peculiar motions of the studied objects .
we present here a careful analysis of the observing style , calibration techniques , data reduction and parallax / proper motion fitting in order to provide the error analysis required to derive the most precise measurements of the distance to iras 00420@xmath15530 .
between 2005 november and 2006 september , we made twelve 4-hour vlba observations ( approximately monthly ) of the h@xmath0o masers in iras 00420@xmath15530 to measure their annual parallax and proper motion .
the large number of monthly epochs were chosen to ensure that the parallax and proper motion fits would not be confused by maser spots fading and disappearing between epochs . in the end
, most of the spots persisted through most epochs , and this was not a problem .
the observations were made from the vlba s dynamic observing queue , wherein our standardized observing script was initiated opportunistically and periodically in real - time according to availability of the array , scientific ranking ( relative to fixed - schedule programs and other dynamically scheduled observations ) and other operational constraints , such as observing direction and weather . since iras
00420@xmath15530 is at relatively high declination ( and thus circumpolar at most vlba sites ) , our four - hour observation tended to rise to the top of the dynamic queue quite easily and often as a convenient schedule - filler , and so the interval between epochs was often shorter than one month .
the properties of the observations are listed in table [ tab1 ] .
cllll a & 2005 nov 24 & 02:3006:30 & @xmath13 @ -17.5@xmath14 & brewster did not participate + b & 2005 dec 19 & 00:5104:51 & @xmath15 @ -16.6@xmath14 & + c & 2006 jan 08 & 23:3303:33 & @xmath16 @ -16.3@xmath14 & + d & 2006 jan 26 & 22:2202:22 & @xmath17 @ -23.7@xmath14 & + e & 2006 feb 22 & 20:3200:32 & @xmath18 @ -18.3@xmath14 & + f & 2006 mar 17 & 19:0123:01 & @xmath19 @ -19.3@xmath14 & + g & 2006 apr 08 & 17:3521:35 & @xmath20 @ -14.3@xmath14 & + h & 2006 may 08 & 15:3719:37 & @xmath21 @ -24.9@xmath14 & + i & 2006 may 31 & 13:5117:51 & @xmath22 @ -13.3@xmath14 & + j & 2006 jun 22 & 12:4016:40 & @xmath22 @ -15.3@xmath14 & + k & 2006 aug 04 & 09:5113:51 & @xmath23 @ 21.7@xmath14 & mauna kea did not participate ; epoch not used + l & 2006 sep 02 & 07:5711:57 & @xmath24 @ -16.7@xmath14 & + four 16 mhz spectral windows were simultaneously observed in the vlba s 22 ghz band .
one spectral window ( the second ) was set up to cover the h@xmath0o maser lines in iras 00420@xmath15530 at @xmath2546 km s@xmath10 .
the other three windows were distributed across the 500 mhz instantaneous if bandwidth of the vlba to provide optimal delay sensitivity for calibration of the troposphere .
in addition to iras 00420@xmath15530 , nine compact background radio - loud quasars were observed as calibrators .
two of these calibrators were for phase - referencing purposes to provide for accurate relative astrometry ( see table [ tab2 ] and figure [ calgeo ] ) .
the first of these calibrators ( j0042@xmath15708 , @xmath41.41@xmath14 from iras 00420@xmath15530 ) was observed for primary phase referencing . for most of the observation ,
iras 00420@xmath15530 and j0042@xmath15708 were observed alternately on a @xmath460s timescale ( 30s dwell on each ) .
once per hour , an alternate calibrator ( j0047@xmath15657 , @xmath40.66@xmath14 from j0042@xmath15708 ) was observed to permit verification of the phase - referencing calibration .
lllll iras 00420@xmath15530 & 00@xmath2644@[email protected] & [email protected] & up to @xmath470.0 & science target : h@xmath0o masers + j0042@xmath15708 & 00@xmath2642@[email protected] & [email protected] & @xmath40.18 & primary phase - reference calibrator + j0047@xmath15657 & 00@xmath2647@[email protected] & [email protected] & @xmath40.10 & secondary phase - reference calibrator + the other seven calibrators , chosen from the vlba calibrator list @xcite were observed as a group once per hour ( five times in four hours ) to enable determination of the residual macroscopic troposphere delay error at each antenna .
the accuracy of phase - referencing calibration at the vlba is limited by an zenith angle - dependent delay error arising from inaccuracies in the macroscopic troposphere model used at the vlba correlator @xcite .
the seasonal zenith troposphere delay model is typically in error by many centimeters , and this leads to a significant time - dependent delay error between the science target and phase - reference calibrator since they are observed at slightly different zenith angles . as a result , the phase - referencing determined for the calibrator does not fully calibrate the target unless this macroscopic component is compensated .
this effect is calibrated by observing a set of bright ( @xmath29 jy ) calibrators at a wide range of zenith angles and fitting the observed residual delays for the zenith troposphere error , assuming the delay increases away from the zenith as @xmath30 , where @xmath31 is the zenith angle . the standard approach ( e.g. , * ? ? ?
* ) is to select approximately one dozen such calibrators distributed over the entire sky and observe them 1@xmath53 times over the course of the observation to get a few instantaneous measurements of the zenith delay .
such an approach does not permit tracking the delay error in time very accurately and makes no allowance for any azimuth dependence of the delay error , such as might be expected as weather changes over an antenna .
our approach was to select a smaller set of calibrators distributed sufficiently in zenith angle ( @xmath32 ) at the approximate azimuth of our target , and observe them more continuously during the observation .
after each observation , the data were correlated in two passes .
the first pass correlated all four spectral windows at low spectral resolution ( 32 channels ) .
the second pass correlated only the spectral window containing the h@xmath0o maser lines with 1024 channels , yielding a channel spacing of 15.625 khz , and thus a spectral resolution in the lsr of 18.75 khz , or 0.253 km s@xmath10 .
the visibility calibration was performed in nrao aips @xcite , following the exact same procedure for each epoch .
the low - spectral - resolution data were processed first , as follows .
first , several standard a priori calibrations were applied , including revised earth orientation parameter corrections ( eop , important for good astrometry ) , parallactic angle phase , quantized sampling bias amplitude corrections , and system temperature / gain amplitude calibration .
second , a scan on one of the bright macroscopic troposphere calibrators ( j2202@xmath14216@xmath33bl lac , @xmath43.0 jy ) was selected to serve as a reference for all subsequent phase and delay calibration solutions . for this scan , the net residual delay and
non - linear bandpass was determined in each spectral window .
this solution , applied to the rest of the data , removes constant instrumental phase and delay errors , introduces a tropospheric delay residual offset ( whatever the troposphere delay was toward j2202@xmath14216 in this scan ) , and leaves behind only relative ( and relatively small ) delay residuals that are time- and direction - dependent .
these errors include any residual instrumental clock drift , zenith - angle - dependent relative macroscopic troposphere delay errors ( to be solved using the seven bright calibrators ) , and time - dependent ( phase- ) delay errors ( to be phase - referenced ) .
the macroscopic troposphere delay residual calibration is obtained by applying the calibrations described above and solving for the phase in each of the four spectral windows and in each of the scans on the seven bright calibrators .
these phases are then fit for multi - band delays . in general , the likelihood of ambiguity resolution problems in the multi - band delay determination are small , since these delays are residuals , and the expected zenith delay errors are expected to be only a fraction of a nanosecond .
the multi - band delays are then used to jointly solve for the residual instrumental clock drift and time - dependent zenith troposphere delay error , assuming the @xmath30 model .
the separation of these two effects depends upon observing the calibrators out of zenith angle order , else a monotonic clock drift would be difficult to distinguish from a component of the elevation - dependent troposphere delay residual .
solutions with constant , linear , quadratic , and cubic time - dependence ( and all combinations ) in both clock and troposphere delay terms were attempted .
the higher - order fits tended to diverge .
optimal fits were obtained for quadratic troposphere and linear clock .
typical zenith delay residuals were found to be a few to 15 cm , with variations over four hours at the 1@xmath52 cm level .
the overall accuracy of this delay calibration is approximately [email protected] cm , and is limited by the simplicity of the zenith angle dependence and unmodeled local time - variation of the troposphere delay ( of the sort we would otherwise phase - reference ) . since this is a residual effect , negative troposphere delay solutions are possible .
also , the troposphere delay solution is not referenced to any specific antenna since the zenith angle dependence and differential zenith angle sampling at each antenna breaks the usual degeneracy .
figure [ mtpherr ] shows , for epoch a , the systematic phase - referencing errors for iras 00420@xmath15530 and j0047@xmath15657 due to their time - dependent zenith angle difference relative to j0042@xmath15708 and the ( mean ) zenith delay model calibration determined as described above .
the zenith delay errors range in magnitude from 0.6 cm for kp to 13.5 cm for hn ( br did not participate , and sc was excluded ) .
the errors are generally larger for iras 00420@xmath15530 since it has a larger separation from j0042@xmath15708 , and at a relatively unfavorable position angle ( near transit ) that maximizes the zenith angle difference . only for the easternmost antennas in the array ( hn , nl ) , which are observing farthest from transit ,
is the position angle of j0047@xmath15657 relative to j0042@xmath15708sufficient to excessively amplify this systematic phase error . the vlba station at st .
croix ( sc ) was generally the worst performing station in the delay calibration , yielding poor multi - band delay fits and/or poor zenith delay solutions .
this is caused by excessive and highly variable tropospheric water vapor at sc s sea - level location and humid climate .
since this station also yields the least accurate phase referencing solutions for the same reasons , it was excised from all epochs .
this results in the loss of many relatively long baselines , but the poor calibration of these baselines would be of questionable value . at this point , the high - spectral - resolution data were processed , first with all of the same _ a priori _ calibrations as the low - resolution data .
the tropospheric delay calibration solution ( including the clock drift ) was then transferred from the low - resolution dataset , to be used as an additional a priori calibration for the high - resolution data .
the same strong calibrator scan was selected to obtain the global instrumental reference calibration , except a polynomial bandpass was used since the per - channel snr was insufficient for a sampled bandpass at the high spectral resolution .
finally , the phase - reference calibration was determined on j0042@xmath15708 .
the full calibration was then applied to iras 00420@xmath15530 , j0042@xmath15708 , and the secondary phase - reference calibrator , j0047@xmath15657 . after re - sampling the velocity axis for each epoch to the lsr
, the data were exported in uvfits format .
the data for each epoch were then imaged in difmap @xcite , selecting the channels with maser emission .
first , a relatively low spatial resolution image of a wide field - of - view was generated to spatially locate significant maser regions , then each of these regions were imaged at full resolution .
a simple multi - field clean deconvolution script was developed in difmap to optimize the imaging of channels with signal in more than one region .
figure [ skyfig ] shows the spatial distribution of the maser regions ( integrated clean components from epoch a ) , integrated in velocity .
there are nine major maser spot regions , two of which ( 4 and 7 ) contain two distinct maser spots , with significant emission in a total of 44 channels .
the typical imaging sensitivity achieved per channel was 100@xmath5200 mjy / beam , with maser peaks ranging up to @xmath475 jy beam@xmath10 .
figure [ integspecfig ] shows the spectrum of iras 00420@xmath15530 integrated over all maser spot regions , for a representative set of epochs ( a , d , g , and j ) , indicating the overall variability of the source .
figure [ dynspecfig ] shows separately the time - dependent spectra for maser regions 14 . a variety of spectral structure and behavior is observed .
regions 1 and 3 show strong , stable , and simple spectral structure .
region 2 shows a monotonic drift of five channels in the peak of emission .
region 4 shows two distinct spectral components : the feature at @xmath34 km s@xmath10 drifts by about one channel and fades , while the feature at @xmath35 km s@xmath10 , indistinct at epoch a , grows in strength with epoch and is stationary in velocity .
the remaining regions ( not shown ) are generally weaker , and show a similar range of structure and variability . of these
, maser region 5 is the most stable .
as shown below , the spectral structure and variability has serious consequences for obtaining reliable parallax and proper motion fits .
5530 at epochs a , d , g , and j. [ integspecfig ] ] ] to obtain an estimate of the fundamental phase - referencing accuracy of these observations , continuum images ( bandwidth : 14 mhz ) of the secondary phase - reference calibrator , j0047@xmath15657(snr@xmath420 ) , were generated .
for the ensemble of epochs ( excluding epoch k ) , the rms peak position of j0047@xmath15657 is @xmath3630 @xmath37as in both right ascension and declination , with no evidence of any significant systematic motion .
( without the macroscopic troposphere solution applied , the declination rms increases by a factor of 2 , indicating the importance of this calibration . )
this estimate establishes a basis for estimating the astrometric precision ( per snr ) of the maser observations .
specifically , a snr of @xmath420 nominally corresponds to an astrometric precision of @xmath430 @xmath37as . this error estimate can not be transferred and scaled directly to the observations of iras 00420@xmath15530 .
a variety of systematic errors render themselves differently for iras 00420@xmath15530 and j0047@xmath15657 .
j0047@xmath15657 is only 0.66@xmath14 from j0042@xmath15708 , the primary calibrator , less than half the 1.41@xmath14 j0042@xmath15708iras 00420@xmath15530 separation , and so the phase - referencing calibration is nominally more accurate for j0047@xmath15657 .
furthermore , near transit at the center of the vlba , the position angle of the j0042@xmath15708j0047@xmath15657 separation is mostly in azimuth while that of j0047@xmath15657iras 00420@xmath15530 is mostly in zenith angle ( see figure [ calgeo ] ) , making any residuals in the zenith angle - dependent macroscopic troposphere calibration more important for iras 00420@xmath15530 . as described below , these and other systematic errors place more stringent limits on the astrometric precision of iras 00420@xmath15530 , despite the fact that most of the maser spots used in the parallax fitting have higher snr than j0047@xmath15657 .
parallax and proper motion calculations rely on measurements of the time - dependent direction of consistently identified , small , physically coherent objects relative to a distant , presumed stationary , background reference , in this case , the phase reference calibrator .
masers are among the most compact observable objects available , but at the resolution of the vlba , they are generally not strictly point - like .
furthermore , it is not clear to what degree masers can be characterized as distinct physical objects since they likely arise not from fixed discrete objects , but rather from coherent velocity structures within a presumably larger physical structure ( e.g. , gas cloud ) . indeed ,
most of the maser spots in iras 00420@xmath15530 are clearly not strictly point - like and are variable in their marginally resolved structure and flux density from epoch to epoch .
furthermore , the spectral structure of each maser spot is generally non - trivial , with the peak for some spots clearly drifting substantially in velocity from epoch - to - epoch .
these factors make reliable epoch - to - epoch identification of distinct physical features difficult .
therefore , rather than fitting complicated multiple point and resolved gaussian model components to the images or visibilities , and thereby introducing an additional unphysical component in the analysis , we have adopted the simple peak position ( fitted from image pixels ) as the best estimate of the maser position , and have accepted the inevitable intrinsic variation of this estimate ( a form of traditional confusion within the synthesized beam ) as an additional source of error in our analysis . for each of eleven distinct spots in 9 spot regions , and in each channel with significant power ( see below ) ,
we have fitted the sequence of peak positions for parallax and two - dimensional proper motion , according to the standard equation : @xmath38 where @xmath39 is the annual parallax , @xmath40 is time , @xmath41 is the 2-dimensional proper motion , @xmath42 denote the maser position at @xmath43 , and @xmath44 is a 2-dimensional function of direction and time describing the parallax ellipse ( the reflection of the earth s orbit onto the sky ) . in using this equation , we nominally assume that any non - linear internal motions with the maser spot regions are negligible .
this yielded 44 estimates of the proper motion and parallax of iras 00420@xmath15530 , and the results are listed in table [ tab3 ] . an example fit is shown in figures [ parallaxskyfig ] and [ parallaxtimefig ] . crrlrrrrrr 1 & -43.4 & 11 & abcdefghij l & * 0.44 * ( 0.02 ) & -2.06 ( 0.08 ) & -1.88 ( 0.05 ) & 32 & 45 & 55 + 1 & -43.6 & 11 & abcdefghij l & * 0.44 * ( 0.02 ) & -2.10 ( 0.07 ) & -1.95 ( 0.05 ) & 29 & 38 & 47 + 1 & -43.9 & 11 & abcdefghij l & * 0.44 * ( 0.02 ) & -1.94 ( 0.06 ) & -1.72 ( 0.04 ) & 31 & 29 & 42 + 1 & -44.1 & 11 & abcdefghij l & * 0.48 * ( 0.02 ) & -2.05 ( 0.07 ) & -1.63 ( 0.05 ) & 23 & 43 & 48 + 1 & -44.3 & 11 & abcdefghij l & * 0.49 * ( 0.03 ) & -2.09 ( 0.09 ) & -1.65 ( 0.07 ) & 28 & 56 & 62 + & & & & & & & & & + 3 & -45.8 & 11 & abcdefghij l & * 0.48 * ( 0.01 ) & -2.55 ( 0.05 ) & -0.89 ( 0.04 ) & 19 & 30 & 36 + 3 & -46.0 & 11 & abcdefghij l & * 0.45 * ( 0.01 ) & -2.52 ( 0.05 ) & -0.84 ( 0.04 ) & 19 & 31 & 36 + 3 & -46.2 & 11 & abcdefghij l & * 0.45 * ( 0.01 ) & -2.52 ( 0.05 ) & -0.89 ( 0.04 ) & 18 & 32 & 37 + & & & & & & & & & + 4a & -46.6 & 6 & abcdef & 0.50 ( 0.06 ) & -2.67 ( 0.13 ) & -1.92 ( 0.25 ) & 26 & 19 & 32 + 4a & -46.8 & 6 & abcdef & 0.49 ( 0.06 ) & -2.66 ( 0.13 ) & -1.97 ( 0.25 ) & 25 & 20 & 32 + 4a & -47.2 & 6 & abcdef & 0.50 ( 0.05 ) & -2.62 ( 0.10 ) & -1.93 ( 0.20 ) & 23 & 11 & 25 + & & & & & & & & & + 4b & -47.4 & 8 & defghij
l & * 0.43 * ( 0.02 ) & -2.34 ( 0.08 ) & -1.85 ( 0.08 ) & 18 & 27 & 32 + 4b & -47.6 & 8 & defghij l & * 0.44 * ( 0.02 ) & -2.40 ( 0.08 ) & -1.88 ( 0.08 ) & 22 & 23 & 32 + & & & & & & & & & + 5 & -43.2 & 10 & abcdefghij & * 0.50 * ( 0.02 ) & -2.65 ( 0.09 ) & -1.05 ( 0.06 ) & 32 & 19 & 38 + 5 & -43.4 & 10 & abcdefghij & * 0.47 * ( 0.03 ) & -2.48 ( 0.13 ) & -1.18 ( 0.08 ) & 40 & 30 & 50 + & & & & & & & & & + 7b & -43.9 & 7 & abcdefg & 0.62 ( 0.05 ) & -5.79 ( 0.17 ) & -0.53 ( 0.19 ) & 25 & 28 & 38 + 7b & -44.1 & 7 & abcdefg & 0.51 ( 0.04 ) & -5.33 ( 0.12 ) & -0.77 ( 0.14 ) & 20 & 18 & 27 + 7b & -44.3 & 7 & abcdefg & 0.51 ( 0.04 ) & -5.29 ( 0.12 ) & -0.77 ( 0.13 ) & 15 & 19 & 25 + 7b & -44.5 & 7 & abcdefg & 0.50 ( 0.04 ) & -5.29 ( 0.14 ) & -0.81 ( 0.14 ) & 21 & 17 & 27 + & & & & & & & & & + 8 & -47.4 & 7 & abcdefg & 0.59 ( 0.10 ) & -5.51 ( 0.31 ) & -1.23 ( 0.34 ) & 61 & 48 & 77 + 8 & -47.6 & 7 & abcdefg & 0.56 ( 0.09 ) & -5.71 ( 0.28 ) & -1.18 ( 0.33 ) & 63 & 36 & 73 + 8 & -47.9 & 7 & abcdefg & 0.48 ( 0.09 ) & -5.65 ( 0.27 ) & -1.39 ( 0.32 ) & 63 & 28 & 69 + 8 & -48.1 & 7 & abcdefg & 0.41 ( 0.08 ) & -5.57 ( 0.24 ) & -1.60 ( 0.28 ) & 54 & 23 & 59 + & & & & & & & & & + 9 & -49.3 & 8 & abcdefgh & * 0.43 * ( 0.04 ) & -5.01 ( 0.16 ) & -1.15 ( 0.13 ) & 35 & 30 & 46 + 9 & -49.5 & 10 & abcdefghij & * 0.43 * ( 0.03 ) & -4.99 ( 0.13 ) & -1.21 ( 0.09 ) & 32 & 41 & 52 + 9 & -49.8 & 10 & abcdefghij & * 0.48 * ( 0.03 ) & -5.17 ( 0.11 ) & -1.24 ( 0.08 ) & 36 & 26 & 44 + 9 & -50.0 & 8 & abcdefgh & * 0.48 * ( 0.03 ) & -5.14 ( 0.10 ) & -1.42 ( 0.09 ) & 20 & 20 & 28 + 2 & -44.5 & 10 & abcdefghij & 0.46 ( 0.04 ) & -1.21 ( 0.16 ) & -2.48 ( 0.11 ) & 32 & 56 & 65 + 2 & -44.7 & 10 & abcdefghij & 0.52 ( 0.04 ) & -1.53 ( 0.14 ) & -2.42 ( 0.09 ) & 34 & 47 & 58 + 2 & -44.9 & 10 & abcdefghij & 0.61 ( 0.04 ) & -1.79 ( 0.17 ) & -2.83 ( 0.10 ) & 36 & 57 & 67 + 2 & -45.1 & 10 & abcdefghij & 0.48 ( 0.07 ) & -1.22 ( 0.28 ) & -3.50 ( 0.16 ) & 58 & 100 & 115 + 2 & -45.3 & 10 & abcdefghij & 0.35 ( 0.10 ) & -0.72 ( 0.41 ) & -3.73 ( 0.24 ) & 117 & 120 & 167 + 2 & -45.5 & 10 & abcdefghij & 0.25 ( 0.15 ) & -0.92 ( 0.59 ) & -2.83 ( 0.36 ) & 144 & 239 & 279 + & & & & & & & & & + 6 & -44.7 & 10 & abcdefghij & 0.83 ( 0.51 ) & -4.40 ( 2.10 ) & -11.01 ( 1.11 ) & 161 & 901 & 915 + 6 & -44.9 & 10 & abcdefghij & 1.11 ( 0.44 ) & -5.34 ( 1.70 ) & -10.22 ( 0.93 ) & 251 & 779 & 819 + 6 & -45.1 & 10 & abcdefghij & 0.90 ( 0.45 ) & -4.61 ( 1.71 ) & -10.27 ( 0.99 ) & 211 & 797 & 824 + 6 & -45.3 & 10 & abcdefghij & 0.82 ( 0.42 ) & -4.28 ( 1.56 ) & -10.18 ( 0.95 ) & 199 & 796 & 820 + & & & & & & & & & + 7a & -43.2 & 7 & abcdefg & 0.16 ( 1.60 ) & -13.68 ( 5.07 ) & 0.25 ( 5.43 ) & 908 & 889 & 1271 + 7a & -43.4 & 7 & abcdefg & -2.89 ( 2.30 ) & 2.45 ( 6.92 ) & -14.79 ( 8.06 ) & 1307 & 1227 & 1793 + 7a & -43.6 & 7 & abcdefg & -0.43 ( 1.79 ) & -11.18 ( 5.47 ) & -3.06 ( 6.35 ) & 961 & 1014 & 1397 + 7a & -43.9 & 7 & abcdefg & -1.06 ( 1.89 ) & -8.59 ( 6.32 ) & -4.96 ( 6.32 ) & 873 & 1017 & 1340 + for the fitting , only channels with maser spot detections greater than @xmath412 jy ( corresponding to snr @xmath45 10 , typically ) were included .
additionally , after a trial fit , partial epoch ranges for some spots were discarded where it was evident that the assumption of a consistent discrete physical component following the relatively simple parallax and proper motion trajectory was violated .
presumably , the peak derived from our finite - resolution observations shifted abruptly as a result of intrinsic structural changes .
epoch k was excluded entirely since the mauna kea station , which provides the longest baselines , did not participate .
the fits for three maser regions ( 2 , 6 , & 7a ) were rejected outright from further analysis on the basis of poor consistency among channels or unphysical ( negative ) parallax estimates .
each of these regions show substantial spectral complexity , including variation in the velocity of peak emission , indicating that they are non - ideal test - particles for parallax measurements .
no degree of judicious epoch selection makes these maser regions recoverable for parallax estimation purposes .
determining proper epoch- and maser - dependent weights for the fitting process is non - trivial , since accurate _
total _ errors , including systematics , for the peak position estimates are not readily available _ a priori_. formally , for a well - isolated point source with no systematic errors , the precision of the peak position estimates would be inversely proportional to the snr of the peak emission , and the resulting formal weights would be proportional to the _ square _ of the snr . in practice , such an estimate of astrometric precision must be a lower limit due to systematic errors arising from the complicating maser structure considerations noted above , as well as to per - epoch residuals in the macroscopic troposphere and phase - referencing calibration .
additionally , the wide range of flux densities among maser regions and epochs ( @xmath4260 jy ) result in a range of weights of 2 orders of magnitude or more for some maser spots .
such weights could have a deleterious effect on the effective sampling of the parallax ellipse during the year , and lead to complicated biases in the parallax estimates .
therefore , we adopted weights proportional to snr ( i.e. , the square root of the formal weights ) for the final fits .
for most of the fits , this choice was not consequential ; use of the formal weights yields parameter estimates not significantly different from that obtained with snr weights .
our parallax measurement is not limited by per - epoch snr , but rather by _ per - epoch _ systematic errors in the astrometry .
while the proper motions may be expected to vary among the spot regions ( and even among channels in a single region ) due to intrinsic internal motions , the parallax estimates should be consistent for all of them since the range of maser emission is very small ( @xmath46 mas ) relative to its distance ( kiloparsecs ) , and is unlikely to be significantly deeper than it is wide . in practice ,
the variation of the parallax estimates among maser spot regions and channels is more complicated .
while several ( but not all ) maser regions show impressive internal consistency , the dispersion of parallax estimates over the entire ensemble is considerably larger than the formal errors would imply .
this fact is related to residual systematic errors affecting the fits which originate in the complexity and evolution of the spectral and spatial structure of the maser regions , residual calibration errors , and the details of the time - dependent sampling of the parallax ellipse . at any single epoch
, calibration errors will affect different maser spot regions uniformly , and so are systematic , i.e. , relative position measurements among masers at one epoch have no information about the magnitude of such errors .
over many epochs ( up to eleven , in this case ) , however , we assume such errors can be treated as approximately stochastic . similarly , after excising obviously discrepant epochs as described above , we treat any _ non - linear _ intrinsic variations in the peak position due to internal structural and flux density evolution as approximately stochastic over many epochs .
these errors limit the precision of the parallax estimate , but are unlikely to bias it substantially . with up to eleven epochs per fit , we assert that the rms residuals determined in the fits are a reasonable estimate of the mean per - epoch astrometric precision achieved in our observations , typically @xmath47as in both dimensions .
the error bars in figure [ parallaxtimefig ] reflect the relative weights used in the fit , and are conservatively scaled so that their minimum is 30 @xmath37as . of the remaining 27 fits in 8 maser regions ,
most yield internally consistent parallax and proper motions ( with channel ) , but the overall range of parallax results remains relatively large : 0.41 - 0.62 mas , with formal errors in the range 0.01 - 0.1 mas ( see figure [ parallaxhist ] which shows the distribution in the range 0.40 - 0.52 mas ) .
in addition to several obvious outliers ( there are three such outliers greater than 0.55 mas and not shown in figure [ parallaxhist ] ) , there is some evidence of maser region - dependent systematic errors .
a careful analysis of the time - dependent weighted sampling shows that those maser regions for which the later epochs were undetected or discarded tend to yield systematically higher parallax estimates ( near or in excess of 0.50 mas ) compared to those with more complete sampling .
fitting trials on the completely sampled maser regions ( e.g. , region 3 ) using _ only _ the early epochs confirm this effect .
we therefore have discarded maser regions 4a , 7b , and 8 as insufficiently sampled and inadequate for further consideration .
the remaining 16 maser region channels are shown in the filled histogram of figure [ parallaxhist ] .
the distribution still appears bimodal , but the two states can not be segregated according to maser region .
the unweighted dispersion of the remaining measurements is 0.02 mas , broadly consistent with the typical formal errors in the individual fits , which arise from a combination of pure statistics and per - epoch systematics that are stochastic over many epochs .
the individual parallax solutions for the remaining 16 maser region channels have been averaged with weights derived from their formal errors to yield a net parallax estimate of [email protected] mas .
figure [ finalest ] shows the final parallax signature in right ascension and declination as a function of time ( bold curve ) also plotted are the per - channel data ( filled and open dots , with error bars as in figure [ parallaxtimefig ] ) and the range of individual channel fits ( hashed region ) ; the fitted proper motions and relative position offsets from the individual solutions have been removed .
it is clear that several epochs ( e.g. , epoch c in right ascension , epochs i and l in declination ) are affected by a substantial residual systematic offset in one dimension or the other , but that for most epochs , the magnitude of these systematics is similar or smaller than the statistical distribution of measurements .
since the parallax solution has been derived directly from the data , it is difficult to accurately ascertain the true balance between the per - epoch systematic and statistical contributions .
although the formal errors in each individual maser channel solution nominally account for both sources of error ( assuming the per - epoch systematics are approximately stochastic over many epochs ) , we nonetheless conservatively double the formal error estimate and adopt a final parallax estimate of [email protected] mas .
the distance to iras 00420@xmath15530 is therefore [email protected] kpc , a measurement with a precision of @xmath33% .
with direct measurements of the distance and three - dimensional space velocity of objects such as iras 00420@xmath15530 using the vlba , it becomes possible to more accurately evaluate the veracity of the kinematic distances derived from galactic rotation curves , and to do so over a larger portion of the galaxy than with any other available means .
this is especially important outside the solar circle where rotation curve analyses have had to rely principally on photometric and other indirect distance estimates .
a large range of distance estimates for iras 00420@xmath15530 have been used in the literature for various astrophysical analyses .
most tabulated distances have been in the range @xmath48@xmath49 kpc , and are presumably kinematic in origin .
our h@xmath0o maser parallax distance of 2.17 kpc is , in fact , most consistent with the oldest published distance estimate , a photometric distance of 1.7 kpc @xcite . the direct geometric distance estimate reported here is more than a factor of two closer than the kinematic distance of @xmath44.7 kpc implied by a radial velocity of @xmath50 km s@xmath10 using the rotation curve of @xcite (
henceforth , bb93 ) .
however , those authors acknowledge systematic offsets from their rotation curve in the sample they use to derive it ( hii regions and reflection nebulae ) , including the fact that it overestimates the ( galactocentric ) orbital velocities of most objects within @xmath43 kpc outside the solar circle .
furthermore , at the galactic longitude and distance of iras 00420@xmath15530 , in particular , their line - of - sight velocity residuals show a systematic deficit of @xmath425 km s@xmath10 ; their rotation curve nominally predicts a radial velocity of @xmath525 km s@xmath10 for iras 00420@xmath15530 , @xmath421 km s@xmath10 slower than observed .
similar large - scale systematic residuals occur at many outer galaxy longitudes , and they attribute this to non - circular galactic orbits , or more generally , the inadequacy of an axisymmetric model for the velocity field of the outer galaxy . in short , although the rotation curve predicts galactic orbital velocities with errors of only 10@xmath515% ( out of 200@xmath5250 km s@xmath10 ) , it is a much poorer predictor of _ line - of - sight _ velocities measured at the sun , and so kinematic distance estimates for the outer galaxy will be generally unreliable . since our vlba measurements provide the three - dimensional space velocity ( relative to the sun ) of iras
00420@xmath15530 , it is possible to explore the nature of its galactic orbit in detail . adopting the maser component in region 3 , @xmath51 km s@xmath10 as representative for iras 00420@xmath15530 we find a _
heliocentric _ line - of - sight velocity of @xmath52 km s@xmath10 , and , at a distance of 2.17 kpc , a ( heliocentric ) physical velocity of @xmath53 km s@xmath10 in galactic longitude and @xmath54 km s@xmath10 in galactic latitude ( at the distance of iras 00420@xmath15530 , 1 milliarcsecond year@xmath10 @xmath33 10.3 km s@xmath10 ) .
figure [ galplanobs ] illustrates the in - plane components of these measurements in a plan view of the region of the galaxy containing the sun and iras 00420@xmath15530 .
formally , the error in these velocities is @xmath40.5 km s@xmath10 , but there is a full range of @xmath410 km s@xmath10 in the line - of - sight and latitude components , and somewhat less in the longitude components among the brightest maser spot regions .
we assume tentatively that the internal motion of this component is not large enough to excessively bias our analysis ( this point will be addressed in greater detail below ) .
the following analysis is limited by the accuracy of knowledge of the absolute orbital speed of the galaxy and the sun s peculiar motion relative to the lsr .
we have provisionally assumed standard iau values for the lsr parameters , namely that it is @xmath55 kpc from the galactic center and moving at @xmath56 km s@xmath10 on a circular orbit @xcite .
the sun s peculiar motion relative to the lsr has been assumed to be @xmath57 km s@xmath10 ( toward the galactic center ) , @xmath58 km s@xmath10 ( along the orbit ) , and @xmath59 km s@xmath10 ( out of the plane ) as measured by @xcite using hipparcos data .
the calculations have be done in three dimensions , but with no correction for the sun s unknown distance from the galactic plane . converting to the galactocentric frame by correcting for @xmath60 and solar peculiar motions
, we find that iras 00420@xmath15530 is moving toward the galactic center at 23.0 km s@xmath10 , around the galaxy ( circular component ) at 202.0 km s@xmath10 , and toward the plane ( from below ) at 5.8 km s@xmath10 .
the in - plane components of iras 00420@xmath15530 s velocity vector are shown in figure [ irasgal ] .
thus , iras 00420@xmath15530 is moving significantly slower than the circular speed of @xmath4224 km s@xmath10 predicted by the bb93 rotation curve , and has a substantial non - circular component toward the galactic center . by assuming a nominal orbital speed for iras 00420@xmath15530
, we can shift our velocity measurements to a local standard of rest orbiting the galaxy at the position of iras 00420@xmath15530 .
figure [ irascorot220 ] shows the resulting in - plane peculiar motion of iras 00420@xmath15530 , assuming its orbital speed is 220 km s@xmath10 .
( the bb93 rotation curve prediction is insignificantly larger ; essentially we nominally assume the rotation curve is flat in this region of the galaxy . ) also shown is the peculiar motion of w3(oh ) ( also in its _ own _ standard of rest ) as reported by @xcite , also assuming an orbital speed of 220 km s@xmath10 .
the peculiar motions of iras 00420@xmath15530 and w3(oh ) are in almost exactly the same direction , and very nearly in the direction of the sun . combined with the large - scale systematic line - of - sight velocity deficit observed in the bb93 rotation curve residuals ,
this is highly suggestive of a large - scale anomalous motion of the material in this region of the galaxy , i.e. , the perseus spiral arm , in a direction toward the sun .
such anomalous motion could be related to the formation and propagation of the perseus spiral arm , and is evidence of the inadequacy of axisymmetric models of galactic rotation .
an obvious alternative to this conclusion is that the sun has a large velocity component in the direction of these objects , either in terms of the lsr orbital speed , the solar peculiar motion , or both .
indeed , the orbital speed component of the sun s velocity is the least well - understood component of its motion due to uncertainties in the overall scale of the galaxy ( @xmath61 and @xmath60 ) , and the component of solar peculiar motion in the direction of galactic rotation .
however , the velocity adjustments required of the alternative scenarios are generally implausible . increasing the galactic orbital speed for the solar lsr to 250 km s@xmath10 ( the value favored by @xcite for @xmath62 kpc according to measurements of the proper motion of sgr a@xmath63 ) and maintaining the flat rotation curve out to iras 00420@xmath15530 and w3(oh ) reduces the radial ( toward the galactic center ) component of the anomalous velocity by only @xmath46 km s@xmath10 , without substantially changing the orbital component .
the anomaly is even less sensitive to adjusting _ both _ the orbital speed and @xmath61 in such a way as to keep the _
angular _ speed ( @xmath64 ) of the solar lsr constant @xcite .
this insensitivity is a result of the relatively small angle ( @xmath411@xmath14 ) subtended at the galactic center by the sun and iras 00420@xmath15530 , and that this angle changes very slowly for a fixed distance to iras 00420@xmath15530 ( our measurement ) over the plausible range of @xmath61 ( 8 @xmath36 1 kpc ) . introducing a difference in nominal orbital speed between the sun and these sources sufficient to account for the anomaly is also implausible , implying zero additional mass outside the solar circle , and would not compensate for the radial component of the anomaly . finally , a sufficient correction to just the solar peculiar motion would require a change to the radial component of @xmath420 km s@xmath10 , entirely inconsistent with the hipparcos measurements of @xcite .
it is also possible that we have failed to model a large internal motion of the h@xmath0o masers in iras 00420@xmath15530 that happens to lie in the direction of the apparent anomalous motion , a possibility noted by @xcite for w3(oh ) .
however , we consider this unlikely given the relatively modest range of three - dimensional velocities observed for the main h@xmath0o masers in iras 00420@xmath15530 , and that they are near ( @xmath65 5 km s@xmath10 ) the systemic velocity indicated by other molecular species in iras 00420@xmath15530 .
the maser channel chosen for this analysis ( region 3 , @xmath66 km s@xmath10 ) is reasonably representative of the available information for iras 00420@xmath15530 .
other choices are possible , but either can be excluded on grounds of insufficient constraints or do not substantively change the result .
the maser regions excluded from the parallax analysis have proper motion estimates biased toward more negative velocities in longitude due to poor sampling of the parallax ellipse in the late epochs . of the remaining 16 well - sampled maser channels used in the parallax analysis , only region
9 differs in the longitude component of its proper motion by more than @xmath45 km s@xmath10 ( at the longitude of iras 00420@xmath15530 , equatorial and galactic coordinates are approximately parallel , so @xmath67 ) .
the dispersion in the latitude component is somewhat larger , but it is irrelevant to the space velocity analysis within the galactic plane .
region 9 has a velocity in longitude @xmath425 km s@xmath10 more negative than region 3 . alternatively choosing region 9 as representative ameliorates the anomaly in the circular orbital component of the velocity found in the above analysis for region 3 quite well , but also _ increases _ the anomalous component in the direction of the galactic center .
while we consider the remarkable consistency with the w3(oh ) anomalous space velocity compelling , it is not , by itself , sufficient evidence of any interpretation of possible internal motions in iras 00420@xmath15530 . without an absolute reference within the source , internal motion biases
can not be ruled out , but an anomalous motion in the general direction of the sun and the galactic center can not be excluded by the available h@xmath0o maser proper motion data .
distance and three - dimensional space velocity measurements of a larger sample of such objects in this and other regions of the galaxy ( e.g. , * ? ? ?
* ) are necessary to resolve such issues conclusively , and establish the true bulk motion of material in the perseus spiral arm .
such observations could possibly also better constrain the galactic orbital speed and solar peculiar motion , but will be limited by the isotropy and homogeneity ( i.e. , the apparent lack of galactic axisymmetry ) of the available sample .
@xcite have measured the distance to ngc 281a star - forming region only @xmath68 from iras 00420@xmath15530using vera @xcite observations of h@xmath0o masers and a similar analysis ( using the same phase - reference calibrators ) .
they find a distance for ngc 281 of [email protected] kpc , 650 pc further than iras 00420@xmath15530 , and postulate that these sources lie on opposite sides of the surface of an hi super - bubble that is expanding out of the plane of the galaxy .
their model is geometrically appealing , and consistent with a substantial anomalous velocity component toward the sun for iras 00420@xmath15530 , on the near side of the bubble .
ngc 281 is on the opposite side of the bubble , with line - of - sight velocity consistent with their model .
however , their model also predicts a large velocity out of the plane for iras 00420@xmath15530 , which we do not observe .
unmodeled internal motions of the h@xmath0o masers in iras 00420@xmath15530 perpendicular to the galactic plane could explain this discrepancy , but the available information can not resolve this ambiguity .
w3(oh ) is not associated with the super - bubble in their model .
using the vlba , we have measured the parallax and three - dimensional space velocity of h@xmath0o masers in iras 00420@xmath15530 .
the parallax of [email protected] mas provides a purely geometric distance estimate of [email protected] kpc ( @xmath69% ) , a factor of two closer than popularly assumed kinematic distance estimates suggest , yet consistent with an older , far - less accurate photometric distance . at this distance
, the space velocity of iras 00420@xmath15530 shows a substantial anomalous component in the direction of the sun , consistent with similar observations of w3(oh ) , implying a significant non - circular systematic motion of the perseus spiral arm of the galaxy .
parallax observations of larger samples of maser sources using the vlba have the potential to extend the distance estimates of a few % accuracy to a large fraction of the galaxy , and will contribute to more accurate non - axisymmetric models of galactic rotation .
the national radio astronomy observatory is operated by associated universities inc .
, under cooperative agreement with the national science foundation .
the authors wish to thank m. reid for useful discussions .
brand , j. and blitz , l. 1993 , 275 , 67 brand , j. , et al .
1994 , suppl 103 , 541 brisken , w. f. , benson , j. m. , goss , w. m. , & thorsett , s. e. 2002 , 571 , 906 chatterjee s. , cordes , j. m. , vlemmings , w. h. t. , arzoumanian , z. , goss , w. m. , & lazio , t. j. w. 2004 , 604 , 339 dehnen , w. & binney , j. j. 1998 , , 298 , 387 greisen , e. w. 2003 , astrophysics and space science library , 285 , 109 hachisuka , k. , et al .
2006 , 645 , 337 honma , m. , kawaguchi , n. , & sasao , t. 2000 , 4015 , 624 kerr , f. j. & lynden - bell , d. 1986 , , 221 , 1023 ma , c. , et al . 1998 , , 116 , 516 molinari , s. , brand , j. , cesaroni , r. , & palla , f. 1996 , 308 , 573 molinari , s. , testi , l. , rodrguez , l. f. , & zhang , q. 2002 , 570 , 758 neckel , t. & staude , h. j. 1984 , 131 , 200 reid , m. j. 2008 , , 212 , 6303 reid , m. j. & brunthaler , a. 2004 , , 616 , 872 reid , m. j. , readhead , a. c. s. , vermeulen , r. c. , & treuhaft , r. n 1999 , 524 , 816 sato , m. , et al .
2008 , , in press shepherd , m. c. 1997 , astronomical data analysis software and systems vi , 125 , 77 van langevelde , h. j. , vlemmings , w. , diamond , p. j. , baudry , a. , & beasley , a. j. 2000 , 357 , 945 xu , y. , reid , m. j. , zheng , x. w. , & menten , k. m. 2006 , science 311 , 54 zhang , q. , hunter , t. r. , brand , j. , sridharan , t. k. , cesaroni , r. , molinari , s. , wang , j. , & kramer , m. 2005 , 625 , 864
|
we have used the vlba to measure the annual parallax of the h@xmath0o masers in the star - forming region iras 00420@xmath15530 .
this measurement yields a direct distance estimate of @xmath2 kpc ( @xmath33% ) , which disagrees substantially with the standard kinematic distance estimate of @xmath44.6 kpc ( according to the rotation curve of * ? ? ?
* ) , as well as most of the broad range of distances ( [email protected] kpc ) used in various astrophysical analyses in the literature . the 3-dimensional space velocity of iras 00420@xmath15530 at this new , more accurate distance implies a substantial non - circular and anomalously slow galactic orbit , consistent with similar observations of w3(oh ) @xcite , as well as line - of - sight velocity residuals in the rotation curve analysis of @xcite .
the perseus spiral arm of the galaxy is thus more than a factor of two closer than previously presumed , and exhibits motions substantially at odds with axisymmetric models of the rotating galaxy .
|
most of the dense molecular gas in galaxies is assembled into giant molecular clouds ( gmcs ) with masses from 10@xmath15 m@xmath6 to 10@xmath16 m@xmath6 and radii from 50 pc to a few hundred pc ( e.g. , fukui & kawamura 2010 ) .
as stars , the `` atoms '' of galaxies , are exclusively formed in these gmcs , physical processes that govern the conversion of hi into h@xmath0 clearly play an important role in determining the properties of gmcs and consequently the initial conditions for star formation .
therefore , understanding the properties of the interstellar regions where gas transitions from h@xmath17h@xmath0-dominated ( hi h@xmath0 transition hereafter ) is an important step toward a complete theory of star formation . from an observational point of view , the hi
transition has been investigated in the galaxy mainly via ultraviolet ( uv ) absorption measurements toward background sources .
for example , savage et al .
( 1977 ) and gillmon et al .
( 2006 ) measured h@xmath0 column densities , @xmath18(h@xmath0 ) , along the lines of sight to early - type stars and agns using the _ copernicus _ and _ fuse _ satellites and combined them with hi column densities , @xmath18(hi ) , to estimate h@xmath0 fraction , @xmath19 = 2@xmath18(h@xmath0)/(@xmath18(hi@xmath20(h@xmath0 ) ) .
they found that @xmath19 sharply rises at @xmath18(h ) = @xmath18(hi@xmath20(h@xmath0 ) @xmath3 ( 35 ) @xmath11 10@xmath12 cm@xmath7 .
several far - infrared ( fir ) studies of interstellar clouds have revealed the presence of h@xmath0 based on the comparison between dust and atomic gas contents ( e.g. , reach et al .
1994 ; meyerdierks & heithausen 1996 ; douglas & taylor 2007 ) .
reach et al .
( 1994 ) found a significant positive deviation from a linear relation between ir intensity at 100 @xmath21 m and @xmath18(hi ) for cirrus clouds , indicating a substantial amount of h@xmath0 .
the threshold @xmath18(hi ) at which @xmath18(h@xmath0 ) implied by the fir data becomes significant was found to be @xmath34 @xmath11 10@xmath12 cm@xmath7 , consistent with the results from savage et al .
( 1977 ) and gillmon et al .
( 2006 ) .
more recently , the hi h@xmath0 transition has been observationally inferred by an absence of high @xmath18(hi ) . for example , barriault et al . ( 2010 ) observed two high galactic latitude clouds in the oh and hi transitions and found that @xmath18(hi ) saturates at @xmath35 @xmath11 10@xmath12 cm@xmath7 for high values of @xmath18(oh ) , suggesting that there is molecular gas not traced by hi .
extragalactic studies have found this saturation as well .
wong & blitz ( 2002 ) , blitz & rosolowsky ( 2004 , 2006 ) , bigiel et al .
( 2008 ) , and wong et al .
( 2009 ) observed nearby galaxies in the co and hi transitions and found that hi surface density , @xmath1 , saturates at @xmath310 m@xmath6 pc@xmath7 .
most extragalactic studies have a spatial resolution of @xmath3700 pc at best , where most gmcs are unresolved and several are likely to be blended along a line of sight .
in addition , at this resolution it is impossible to compare h@xmath0 with the corresponding hi associated with the cloud envelopes . from a theoretical point of view , the hi h@xmath0 transition has been studied in the context of the structure of photodissociation regions ( pdrs ) . as uv photons filling up the bulk of the interstellar medium ( ism ) photodissociate h@xmath0 , predominantly molecular gas is only found in the dense regions where high enough shielding is achieved by both dust grains and h@xmath0 self - shielding .
these molecular regions are bounded by pdrs where the gas predominantly exists in the atomic phase ( e.g. , hollenbach & tielens 1997 ) .
the structure of pdrs has been mainly investigated for 1d geometries that include uni - directional or bi - directional beams of radiation propagating into semi - infinite slabs or purely radial radiation fields hitting the surfaces of spherical clouds ( e.g. , van dishoeck & black 1986 ; draine & bertoldi 1996 ; browning et al . 2003 ) . however , this 1d approximation is not appropriate for gmcs that are embedded in a radiation field produced by the combination of many single stars and star clusters .
in addition , when gmcs are surrounded by large hi envelopes , which have been frequently observed in the galaxy ( e.g. , wannier et al . 1983 , 1991 ) , the curvature of the pdrs and the finite size of the molecular regions are no longer negligible .
a higher dimensional approach is required in this case , but as it requires a treatment of the angular dependence of the radiation field , all previous studies have been purely numerical ( e.g. , spaans & neufeld 1997 ; liszt & lucas 2000 ; liszt 2002 ) .
these numerical studies do not provide a simple analytic estimate of the structure of pdrs that allows us to extrapolate the models over a wide range of ism environments and to easily compare with observations .
recently , krumholz et al . ( 2008 , 2009 ; kmt08 and kmt09 hereafter ) performed analytic studies on a pdr in a spherical cloud that is bathed in a uniform and isotropic interstellar radiation field ( isrf ) .
they investigated the steady - state equilibrium formation and photodissociation of h@xmath0 and approximated the hi
h@xmath0 transition as being infinitely sharp .
their calculation of where exactly inside a molecular cloud this transition happens resulted in two important analytic predictions .
first , they found that a certain amount of @xmath1 is required to provide enough shielding for h@xmath0 formation .
for gmcs with solar metallicity , this is @xmath1 @xmath3 10 m@xmath6 pc@xmath7 . once the minimum @xmath1 is achieved to shield h@xmath0 against photodissociation , all excess hi is converted into h@xmath0 and the h@xmath0 abundance increases .
second , they derived an analytic function for h@xmath0-to - hi ratio , @xmath22 = @xmath2/@xmath1 , and showed that @xmath22 depends on the total gas surface density and metallicity but is interestingly almost independent of the strength of the isrf . this is a very powerful prediction , suggesting that for a given ism environment , the total gas surface density determines the h@xmath0-to - hi ratio . if confirmed by observations , this `` easy '' prescription of h@xmath0 formation will be invaluable for numerical simulations of galaxy formation and evolution . from a purely practical perspective , as the above predictions are functions of direct observables ( total gas surface density and metallicity ) , they are much easier to test observationally .
most of the previous models of h@xmath0 formation ( e.g. , van dishoeck & black 1986 ; draine & bertoldi 1996 ; browning et al . 2003 ) require the intensity of the uv radiation field and the number density of h nuclei as input parameters , which can not be easily determined by observations .
kmt09 s predictions specifically focus on the properties of gmcs on galactic scales ( @xmath3kpc ) .
several recent studies investigated the hi and h@xmath0 properties of nearby spiral and dwarf galaxies on kpc scales and found reasonable agreement with kmt09 s prediction of @xmath22 as a function of total gas surface density and metallicity ( e.g. , kmt09 ; fumagalli et al .
2009 , 2010 ) . on the other hand , kmt09 s simplified treatment of the thermal and chemical processes in pdrs may be less suited for studies of individual pdrs .
for example , kmt09 s model does not consider the temperature dependence of h@xmath0 formation rate coefficients , h@xmath0 dissociation by cosmic rays , or absorption of lyman - werner ( lw ) photons by molecules other than h@xmath0 .
nevertheless , kmt09 s model provides strong physical insights into the fundamental principles of h@xmath0 formation . in this paper
, we test kmt09 s predictions on the perseus molecular cloud ( perseus hereafter ) on sub - pc scales ( @xmath30.4 pc ) for the first time in an attempt to understand the fundamental principles of h@xmath0 formation in gmcs .
perseus is a nearby molecular cloud located at a distance ranging from 200 pc to 350 pc ( herbig & jones 1983 ; ernis 1990 ) and has roughly a projected angular size of 6@xmath23 by 3@xmath23 on the sky ( based on co emissionco(10 ) data are quoted as co. ] ) .
we adopt the distance to perseus of 300 pc throughout this paper .
perseus is part of the large taurus
perseus molecular cloud complex and lies below the galactic plane ( @xmath24 ) , resulting in slightly less confused hi spectra compared to other galactic gmcs . with @xmath25 m@xmath6 ( sancisi et al .
1974 ; ungerechts & thaddeus 1987 ) , this is a low mass cloud with an intermediate level of star formation ( bally et al .
figure [ f : intro ] shows the @xmath4-band extinction image of perseus with co contours to emphasize the locations of several prominent dark and star - forming regions .
ic348 ( age = 0.712 myr ; herbig 1998 ) and ngc 1333 ( age < 1 myr ; lada et al .
1996 ) are active star - forming regions , while b5 , b1e , b1 , l1455 , l1448 , and l1451 are dark regions with low levels of star formation .
a wealth of multi - wavelength observations exist for perseus ; we provide details in section 2 .
one of the recent observational advances is the high resolution hi data set of perseus and its outskirts provided by the galfa hi survey ( peek et al .
this data set nicely matches the angular resolution of fir images , @xmath34@xmath26 , allowing us to estimate @xmath22 across and around perseus and to compare the results to kmt09 s predictions .
this paper is organized in the following way .
we start with a description of the data sets used in this study ( section 2 ) .
we then describe details of our methodology to derive @xmath18(hi ) and @xmath18(h@xmath0 ) images ( sections 3 and 4 ) . in section 5 ,
we examine the dust temperature image closely and discuss the implications for the strength of the radiation field across perseus . in section 6 , we summarize kmt09 s model and predictions . in section 7
, we compare our @xmath1 and @xmath22 images with kmt09 s predictions .
we then compare the derived h@xmath0 distribution with the existing co observations and investigate the existence of the `` co - dark '' h@xmath0 gas ( section 8) .
finally , we discuss and summarize our conclusions ( sections 9 and 10 ) .
[ [ hi ] ] hi ~~ we use hi data from the galactic arecibo l - band feed array hi survey ( galfa hi ) .
hi uses alfa , a seven - beam array of receivers mounted at the focal plane of the 305-m arecibo telescope , to map hi emission in the galaxy .
each of seven dual polarization beams has an effective beamsize of 3.9@xmath26 @xmath11 4.1@xmath26 and a gain of 8.511 jy k@xmath27 ( peek et al .
the galfa hi spectrometer , galspect , has a velocity resolution of 0.184 km s@xmath27 ( 872 hz ) and covers @xmath28 km s@xmath27 < @xmath29 < @xmath30 km s@xmath27 ( 7 mhz ) in the local standard of rest ( lsr ) frame . galfa
hi adopts two different scanning modes , the drift and basketweaving modes . in the drift mode ,
the telescope is fixed at azimuth of 0@xmath23 or 180@xmath23 and the sky is allowed to drift by . in the basketweaving mode ,
on the other hand , the telescope is fixed at the meridian and is driven up and down in zenith angle over the chosen declination ( dec ) range . for each day of observations ,
the starting point of the telescope is shifted in right ascension ( ra ) by @xmath312@xmath26 such that the entire proposed region is covered .
alfalfa ( giovanelli et al . 2005 ) and galfacts ( guram & taylor 2009 ) , two large area surveys with which galfa hi commensally observes , use the drift and basketweaving mode , respectively .
galfa hi combines hi data from many individual projects , both in solo and commensal modes .
reduction of hi data is performed with the galfa hi standard reduction ( gsr ) pipeline .
full details of gsr are presented in peek et al .
( 2011 ) , but here we briefly describe its most important steps .
first , hi spectra are corrected for the intermediate frequency ( if ) gain by dividing them by a reference spectrum .
this reference spectrum is obtained via the least - squares frequency switching ( lsfs ) technique ( heiles 2005 ) .
the lsfs observation is performed at the beginning of each observing period and the obtained if gain spectrum is used for the whole period s data .
second , the hi spectra are searched for baseline ripples produced by reflections in the signal chain , the primary reflector , and the gregorian dome .
this initial calibration is done by examining the hi data in the fourier space and removing the fourier components that correspond to the reflection patterns .
third , the gain calibration is applied to the hi spectra .
the relative gains of each beam over each day of observations are determined by finding the points where beam tracks intersect and comparing the hi spectra of each beam .
this gain calibration is most effective for the scans obtained in the basketweaving mode , where a large number of intersection points are produced .
lastly , the fully calibrated hi spectra are gridded onto the image plane in cartesian coordinates .
the gridded hi data are then calibrated for the first sidelobe and the final brightness temperature scale .
the lower angular resolution hi data from the leiden
argentine bonn ( lab ) survey ( kalberla et al . 2005 ) are used for the final brightness temperature calibration .
the galfa
hi first data release ( dr1 ) has been presented by peek et al .
the dr1 data cover 7520 deg@xmath31 of sky and have been produced using 3046 hours of data obtained from 12 individual projects .
rms noise in a 1 km s@xmath27 channel ranges from 140 mk to 60 mk , with a median of 80 mk .
the dr1 data can be downloaded at https://purcell.ssl.berkeley.edu/. the hi data used in this paper are part of dr1 and have been obtained from togs2 ( turn - on galfa spectrometer2 ; pi : c. heiles & s. stanimirovi ) , a2004 ( hi survey of the taurus molecular cloud ; pi : p. goldsmith ) , and a2174 ( hi survey of the perseus molecular cloud ; pi : l. knee )
. we combined the scans from these three projects and produced an hi cube centered at ( ra , dec ) = ( 03@xmath3229@xmath3352@xmath34,@xmath3530@xmath3634@xmath261@xmath37 ) in j2000 with a size of 14.8@xmath36 @xmath11 9@xmath36 .
we use 60 @xmath21 m and 100 @xmath21 m images from the improved reprocessing of the _ iras _ survey ( iris ) ( miville - desch@xmath38nes & lagache 2005 ) .
iris provides an improved zero point calibration and correction for the zodiacal light and striping in the images , as well as increased angular resolution compared to the original _ iras _ images .
the angular resolution at 60 @xmath21 m and 100 @xmath21 m is 4@xmath26 and 4.3@xmath26 , respectively , comparable to that of the galfa hi data .
the noise level in the iris images is approximately 0.07 mjy sr@xmath27 and 0.48 mjy sr@xmath27 at 60 @xmath21 m and 100 @xmath21 m , respectively .
[ [ co ] ] co ~~ we use the co integrated intensity image provided by dame et al . ( 2001 ) .
dame et al .
( 2001 ) combined co surveys of the galactic plane and local molecular clouds and produced a composite survey of the entire galaxy at an angular resolution of 8.4@xmath26 .
the co data were obtained by the 1.2-m telescope at the harvard - smithsonian center for astrophysics ( cfa ) .
the spectra were sampled with an angular spacing of 7.5@xmath26 and the final data cube has a uniform rms noise of 0.25 k per a 0.65 km s@xmath27 channel . to estimate co integrated intensity , @xmath39 , dame et al .
( 2001 ) integrated co emission from @xmath40 km s@xmath27 to @xmath41 km s@xmath27 .
see section 2 of dame et al .
( 2001 ) for details of the observations and analysis .
while we use the cfa co data for the most of our analysis , @xmath42co(10 ) data provided by the complete survey ( ridge et al .
2006a ) are used in section 7 .
as @xmath42co(10 ) is a better tracer of dense molecular gas than co , we use the @xmath43 image with a size of 6.25@xmath44 to set the boundaries of the dark and star - forming regions in perseus .
the @xmath42co(10 ) data were obtained by the 14-m five college radio astronomy observatory ( fcrao ) telescope at an angular resolution of 44@xmath37 . to study perseus , ridge et al .
( 2006a ) integrated @xmath42co(10 ) emission from @xmath45 km s@xmath27 to @xmath46 km s@xmath27 with a velocity resolution of 0.07 km s@xmath27 and estimated an average rms noise per channel of 0.17 k. we use the @xmath48 image provided by the complete survey .
this image was constructed from the two micron all sky survey ( 2mass ) point source catalog using the nicer algorithm ( lombardi & alves 2001 ) at an angular resolution of 5@xmath26 .
the nicer algorithm estimates reddening along a line of sight by comparing the observed colors of background stars to the intrinsic colors ( measured in a nearby control field , where the extinction is negligible ) .
note that this method does not assume a grain - size distribution or dust emissivity and therefore provides the most reliable estimate of the total gas column density in a molecular cloud ( e.g. , goodman et al .
2009 ; gps09 hereafter ) . in this paper , however , we do not directly use the 2mass @xmath48 image to derive the total gas column density because it only covers the central region of perseus , while we are mainly interested in the cloud outskirts .
instead , we use the 2mass @xmath48 image to calibrate our @xmath49 image derived from the iris data ( see section 4.4 ) . in this paper , we combine the data sets described above in several ways . in each case
, we first bring all input data sets to common angular resolution determined by the lowest angular resolution data set .
this is done by performing convolution with an appropriate gaussian kernel .
for example , to determine the velocity range of hi emission based on the correlation between @xmath18(hi ) and the dust column density , we smooth the hi data to the angular resolution of the 2mass @xmath48 image , 5@xmath26 ( section 3.2 ) . to derive @xmath18(h@xmath0 ) , the iris and hi data are convolved with a gaussian kernel to have an angular resolution of 4.3@xmath26 .
consequently , our final @xmath1 , @xmath2 , and @xmath22 images have an angular resolution of 4.3@xmath26 ( sections 4.5 and 7 ) .
the main component of the hi emission in perseus is centered at @xmath29 = @xmath50 km s@xmath27 .
compared to the molecular gas traced by co , the peak of the hi emission is offset by 23 km s@xmath27 . as an example , we compare the hi and co spectra at the position of ic348 in figure [ f : hi_co_spectra ] .
there are ( at least ) two components in the hi spectrum and the stronger one peaks at @xmath51 @xmath355 km s@xmath27 .
interestingly , the location of the dip between the two hi components corresponds to the co peak at @xmath51 @xmath358 km s@xmath27 .
this dip is likely from hi self - absorption ( hisa ) and indicates the presence of very cold neutral medium ( spin temperature @xmath52 k ; e.g. , gibson 2010 ) .
the correction for hisa requires a knowledge of the relative contributions from cold and warm gas to the hi emission , which can be obtained via hi absorption measurements .
however , only two hi absorption measurements exist for perseus ( e.g. , heiles & troland 2003a ) .
this is clearly too sparse to interpolate the relative contributions from cold and warm gas throughout the whole molecular cloud .
therefore , we do not apply the correction for hisa to @xmath18(hi ) but we investigate the optical depth effect in section 7.1.1 by applying the first order opacity correction .
to determine the velocity range of hi emission associated with perseus , we investigate the correlation between @xmath18(hi ) and the dust column density .
the correlation between hi emission and fir emission has been well - established for high galactic latitudes ( e.g. , boulanger et al .
we use the 2mass @xmath48 image as a tracer of the dust column density and smooth the hi data to the angular resolution of the 2mass @xmath48 image .
we fix the central velocity of hi emission at @xmath535 km s@xmath27 based on section 3.1 and integrate hi emission using a velocity width @xmath54 increasing from 2 km s@xmath27 to 80 km s@xmath27 with 2 km s@xmath27 intervals . for each derived @xmath18(hi ) image , we calculate the correlation coefficient between @xmath18(hi ) and 2mass @xmath48 . only data points with signal - to - noise ratio ( snr ) @xmath55 in the 2mass @xmath48 image are used for this calculation . figure [ f : hi_av_corr ] shows the correlation coefficient as a function of @xmath54 . although the correlation coefficient is never very high , there is a clear peak at @xmath56 km s@xmath27 .
the lack of strong correlation between @xmath48 and @xmath18(hi ) has been noticed before and is mostly ascribed to the presence of molecular gas ( e.g. , reach et al .
1994 ; dame et al . 2001 ; leroy et al .
we take the correlation coefficient of 0.15 ( half of the maximum ) as a threshold and use the corresponding @xmath54 of 20 km s@xmath27 to derive @xmath18(hi ) under the assumption of optically thin hi gas(hi ) is calculated by @xmath18(hi ) = 1.823 @xmath11 @xmath57 ( cm@xmath7 ) , where @xmath58 is the brightness temperature . ] .
the derived @xmath18(hi ) ranges from 3.5 @xmath11 10@xmath12 cm@xmath7 to 1.3 @xmath11 10@xmath59 cm@xmath7 , with a median of 7.9 @xmath11 10@xmath12 cm@xmath7 .
we present the @xmath18(hi ) image in figure [ f : hi_col ] .
sancisi et al .
( 1974 ) and imara & blitz ( 2011 ) estimated @xmath18(hi ) for perseus independently and their results are essentially consistent with ours .
sancisi et al .
( 1974 ) obtained hi data for perseus using the 25-m dwingeloo telescope and estimated @xmath18(hi ) by integrating hi emission from @xmath60 km s@xmath27 to @xmath46 km s@xmath27 .
in addition , they attempted to subtract the foreground and background @xmath18(hi ) by comparing the hi emission in the direction of perseus to the hi emission measured in a nearby field spatially offset by a few degrees .
they found an average @xmath18(hi ) associated with perseus @xmath3 5 @xmath11 10@xmath12 cm@xmath7 .
this is close to our median @xmath18(hi ) .
on the other hand , imara & blitz ( 2011 ) used hi data from the lab survey and estimated @xmath18(hi ) by integrating hi emission from @xmath61 km s@xmath27 to @xmath62 km s@xmath27 . to determine the velocity range for @xmath18(hi ) , they assumed that hi spectra can be approximated with gaussian functions and set @xmath54 to be equal to four times the standard deviation of the gaussian component associated with perseus . in summary ,
both sancisi et al .
( 1974 ) and imara & blitz ( 2011 ) derived @xmath18(hi ) under the assumption of optically thin hi gas and their velocity ranges for @xmath18(hi ) are consistent with ours .
before further processing , we remove point sources from the 60 @xmath21 m and 100 @xmath21 m images . we identify compact point sources from the _ iras _ point source catalog that have @xmath64 mjy sr@xmath27 and remove them by applying a circular mask with a radius equal to twice the angular resolution of the 100 @xmath21 m image .
these masks are shown as blank pixels in figures [ f : ir100 ] , [ f : dust_map ] , [ f : rh2 ] , and [ f : h2_co ] .
as perseus belongs to the taurus
perseus molecular cloud complex and our data cover a large area on the sky ( @xmath315@xmath65 ) , there is a concern of including molecular gas not associated with perseus in our analysis . to investigate this , we plot @xmath66 and @xmath67 as a function of @xmath18(hi ) in figure [ f : hi_ir ] . as @xmath66/@xmath18(hi ) and @xmath67/@xmath18(hi ) are approximations of dust - to - gas ratio ( dgr ) , an overall linear relation , implying a single dgr , is a reasonable expectation for a single molecular cloud .
figure [ f : hi_ir ] shows that the data points are segregated into two possible distinct features at @xmath18(hi ) @xmath3 6 @xmath11 10@xmath12 cm@xmath7 .
the majority of the data points shows a global linear relation between @xmath18(hi ) and fir emission .
these data points are shown in gray and we assume that they mostly trace perseus . however , the data points in black have a slightly different slope .
we selected the black data points as having @xmath18(hi ) < 6 @xmath11 10@xmath12 cm@xmath7 . in figure
[ f : ir100 ] , we show the spatial distribution of the black data points by overlaying them on the 100 @xmath21 m image . clearly , they are all found at the southeast edge of the image in the direction of taurus .
taurus is centered at ( ra , dec ) = ( 04@xmath3230@xmath33,@xmath68 ) and has @xmath69 = @xmath70 km s@xmath27 ( e.g. , narayanan et al . 2008 ) . as a distinct molecular cloud
, taurus would have a different dgr . to avoid any possible contamination , we exclude ( blank ) the black data points from the iris images and further analysis .
in addition to the above consideration , the warm dust ring that stands out in the dust temperature image ( see section 4.3 ) is an additional source of contamination and we exclude it from further analysis . for optically thin dust grains at an equilibrium temperature @xmath73
, the optical depth at 100 @xmath21 m , @xmath49 , is determined by : @xmath74 where @xmath67 is the measured intensity at 100 @xmath21 m and @xmath75 is the intensity of a blackbody of temperature @xmath73 at 100 @xmath21 m . to estimate @xmath49
, we first estimate @xmath73 using the ratio @xmath76 . under the assumption that the mass absorption coefficient , @xmath77 , has a wavelength dependence such as @xmath78 ,
@xmath73 can be estimated from : @xmath79 where @xmath80 is the planck constant and @xmath81 is the boltzmann constant .
the emissivity spectral index , @xmath82 , can vary between 1 ( for amorphous carbon ) and 2 ( for metallic and crystalline dielectric material ) , depending on the grain composition and size . throughout this paper , we adopt @xmath83 based on the discussions in schnee et al .
( 2005 ; s05 hereafter ) . to estimate @xmath73
, we smooth the 60 @xmath21 m image to the angular resolution of the 100 @xmath21 m image .
@xmath66 may include a significant contribution from stochastically heated very small grains ( vsgs ) that must be accounted for .
to do this , we adopt the method used by s05 .
first , we smooth the 60 @xmath21 m and 100 @xmath21 m images to the angular resolution of the @xmath73 map derived by schlegel et al .
( 1998 ; sfd98 hereafter ) .
sfd98 used the low resolution dirbe 100 @xmath21 m and 240 @xmath21 m data to derive @xmath73 ; the contribution from vsgs is expected to be negligible at these wavelengths ( e.g. , li & draine 2001 ) .
second , we calculate @xmath73 based on 100% of @xmath67 and a lesser percentage of @xmath66 , assuming @xmath83 . the fraction of @xmath66 is chosen to minimize the average difference between our @xmath73 and sfd98 s @xmath73 .
we find that the best matching fraction is 22% , meaning that 78% of @xmath66 is due to vsgs .
this result is consistent with s05 , who found 26% , but lower than the prediction of interstellar dust model for the solar neighborhood ( @xmath340% ; dsert et al .
1990 ) .
we present the @xmath73 image in figure [ f : dust_map ] .
our @xmath73 image is generally consistent with the @xmath73 image derived by s05 , but shows systematically slightly lower @xmath73 .
the difference between our @xmath73 and s05 s @xmath73 , @xmath84 , has a median of @xmath85 k. this discrepancy likely arises from the fact that s05 adopted a @xmath73-dependent formula for @xmath82 and used a different fitting algorithm .
their @xmath73-dependent formula for @xmath82 considered the observed inverse relation between @xmath73 and @xmath82 and had two free parameters ( see equation 2 of s05 ) .
to simultaneously fit these two free parameters , s05 used the idl routine amoeba , which adopts the downhill method of nelder & meade ( 1965 ) , while we estimated the best matching parameter using the least - squares method .
@xmath73 across perseus ranges from @xmath316 k to @xmath322 k. the lowest @xmath73 regions spatially correlate with the highest @xmath48 , in agreement with the expectation for dense and cold molecular gas .
several localized regions with high @xmath73 stand out ; two are the star - forming regions , ic348 and ngc1333 , while the third is a ring - like feature located at ( ra , dec ) = ( 03@xmath3239@xmath3330@xmath34,@xmath3532@xmath36 ) with a radius of @xmath356@xmath26 .
ridge et al .
( 2006b ) investigated this ring in detail using _ iras _ , _ msx _ , 2mass @xmath48 , and co data and concluded that the ring , the product of an hii region driven by the b - type star hd 278942 , is likely located behind ic348 .
however , they also argued that the ring is interacting with perseus , based on the spatial correlation between 8 @xmath21 m emission and 2mass @xmath48 .
the ring is visible in the 2mass @xmath48 image but at a much lower contrast , implying that the ring is not as significant column density feature as one might infer from the iris data .
therefore , we exclude it from further analysis by applying a circular mask of radius of 56@xmath26 .
this mask size is consistent with what ridge et al .
( 2006b ) used to separate the ring from the perseus cloud component , @xmath350@xmath26 . at this point , we refined the zero point of the @xmath49 image by assuming that the dust column density traced by @xmath49 is proportional to @xmath18(hi ) for the regions dominated by diffuse atomic gas . under this assumption , @xmath49 would approach zero as @xmath18(hi ) approaches zero and any offset from this relation is likely due to the uncertainties in the zero points of the iris images . as a result of this exercise
, we added 1.8 @xmath11 10@xmath86 to the @xmath49 image .
the observed offset in @xmath49 implies a small additional uncertainty in @xmath73 , which is of order of 0.030.6 k. once we estimate @xmath49 using the vsg
corrected @xmath73 , we convert @xmath49 to @xmath48 by : @xmath88 where @xmath89 is the conversion factor that relates the dust column density to the extinction in the @xmath4-band . the conversion factor @xmath89 = 720
is estimated by finding the best agreement between our derived @xmath48 image and the 2mass @xmath48 image for the region where the two images overlap .
our @xmath48 and 2mass @xmath48 have a 1:1 relation but with an average scatter of @xmath30.5 mag .
this scatter around the 1:1 relation is mostly attributed to the variation in @xmath73 along a line of sight ( e.g. , gps09 ) .
we note that the bootstrapping of our @xmath48 image derived from the iris data with the @xmath48 image derived from the 2mass extinction data is motivated by gps09 . for perseus , gps09 inter - compared three methods for measuring total gas column density ; extinction mapping at near - infrared ( nir ) wavelengths , thermal emission mapping at fir wavelengths , and co emission mapping .
they found that the dust - based measures ( nir extinction and fir emission ) are superior to the gas - based measure ( co emission ) and particularly suggested the nir extinction as the best probe of total gas column density .
any molecular transition , such as co emission , traces a limited range of volume densities due to the critical density for excitation , opacity , and depletion .
the dust - based measures have a wider dynamic range , but the nir extinction measure saturates at high extinctions where background sources can not be observed and the fir emission measure is limited at low extinction values .
in addition , the fir emission measure suffers from its intrisic uncertainty , the variation of @xmath73 along a line of sight .
overall , the nir extinction measure is the best probe up to its saturation level .
therefore , our calibration of @xmath49 to the 2mass @xmath48 image would provide a better estimate of the total gas column density , compared to the total gas column density derived solely from the iris data .
finally , we calculate @xmath18(h@xmath0 ) by : @xmath91 where dgr is defined by dgr = @xmath92 and @xmath93 . to measure dgr in perseus , we plot @xmath48 as a function of @xmath94 over the whole cloud in figure [ f : hi_av ] .
as we expect @xmath95 @xmath3 @xmath96 for the regions dominated by diffuse atomic gas ( which correspond to the majority of data points in figure [ f : hi_av ] ) , the fitted slope @xmath97 mag cm@xmath31 is a good measure of dgr .
our estimate is @xmath32 times higher than the typical galactic dgr , 5.3 @xmath11 10@xmath99 mag cm@xmath31 ( e.g. , bohlin et al .
this is not concerning as notable variations have been found for the galactic dgr .
for example , kim & martin ( 1996 ) compiled @xmath48/@xmath18(h ) values for the galaxy derived from uv absorption measurements from literature and found that @xmath48/@xmath18(h ) can be higher than the typical galactic value by a factor of three ( see their figure 4 ) .
we also compare our dgr estimate with values obtained from two independent methods : + ( 1 ) _ copernicus / fuse absorption measurements _ : toward hd 22951 , hd 23180 , and bd@xmath100 , _ copernicus / fuse _ measured h@xmath0 directly in absorption ( savage et al .
1977 ; rachford et al . 2002 ) .
hd 22951 and hd 23180 are b1 stars and bd@xmath100 is a b5 star .
towards bd@xmath100 , snow et al .
( 1994 ) measured @xmath101 mag and combined this with the observed @xmath102 and @xmath103 , resulting in @xmath104 mag cm@xmath31 .
however , extinction parameters are not known for hd 22951 and hd 23180 . considering that hd 22951 , hd 23180 , and bd@xmath100
are close to each other on the sky , if we apply @xmath105 measured for bd@xmath100 to both hd 22951 and hd 23180 , we obtain @xmath106 = ( 4.66 ) @xmath107 mag cm@xmath31 .
while these estimates are close to the typical galactic dgr , we note that they are likely lower limits on the dgr of perseus , considering that most of the lines of sight surveyed by the _ copernicus / fuse _ are composed of two or more diffuse clouds ( e.g. , rachford et al .
2002 , 2009 ) .
rachford et al . ( 2002 ) compared the observed high rotational transition column densities of h@xmath0 to the model of h@xmath0 formation , destruction , and ro - vibrational excitation developed by browning et al .
( 2003 ) and found that it is impossible to reproduce the observed column densities under a single cloud assumption .
instead , models with multiple clouds along the line of sight , and/or multiple pathways of incident uv radiation , match much better the observed column densities .
stars preferentially form in regions with high dgr , where dust grains provide both high shielding against photodissociation and sites for h@xmath0 formation . therefore
, if several diffuse clouds exist along the line of sight toward perseus , the component associated with perseus likely has a higher dgr . in this case , the dgr measured along the whole line of sight is lower than the actual dgr of perseus .
+ ( 2 ) _ upper limit on the dgr _ : equation ( 4 ) itself provides an upper limit on the dgr . if the dgr is too high , a large number of data points will have a negative value of @xmath18(h@xmath0 ) .
for example , if dgr > 1.5 @xmath11 10@xmath108 mag cm@xmath31 , more than 50% of the pixels with @xmath48 > 1 mag will have a negative value of @xmath18(h@xmath0 ) .
therefore , it is most likely that the dgr in perseus is less than 1.5 @xmath11 10@xmath108 mag cm@xmath31 .
+ _ conclusion _ : based on two independent constraints on the dgr , we conclude that our estimate of @xmath97 mag cm@xmath31 is midway between the lower and upper limits and therefore represents a very reasonable estimate .
we estimate the uncertainty in @xmath103 for each pixel by performing a series of monte carlo simulations . in these simulations
, we take into account the uncertainties in @xmath102 and @xmath48 and how they propagate into @xmath103 . for the uncertainty in @xmath102 ,
we generate 1000 @xmath18(hi ) images by using 1000 velocity widths randomly drawn from a gaussian distribution that peaks at 20 km s@xmath27 with 1@xmath109 of 4 km s@xmath27 ( based on section 3.2 ) .
the median 1@xmath109 of @xmath102 due to the systematic uncertainty in using a particular velocity width is 5.6 @xmath11 10@xmath110 cm@xmath7 . for the uncertainty in @xmath48
, we assess the range of @xmath48 by varying the input parameters and deriving @xmath48 each time .
we add / subtract 1@xmath109 noise to / from the 60 @xmath21 m and 100 @xmath21 m images ( 0.07 mjy sr@xmath27 at 60 @xmath21 m and 0.48 mjy sr@xmath27 at 100 @xmath21 m ) and use @xmath111 or 2 .
therefore , we derive the @xmath48 images with 8 different combinations of the input parameters and for each case the contribution from vsgs to @xmath66 and @xmath89 for @xmath112 are determined by finding the best matching parameters . to include the uncertainties of the zero point
, we also derive the @xmath49 images with and without the zero point calibration .
we find that the contribution from vsgs to @xmath66 varies from 0.12 to 0.24 with a median of 0.17 , while @xmath89 varies from 630 to 810 with a median of 728 .
finally , we compare the 16 @xmath48 images and find the maximum and minimum estimates of @xmath48 for each pixel .
the uncertainty in @xmath102 and the maximum / minimum estimates of @xmath48 are incorporated in a monte carlo simulation to produce 1000 @xmath103 images .
we use the distribution of the simulated @xmath103 to estimate the uncertainty in @xmath103 on a pixel - by - pixel basis .
the median 1@xmath109 of @xmath18(h@xmath0 ) is 3.6 @xmath11 10@xmath110 cm@xmath7 .
in addition to the above considerations , several systematic effects are also likely to affect @xmath103 ; ( 1 ) the assumption of a single dust population and temperature along a line of sight , ( 2 ) the assumption of a single dgr , and ( 3 ) the assumption that h@xmath0 is the dominant source of excess dust emission . as it is difficult to estimate the magnitude of these effects , we do not include them in our monte carlo simulations .
@xmath73 across perseus ranges from @xmath316 k to @xmath322 k with a median of 17 k. this is in agreement with sfd98 who found @xmath113 17 k for the same region and s05 who found a slightly higher @xmath113 18 k for the main body of perseus
. similarly , stepnik et al . ( 2003 ) derived @xmath73 @xmath3 17 k for the outskirts of taurus using the dirbe data at 140 @xmath21 m and 240 @xmath21 m .
as shown in figure [ f : dust_map ] , the lowest @xmath73 regions spatially correlate with the a@xmath114 and co images , in agreement with the expectation for dense and cold clouds . the prominent minima in @xmath73 correspond to b5 , b1e , b1 , l1448 , l1451 , and l1455 .
these dark regions contain a number of ir sources and herbig - haro objects that are considered to be associated with pre - main sequence stars ( e.g. , bally et al .
2008 ) , but the luminosities of these objects are so low that their influence on the overall energy budget of the clouds would be negligible
. therefore , @xmath73 within these dark regions is determined by the external isrf and dust grain properties .
low @xmath73 in b5 , b1e , b1 , l1448 , l1451 , and l1455 compared to the surrounding regions suggests that the inside of these regions gas is dense enough to have sufficient shielding against the external isrf , leading to efficient cooling .
for example , b5 has limb - brightening in co , indicating a cold cloud interior surrounded by a photo - heated exterior ( e.g. , beichman et al .
1988 ) . figure [ f : dust_map ] also shows several localized regions with high @xmath73 .
one of the prominent features is the warm dust ring at ( ra , dec ) = ( 03@xmath3239@xmath3330@xmath34,@xmath115 ) , already discussed in section 4.3 .
the other high @xmath73 features include ic348 and ngc1333 . to quantify how much hotter these star - forming regions are , we plot the histograms of @xmath73 in figure [ f : dust_hist ] . for the histograms ,
we use the data points in the rectangular boxes shown in figure [ f : rh2 ] .
each box has either a dark or a star - forming region and extends up to the edge of the image .
the width of each box is set based on the complete @xmath42co(10 ) data to include the whole region of interest .
while all regions have a similar median @xmath73 of 17 k , the star - forming regions have data points with higher @xmath73 .
for example , for ic348 and ngc1333 , 11% and 2% of data points have @xmath73 > 18 k , while for b5 , b1e , and b1 , all data points have @xmath73 < 18 k. the data points with @xmath73 > 18 k are found in the central parts of ic348 and ngc1333 , where b - type stars are located ( blue crosses and red stars in figure [ f : dust_map ] ) , and this implies that the internal radiation field may play a significant role in heating dust grains inside ic348 and ngc1333 .
this possibility is discussed further in the next section .
we now investigate possible heating sources that can result in the measured @xmath73 . as we mentioned in section 1
, kmt09 s model assumes a uniform external isrf .
understanding the origin of isrf is therefore important for comparison between observations and the model .
there are two possible sources of isrf that are responsible for the measured @xmath73 across perseus ; ( 1 ) the uv flux from the b - type stars in perseus ( local ) and ( 2 ) the uv radiation field determined by the overall stellar distribution in the galaxy ( global ) .
\(1 ) we consider two b5 v - type stars in perseus ( red stars in figure [ f : dust_map ] ) as possible sources of the local isrf .
all other stars have spectral types later than b5 ) contribute negligibly to the uv flux .
we do not consider the b4 iv
v - type star ( red triangle in figure [ f : dust_map ] ) due to its uncertain stellar classification ( e.g. , lesh 1969 ; klochkova & kopylov 1985 ; ernis 1990 ) .
nevertheless , inclusion of this star does not change our result . ] .
we assume that the spectral energy distribution of a b - type star is approximately a black body curve ( e.g. , spaans et al .
1994 ) and adopt @xmath116 = 15600 k for a b5 v - type star ( morton & adams 1968 ) . for @xmath117
= 9122066 , two b5 v - type stars emit 2.12 @xmath11 10@xmath118 erg cm@xmath7 s@xmath27 .
if a distance of 15 pc from each star , which corresponds to @xmath31/3 of the longitudinal size of our sky coverage , is assumed , the uv flux from the two b5 v - type stars is reduced to 3.04 @xmath11 10@xmath119 erg cm@xmath7 s@xmath27 ( 1/@xmath120 geometrical dilution ) . for dust grains in thermal equilibrium , @xmath73 is determined by ( lequeux 2005 ) : @xmath121 where @xmath122 is the uv flux and @xmath123 is the size of dust grains .
an absorption efficiency @xmath124 = 1 , appropriate for large dust grains that dominate the fir emission and have a size comparable to uv wavelengths , and a dust emissivity spectral index @xmath82 = 2 are assumed for equation ( 5 ) . therefore ,
dust grains with a radius @xmath123 = 0.1 @xmath21 m in the uv radiation field of 3.04 @xmath11 10@xmath119 erg cm@xmath7 s@xmath27 would have @xmath113 9.3 k. note that we do not consider the decrement of uv flux by dust attenuation . therefore ,
if the b - type stars are the only sources of the isrf , we would expect @xmath125 k in the outskirts of perseus .
\(2 ) we consider the global galactic isrf of porter & strong ( 2005 ) .
porter & strong ( 2005 ) adopted the galactic stellar and dust distribution models and calculated the radiation field from stars at @xmath117 = 0.11000 @xmath21 m .
the isrf was found to be the most intense in the inner galaxy , to decrease toward the galactic poles , but to remain steady up to @xmath35 kpc from the galactic plane . for @xmath126 ( galactocentric radius )
= 0 and @xmath127 ( galactocentric height ) = 0 , the radiation field at 0.1 @xmath21 m is @xmath30.08 ev @xmath128 and for @xmath126 = 0 and @xmath127 = 5 kpc , the radiation field at 0.1 @xmath21 m is @xmath30.01 ev @xmath128 ( see figure 2 of porter & strong 2005 ) .
if we assume that this @xmath387.5% decrement of the uv radiation field at @xmath126 = 0 from @xmath127 = 0 to 5 kpc would be applicable for @xmath129 8 kpc , the uv radiation field at @xmath130 pc ( the location of perseus ; @xmath129 8 kpc and @xmath130 pc ) would be between 2.7 @xmath11 10@xmath131 ( the uv radiation field in the solar neighborhood at @xmath117 = 9122066 ; @xmath132 ) and 3.4 @xmath11 10@xmath86 erg cm@xmath7 s@xmath27 ( 12.5% of @xmath132 ) .
applying equation ( 5 ) , this corresponds to 14 k < @xmath73 < 20 k. as our median @xmath113 17 k is indeed in this range , the uv isrf determined by the overall stellar distribution in the galaxy could be the dominant heating source for dust grains across perseus .
therefore , we conclude that perseus is embedded in the uniform galactic isrf that heats dust grains up to @xmath113 17 k , except for the central parts of ic348 and ngc1333 where the radiation from the b - type stars is dominant .
kmt08 modeled h@xmath0 formation and photodissociation for a spherical cloud that is bathed in a uniform isrf . they assumed that the hi
h@xmath0 transition is infinitely sharp and found that the location of the transition inside an atomic - molecular complex is solely determined by two dimensionless numbers : @xmath134 @xmath135 here @xmath136 @xmath3 0.1 is the fraction of absorptions of lw photons that produce h@xmath0 dissociation rather than decay back to a bound state , @xmath137 is the dust absorption cross section per h nucleus in the lw band , @xmath138 is the number density of lw photons , @xmath139 is the number density of cnm in the hi shielding layer , @xmath140 is the h@xmath0 formation rate coefficient on dust grains , and @xmath141 is the cloud radius .
@xmath142 is the ratio of the rate at which lw photons are absorbed by dust grains to the rate at which they are absorbed by h@xmath0 .
we can consider @xmath142 as a dimensionless measure of the strength of the isrf .
for example , in strong radiation fields , h@xmath0 molecules are easily photodissociated and therefore the absorption of lw photons is dominated by dust grains ( large @xmath142 ) . on the contrary , in weak radiation fields ,
h@xmath0 molecules survive and due to their large resonant cross section the absorption of lw photons is dominated by h@xmath0 molecules ( small @xmath142 ) .
@xmath143 is simply the dust optical depth that the cloud would have if its density is equal to @xmath139 . using @xmath142 and @xmath143
, kmt08 derived @xmath144 , the fraction of the cloud radius where the hi h@xmath0 transition occurs : @xmath145^{1/3 } \\ & \approx 1 - \frac{\psi}{4\tau_{\rm r } - 0.7 \psi }
\\ \end{split}\ ] ] where @xmath146 then , the dust optical depth through the hi shielding layer , @xmath147 , can be determined by : @xmath148 the hi shielding surface density for h@xmath0 formation , @xmath149 , is simply : @xmath150 where @xmath151 is the mass per h nucleus .
to investigate what fraction of the gas in an atomic - molecular complex is molecular , kmt09 simplified the above equations as follows .
first , as both @xmath137 and @xmath140 are the measures of dust surface area , their ratio is nearly unity and drops out of equation ( 6 ) .
second , @xmath139 is determined from the condition of pressure balance between the cold neutral medium ( cnm ) and warm neutral medium ( wnm ) phases : @xmath152 here @xmath153 is the incident uv radiation field and @xmath154 is the metallicity .
the primes indicate that quantities are normalized to the values in the solar neighborhood .
there is a maximum temperature at which cnm can be in pressure balance with wnm and the corresponding density is the minimum cnm density , @xmath155 ( wolfire et al .
2003 ) . typically , @xmath156
@xmath3 110 . as @xmath139 @xmath157 @xmath153 @xmath157 @xmath138 , @xmath153 and @xmath138
drop out of equation ( 6 ) and @xmath142 becomes purely a function of @xmath154 and @xmath156 : @xmath158 to estimate @xmath143 , kmt09 started from the mean column density , @xmath159 , of a spherical cloud that consists of a molecular core of number density @xmath160 and a hi shielding layer of number density @xmath139 : @xmath161.\ ] ] here @xmath162 is the ratio of @xmath160 to @xmath139 .
once @xmath163 , the dust optical depth that the cloud would have if its hi and h@xmath0 are uniformly mixed , is defined by : @xmath164 @xmath143 can be expressed as a function of @xmath163 , @xmath162 , and @xmath144 : @xmath165 } \\ &
= \tau_{\rm c } \frac{1}{\left[1 + ( \phi_{\rm mol } - 1 ) x_{\rm h2}^{3}\right]}. \end{split}\ ] ] as a consequence of these simplified functions , the h@xmath0 mass fraction , @xmath166 , can be expressed with the following analytic function : @xmath167^{-1 } ,
\\ \end{split}\ ] ] where @xmath168 is the h@xmath0 mass and @xmath169 is the total cloud mass .
finally , the h@xmath0-to - hi ratio , @xmath22 , can be expressed as : @xmath170 - 1 .
\\ % & = \sigma_{\rm h2 } / \sigma_{\rm h\textsc{i}}. \\ \end{split}\ ] ] with @xmath171 being a function of @xmath154 and @xmath156 , and @xmath163 being a function of the cloud gas column density , @xmath22 is determined by the total gas column density , @xmath154 , @xmath156 , and @xmath162 , and is independent of the strength of the isrf . while the total gas column density and @xmath154 are direct observables , @xmath156 and @xmath162 are not .
kmt09 adopted @xmath156 = 3 and @xmath162 = 10 as fiducial values .
note that the h@xmath0-to - hi ratio can observationally be expressed as @xmath172 .
another important model prediction is that @xmath149 has a dependence on @xmath154 and is almost independent of the strength of the isrf .
based on equation ( 9 ) and ( 10 ) , @xmath147 slightly increases with @xmath154 via @xmath143 and @xmath171 and @xmath137 increases with @xmath154 via @xmath137 = @xmath173 .
therefore , @xmath149 decreases with @xmath154 to a power less than unity as @xmath174 . for solar metallicity , kmt09 predict @xmath149 @xmath3 10 m@xmath6 pc@xmath7 .
from our derived @xmath18(hi ) and @xmath18(h@xmath0 ) images , we calculate @xmath1 and @xmath2 by : @xmath175 @xmath176 we find that @xmath1 is relatively constant across perseus with a range of 511 m@xmath6 pc@xmath7 , while @xmath2 has a wider range of 073 m@xmath6 pc@xmath7 .
this results in a small dynamic range of @xmath9 , from 8 m@xmath6 pc@xmath7 to 30 m@xmath6 pc@xmath7 across most of the perseus cloud .
we then derive @xmath22 = @xmath2/@xmath1 , which allows us to directly test kmt09 s predictions .
the @xmath22 image is presented in figure [ f : rh2 ] .
we find that the derived @xmath22 ranges from 0 to 10 , with a median of 0.1 .
this suggests that diffuse molecular gas is pervasive in perseus .
for example , the data points with @xmath22 > 1 , which can be considered to be molecular - dominated , comprise only 9% of the total data points .
we note that spatial variations in the @xmath22 distribution are mainly determined by the @xmath2 distribution , due to the small dynamic range of @xmath1 .
the high @xmath22 features spatially correlate with the @xmath48 and co images and correspond to the dark and star - forming regions in perseus . to investigate kmt09 s predictions , we focus on several dark and star - forming regions in perseus and measure @xmath22 radially to probe the cloud envelopes with the lowest @xmath22 . in figure
[ f : sigmahi_sigmatotal ] , we plot @xmath1 as a function of @xmath9 for b5 , ic348 , b1e , b1 , and ngc1333 . for each plot
we use the data points in the rectangular boxes shown in figure [ f : rh2 ] .
the overlaid curves are kmt09 s predictions for particular combinations of input parameters , @xmath154 , @xmath156 , and @xmath162 .
these parameters will be determined and discussed in section 7.2.1 .
we find an almost constant @xmath1 @xmath3 68 m@xmath6 pc@xmath7 for each region .
this is consistent with the small range of @xmath1 found across the whole perseus cloud .
however , the median value of @xmath1 slightly changes between the regions .
for example , b5 has the lowest median @xmath179 m@xmath6 pc@xmath7 and ngc1333 has the highest median @xmath180 m@xmath6 pc@xmath7 .
a similar result for perseus was pointed out by sancisi et al .
( 1974 ) who found that @xmath1 remains approximately constant at @xmath38 m@xmath6 pc@xmath7 for a wide range of interstellar extinction .
several extragalactic studies have also noticed that @xmath1 does not exceed @xmath310 m@xmath6 pc@xmath7 for nearby galaxies ( e.g. , wong & blitz 2002 ; blitz & rosolowsky 2004 , 2006 ; bigiel et al .
2008 ; wong et al .
2009 ) . in kmt09 s model
, h@xmath0 formation requires a certain amount of @xmath149 to shield h@xmath0 against photodissociation .
once this minimum @xmath1 is achieved , h@xmath0 forms out of hi and @xmath1 remains constant . kmt09
s model predicts @xmath149 @xmath3 10 m@xmath6 pc@xmath7 for solar metallicity and this is indeed consistent with what we find in figure [ f : sigmahi_sigmatotal ] .
for all five regions , we find saturation of @xmath5 68 m@xmath6 pc@xmath7 . however , in the case of b5 and ngc1333 , we do not probe the regions of purely atomic gas , which correspond to the linear portion of the model curves in figure [ f : sigmahi_sigmatotal ] . for ic348 , b1e , and b1
, we see hints of a possible turnover between atomic- and molecular - dominated zones at @xmath9 = 58 m@xmath6 pc@xmath7 , yet our observations do not sample well the purely atomic zones .
this inability to probe purely atomic regions is not due to the sensitivity of our @xmath1 and @xmath2 images . in figure
[ f : sigmahi_sigmatotal ] , we show the median 3@xmath109 values of @xmath1 and @xmath9 for the whole perseus cloud as the black dashed lines .
all data points are clearly above the 3@xmath109 values and this suggests that the lack of turnover detection is due to the highly extended atomic envelopes .
for example , in the case of b5 and ngc1333 , we are probing radial profiles up to @xmath31520 pc from the centers of the regions .
therefore , the purely atomic envelopes are likely located more than @xmath320 pc from the molecular peaks . an alternative or additional explanation for the relatively constant @xmath1 we measured could be a high optical depth of hi in perseus . if hi gas is optically thick , the measured brightness temperature @xmath58 levels off at a temperature close to the kinetic temperature @xmath181 .
therefore , under the assumption of optically thin hi gas ( which we used to calculate our @xmath1 ) , @xmath1 would saturate as @xmath1 @xmath157 @xmath58 @xmath3 @xmath181 and therefore be underestimated .
there are a few lines of evidence that indicate the presence of cold , optically thick cnm in perseus .
for example , hisa features were found toward the centers of several dark and star - forming regions ( e.g. , ridge et al .
in addition , heiles & troland ( 2003b ) analyzed the hi emission / absorption spectra obtained toward two radio sources behind perseus , 3c93.1 and nrao140 , and found cnm components in perseus ( based on radial velocity information ) that have optical depths @xmath182 @xmath3 1 and @xmath37 . the best way to estimate
the effect of optical depth is to measure cnm in absorption in the direction of many background radio sources behind perseus .
this provides the total column density of atomic hydrogen gas ( @xmath183 ) along each line of sight , which can be compared with the column density we measured in section 3.2 to derive the optical depth corrections for the whole cloud .
as only two such measurements exist so far , we are in the process of obtaining additional hi absorption measurements . here
we investigate only the magnitude of the optical depth effect .
if we assume @xmath184 = 70 k as measured by heiles & troland ( 2003b ) , the corrected @xmath18(hi ) will need to be calculated using the following equation : @xmath185 if we apply this equation to our @xmath18(hi ) image , we find that @xmath1 changes by a factor of 1.22 .
another approach involves using the statistics of cnm and wnm from heiles & troland ( 2003b ) .
their table 1 summarizes the cnm and wnm column densities for all sources in their survey .
we can then estimate the fraction of cnm column density ( relative to the total column density ) as a function of wnm column density .
using only the table entries for a region centered on perseus and @xmath31200 deg@xmath31 in size , we find that the cnm fraction is close to 50% for wnm column density less than 1.2 @xmath11 10@xmath59 cm@xmath7 .
this again suggests that our @xmath18(hi ) or @xmath1 could be underestimated by a factor of two .
our excercise shows that a high optical depth along each line of sight could result in a factor of two spread in @xmath1 for perseus .
this is not substantial and therefore the almost constant @xmath1 we measured is likely a real physical feature , possibly driven by the conversion of hi into h@xmath0 .
however , the magnitude of the optical depth effect we estimated is approximate and measuring the true @xmath1 distribution across perseus needs a more sophisticated investigation .
we now investigate kmt09 s prediction of @xmath22 vs @xmath9 for the dark and star - forming regions in perseus . in figure
[ f : rh2_sigmatotal ] , we plot @xmath22 as a function of @xmath9 for b5 , ic348 , b1e , b1 , and ngc1333 . for each plot
we use the data points in the rectangular boxes shown in figure [ f : rh2 ] .
we find that the relation between @xmath22 and @xmath9 is remarkably consistent for all dark and star - forming regions .
each plot in figure [ f : rh2_sigmatotal ] shows a very sharp rise of @xmath22 at low @xmath9 , a turnover around @xmath22 @xmath3 1 , and a slow increase toward higher @xmath22 .
for the two star - forming regions , @xmath22 reaches a value of @xmath310 , while for the three dark regions , @xmath22 is less than 5 . in figure
[ f : rh2_sigmatotal ] , we overlay kmt09 s predictions as red solid curves and find excellent agreement . the overlaid curves were derived using equation ( 17 ) with the assumption of @xmath187 and @xmath188 for all dark and star - forming regions . the fiducial value @xmath189 was adopted following the discussion in kmt09 .
we note that @xmath162 does not make a significant change in @xmath22 .
for example , between @xmath187 and 50 , @xmath22 at @xmath9 = 100 m@xmath6 pc@xmath7 only varies by a factor of 1.1 .
our assumption of a single @xmath188 is based on the result of gonzlez hernndez et al .
( 2009 ) and our dgr investigation in section 4.5 .
gonzlez hernndez et al . (
2009 ) studied the chemical composition of ernis 52 , an a3 v - type star in the direction of perseus and derived a metallicity of [ fe / h ] = @xmath190 ( corresponding to 0.71.4 z@xmath6 ) .
they determined a distance to ernis 52 of 231@xmath191 pc and in conjuction with the radial velocity information concluded that ernis 52 is likely a member of ic348 . as a single dgr fits most of the diffuse regions in perseus , there is probably no significant variation of @xmath154 across the cloud .
therefore , we adopt @xmath188 for all dark and star - forming regions in perseus throughout this paper and constrain only @xmath156 from our fitting vs @xmath9 by using both @xmath192 and @xmath156 as free parameters but have found that the results are consistent with the case of @xmath188 and varying @xmath156 within 1@xmath109 . ] .
we performed monte carlo simulations to determine the best - fit curves with consideration of uncertainties in both @xmath22 and @xmath9 .
we added random offsets to @xmath22 and @xmath9 and repeated this process 1000 times . for each variable
, offsets were drawn from a gaussian distribution with the standard deviation equal to the measured noise .
for each realization , we determined the best - fit curve by assuming @xmath187 and @xmath188 and finding @xmath156 that minimizes the sum of the squares of the residuals .
idl routine mpfitfun ( markwardt 2009 ) was used for the fitting . finally , we estimate the median @xmath156 among the simulated 1000 @xmath156 and use it as the best - fit parameter .
we summarize this best - fit @xmath156 for each region in table 1 .
we find that @xmath156 ranges from @xmath36 to @xmath310 .
as @xmath156 determines the typical cnm number density via @xmath193 , @xmath156 = 610 translates into @xmath194 = 6075 k under the assumption of thermal pressure allowing the coexistence of cnm and wnm ( see equation 19 of kmt09 ) .
this temperature range is indeed consistent with typical cnm properties in the solar neighborhood .
heiles & troland ( 2003b ) examined hi emission / absorption spectra obtained toward 79 radio sources and found a median @xmath195 of @xmath370 k for cnm .
while @xmath156 appears to systematically decrease toward the west of perseus , we find no significant difference in @xmath156 between the dark and star - forming regions .
this implies that hi envelopes surrounding these regions have similar @xmath139 .
we estimate @xmath139 = 2030 @xmath128 using @xmath196 = @xmath197 = 0.4 derived from equation ( 5 ) with @xmath73 = 17 k. the similar @xmath139 further suggests a similar normalized radiation field @xmath142 of 11.5 .
as @xmath142 measures the relative importance of dust shielding and h@xmath0 self - shielding , @xmath198 implies that dust shielding and h@xmath0 self - shielding are equally important for h@xmath0 formation in perseus .
kmt09 assume an infinitely sharp hi
h@xmath0 transition and in this case @xmath22 would be zero at the hi shielding column density .
however , in reality , the transition can not be infinitely sharp and kmt09 estimate @xmath22 @xmath3 0.25 at the hi shielding column density by comparing the theoretical hi shielding column densities with the detailed numerical radiative transfer calculations .
for solar metallicity , kmt09 predict ( as shown with the red solid curves ) that @xmath22 @xmath3 0.25 occurs at @xmath9 = 710 m@xmath6 pc@xmath7 .
this is indeed consistent with what we see in figure [ f : rh2_sigmatotal ] ; @xmath22 @xmath3 0.25 at @xmath9 = 612 m@xmath6 pc@xmath7 or @xmath10 = ( 814 ) @xmath11 10@xmath12 cm@xmath7 .
this transition column density is a factor of 23 larger than what previous studies found in the galaxy via uv absorption measurements ( e.g. , savage et al .
1977 ; gillmon et al . 2006 ) .
however , considering that the previous studies defined the transition column density at @xmath22 @xmath3 0.1 and @xmath2 linearly increases with @xmath9 , our result is consistent with the previous studies . to investigate how @xmath22 radially changes from a molecular peak , we plot @xmath22 as a function of distance from the center of each dark and star - forming region in figure [ f : rh2_radial ] .
we use the data points in the rectangular boxes ( see figure [ f : rh2 ] ) and calculate the mean @xmath22 in three - pixel size bins ( bin size = 12.9@xmath26 , corresponding to 1.1 pc at the distance of 300 pc ) to remove small - scale fluctuations . for all five regions , we find two power - law functions ( @xmath199 ) in their radial profiles up to @xmath200 . beyond @xmath200 ,
the radial profiles show slight fluctuations due to the presence of small - scale structure in the @xmath22 image and/or become flat .
we perform a linear least - squares fit to each radial profile up to @xmath200 and summarize the results in table 2 with other properties of the radial profiles .
we list the slopes of power - law functions ( @xmath201 and @xmath202 ) , the radius where the transition from the shallow to steep power - law functions occurs ( @xmath203 ) , @xmath22 at @xmath203 ( @xmath204 ) , the radius where the hi h@xmath0 transition occurs ( @xmath205 ) , the radius where hydrogen is predominantly in the form of hi ( @xmath206 ) , and the thickness of the hi h@xmath0 transition region ( @xmath207 ) .
we define @xmath205 as a radius at @xmath22 @xmath3 0.25 , @xmath206 as a radius at @xmath22 @xmath3 0.1 , and @xmath207 as @xmath208 .
we find no significant difference between the dark and star - forming regions in terms of their @xmath22 radial distribution .
the shallow portion of the radial profiles has a slope of @xmath209 = 0.40.7 , while the steep portion has a slope of @xmath210 = 1.43.7 .
the transition from the shallow to the steep power - law function occurs at @xmath203 @xmath3 3090@xmath26 or @xmath338 pc . at this @xmath203
, @xmath22 ranges from 0.5 to 1.2 .
in addition , we find that h@xmath0 extends up to @xmath211 200@xmath26 or @xmath317 pc from the centers of the dark and star - forming regions .
this suggests that molecular clouds are much more extended than what is suggested by molecular tracers such as co. the hi
h@xmath0 transition , however , seems to be relatively sharp .
we estimate @xmath207 = @xmath208 , which measures the size of diffuse h@xmath0 layer and therefore could be considered as the thickness of the transition region , and find @xmath207 @xmath3 3060@xmath26 , corresponding to @xmath335 pc .
we then estimate the ratio of the thickness of the hi
h@xmath0 transition region to the region size , @xmath212
< @xmath213 = 0.20.4 .
as we do not know the exact size of the dark and star - forming regions , @xmath213 provides only an upper limit on @xmath212 .
the other interesting thing to note is that the @xmath22 radial distribution becomes smoother toward the west of perseus .
namely , @xmath214 becomes smaller from b5 to ngc1333 .
this may be related to @xmath139 decreasing from b5 to ngc1333 ( inferred from @xmath156 decreasing from b5 to ngc1333 with the same @xmath155 ; see section 7.2.1 ) , even though it is not clear how @xmath139 affects the smoothness of @xmath22 radial distribution . in kmt09 s model
, @xmath139 determines the location of the hi
h@xmath0 transition and due to the assumption of the infinitely sharp transition , there is no such relation between @xmath139 and the smoothness of the @xmath22 radial distribution .
this systematic variation of @xmath215 across perseus could be resulted from the change of internal density structure and/or the complicated viewing geometry .
in figure [ f : h2_co ] , we compare the h@xmath0 and co distributions .
the black and gray contours show the cfa @xmath39 data and our @xmath2 data , respectively . to ease the comparison , we plot only 3@xmath109 contours for both data sets .
figure [ f : h2_co ] shows that h@xmath0 is found wherever significant co exists .
however , the reverse is not true ; a considerable amount of h@xmath0 exists beyond the 3@xmath109 contour of @xmath39 .
while h@xmath0 is generally more extended than co , there are significant spatial variations .
for example , co and h@xmath0 trace each other remarkably well in the southwest part of perseus , while h@xmath0 is significantly more extended elsewhere ( particularly around b5 , b1e , b1 , ic348 , and ngc1333 ) . to quantify the relative distribution of h@xmath0 and co , we plot @xmath2 and @xmath39 as a function of distance from the center of each dark and star - forming region in figure [ f : h2_co_radial ] .
we first smooth the @xmath2 image to the angular resolution of the cfa @xmath39 image and use the data points in the rectangular boxes shown in figure [ f : rh2 ] .
we then calculate the mean @xmath2 and @xmath39 in two - pixel size bins ( bin size = 16.8@xmath26 , corresponding to 1.5 pc at the distance of 300 pc ) .
finally , each radial profile is normalized by its peak value and therefore figure [ f : h2_co_radial ] shows the relative distribution of h@xmath0 and co. as spatial averaging improves snr , the radial profiles of @xmath2 and @xmath39 allow us to investigate the relative distribution of h@xmath0 and co more reliably than the visual comparison of 3@xmath109 contours .
we find that all five regions have extended h@xmath0 compared to co at all radii .
for ic348 , b1e , and ngc1333 , h@xmath0 radial profiles show slight fluctuations and/or become flat at @xmath217 , which has been seen in figure [ f : rh2_radial ] .
we measure the radius where each h@xmath0 and co radial profile drops off to 1/4 of its peak value ( @xmath218 and @xmath219 ) and calculate the ratio of @xmath218 to @xmath219 .
we find that the ratio , @xmath220 , ranges from 1.2 to 2 , with a median of 1.4 .
this suggests that on average h@xmath0 is @xmath31.4 times more extended than co in perseus , implying the existence of a substantial amount of `` co - dark '' gas .
the `` co - dark '' gas refers to interstellar gas that exists in the form of h@xmath0 along with ci and cii , but little or no co. such gas has been predicted from theoretical models of pdrs , which showed that the cii co transition occurs deeper into an interstellar cloud than the hi
transition due to the more effective self - shielding for h@xmath0 than co ( e.g. , van dishoeck & black 1988 ; tielens & hollenbach 1985 ) .
recently , wolfire et al .
( 2010 ; whm10 hereafter ) modeled the `` co - dark '' gas and explored how the fraction of h@xmath0 gas mass in this `` co - dark '' component depends on the mass of a cloud ( @xmath169 ) , the uv radiation field ( @xmath196 ) , the metallicity ( @xmath154 ) , and the mean @xmath4-band extinction ( @xmath221 ) .
they used a pdr code to calculate the h@xmath0 , ci , and cii abundances in a spherical cloud and found that the fraction of `` co - dark '' gas , @xmath14 , is @xmath30.3 for solar metallicity over a wide range of @xmath169 and @xmath196 . here
@xmath14 is defined as @xmath14 = ( @xmath222)/@xmath223 , where @xmath224 and @xmath225 is the radius of the h@xmath0 and co part of the cloud , respectively .
formally , @xmath224 is defined as the radius where 2@xmath226 and @xmath225 is defined as the radius where the optical depth from the cloud surface to @xmath225 is 1 .
more simply , when expressed in terms of @xmath227 , @xmath14 depends only on the ratio @xmath228 .
this is because @xmath229 is a measure of the `` co - dark '' gas mass and @xmath230 is a measure of the h@xmath0 gas mass .
interestingly , whm10 found that @xmath229 is a weak function of @xmath196 and is almost constant .
therefore , @xmath14 is strongly dependent only on @xmath230 .
for example , for a cloud with co mass of @xmath231 m@xmath6 and solar metallicity , @xmath232 is expected at @xmath233 mag , while @xmath234 is expected at @xmath235 mag . to compare our observations to whm10 s prediction
, we calculate the h@xmath0 mass within the 3@xmath109 boundary of @xmath39 ( @xmath236 ) and compare it with the h@xmath0 mass within the 3@xmath109 boundary of @xmath2 ( @xmath237 ) .
we find @xmath14 = ( @xmath238)/@xmath239 0.3 .
note that our definition of @xmath14 is different from that of whm10 .
this is because @xmath225 can not be directly measured from observations .
nevertheless , @xmath223 and @xmath240 of whm10 represent the total h@xmath0 and co masses and therefore our @xmath14 is conceptually similar with what whm10 used .
we estimate @xmath241 1.5 mag for the data points within @xmath242 . as whm10 do not provide a prediction of @xmath14 at @xmath221 < 2 mag , we extrapolate it from their figure 11 ;
@xmath243 0.8 is expected at @xmath241 1.5 mag .
note that the expected @xmath243 0.8 at @xmath241 1.5 mag is for a cloud that has a much higher mass ( @xmath244 m@xmath6 ) and a stronger radiation field ( @xmath245 ) than perseus .
considering the predicted insensitivity of @xmath14 to the cloud mass and the strength of radiation field , however , we measure a factor of three lower value , @xmath243 0.3 .
to investigate this discrepancy between our @xmath14 and whm10 s prediction , we first examine the dependence of @xmath14 on the dgr .
as shown in section 4.5 , our estimate of @xmath2 and consequently @xmath14 depends on the dgr via equation ( 4 ) . in the same section
, we discussed the lower and upper limits on the dgr .
if we use the lower limit on the dgr instead of our adopted dgr , @xmath221 decreases to 1.1 mag and @xmath14 increases to 0.6 .
our estimate is a factor of two lower than whm10 s prediction at @xmath221 @xmath3 1.1 mag , @xmath14 @xmath3 0.9 .
similarly , if we use the upper limit on the dgr , @xmath221 increases to 1.9 mag and @xmath14 decreases to 0.2 .
our estimate is now a factor of four lower than whm10 s prediction at @xmath221 @xmath3 1.9 mag , @xmath14 @xmath3 0.8 .
note that `` co - dark '' gas exists in perseus even for these extreme cases and the discrepancy with the model persists .
second , we test if @xmath14 is a strong function of @xmath221 as whm10 predicted . for this purpose
, we divide the @xmath2 image into four regions ( see figure [ f : h2_co ] ) and estimate @xmath14 and @xmath221 for each region .
we use the @xmath2 image derived from our adopted dgr , 1.1 @xmath11 10@xmath108 mag cm@xmath31 . in table 3 , we list the estimated @xmath14 and @xmath221 for the four regions .
we find @xmath14 @xmath3 0.3 for regions 1 and 2 and @xmath14 = 0.10.2 for regions 3 and 4 .
all regions have similar mean extinction , @xmath221 = 1.41.7 mag .
while we have only four data points covering a small range of @xmath221 , we measure a relatively large range of @xmath14 , from @xmath30.1 to @xmath30.3 .
the large variation we find in @xmath14 is not consistent with whm10 , but echos the observations of other diffuse clouds in the galaxy .
for example , the polaris cloud has @xmath241 0.3 mag ( heithausen & thaddeus 1990 ) and @xmath243 0.36 ( abdo et al .
2010 ) . on the other hand
, the pegasus cloud has @xmath241 0.6 mag ( yamamoto et al .
2006 ) and @xmath243 0.6 ( grenier et al .
2005 ) .
this region - to - region variation we find in @xmath14 could be due to the dgr systematically changing across perseus .
for example , a factor of two lower dgr would allow regions 3 and 4 to have similar @xmath14 as regions 1 and 2 .
however , we do not find any evidence for such systematic change of the dgr when examining the correlation between @xmath18(hi ) and fir emission ( see section 4.2 ) . as an alternative
, the complicated viewing geometry of the h@xmath0 and co distributions could result in the regional variation in @xmath14 .
the excellent agreement between our data and kmt09 s predictions ( sections 7.1 and 7.2 ) suggests that kmt09 s model captures the fundamental physics of h@xmath0 formation reasonably well even on sub - pc scales .
this is certainly encouraging , but slightly surprising considering how simple kmt09 s model is .
essentially , for a spherical cloud bathed in a uniform and isotropic isrf , the most important factor for h@xmath0 formation is the hi shielding surface density @xmath149 . once enough shielding against photodissociation
is achieved by @xmath149 , all additional @xmath1 is fully converted into @xmath2 , resulting in a uniform @xmath1 distribution and therefore a linear increase of @xmath2 with @xmath9 . before going into a more detailed discussion of several important aspects of kmt09 s model , we would like to remind the reader that two main predictions summarized in section 6 are not independent .
the predicted behavior of @xmath22 as a function of @xmath9 is a consequence of saturation of @xmath1 .
therefore , the excellent agreement between our data and kmt09 s predictions in figure 12 is entirely driven by the constant @xmath1 .
while we are not able to probe the purely atomic envelope of perseus , the constant @xmath1 of 68 m@xmath6 pc@xmath7 we find is in agreement with kmt09 s prediction of @xmath149 for solar metallicity .
in addition , we caution that our data do not allow us to fully test the dependence of @xmath22 on metallicity and isrf .
kmt09 s model predicts that @xmath22 depends on metallicity , while being almost insensitive to the strength of isrf .
while our findings are in agreement with kmt09 s predictions for solar metallicity and @xmath32/5 of the galactic isrf , the lack of metallicity and isrf variations across perseus does not permit more detailed model testing .
future studies of high - latitude molecular clouds and gmcs in external galaxies with sub - solar metallicity will provide crucial tests of kmt09 s model .
we now turn to several crucial model assumptions and discuss their possible limitations .
the most fundamental principle underlying kmt09 s model is h@xmath0 formation in equilibrium . in equilibrium , h@xmath0 formation and photodissociation
are balanced and h@xmath0 abundance is locally constant .
several authors have investigated the time evolution of h@xmath0 and found a typical timescale of h@xmath0 formation in equilibrium , @xmath246 = 1030 myr ( e.g. , goldsmith et al .
2007 ; liszt 2007 ) as @xmath246 is comparable to or longer than the estimated lifetime of gmcs , @xmath247 myr ( mckee & ostriker 2007 and references therein ) , the role of interstellar turbulence in accelerating h@xmath0 formation has received significant attention in recent years ( e.g. , hartmann et al .
2001 ; bergin et al .
2004 ; glover & mac low 2007 ; glover et al .
the main idea behind the models of h@xmath0 formation in turbulent media is that shocks induced by cloud cloud collisions , spiral density waves , supernovae explosions , etc . ,
compress diffuse atomic gas , allowing h@xmath0 to form in the postshock gas where high enough shielding is achieved against photodissociation . while earlier studies have been limited to 1d , glover & mac low ( 2007 ) and glover et al .
( 2010 ) performed 3d magnetohydrodynamical ( mhd ) simulations of turbulence with detailed chemical modeling .
they showed that a significant amount of h@xmath0 could form within 12 myr , due to the turbulence - driven clumping that accelerates h@xmath0 formation .
however , the h@xmath0 abundance keeps increasing over time and never reaches equilibrium by 2030 myr .
eventually , the gas will become fully molecular unless the h@xmath0 formation is terminated by destruction of gmcs via stellar feedback , e.g. , photoevaporation by hii regions and protostellar outflows .
this sharply contrasts with kmt09 s model .
our results are consistent with the model of h@xmath0 formation in equilibrium and suggest that turbulence may play a secondary role in the h@xmath0 formation .
clearly , further studies are needed to compare abundances of h@xmath0 and other molecular species with the predictions from both stationary and turbulent models .
we find that @xmath1 is uniform across perseus with a range of 68 m@xmath6 pc@xmath7 , consistent with kmt09 s prediction of @xmath149 for solar metallicity .
however , as cnm is far more effective at shielding against photodissociation than wnm , the predicted @xmath149 comes purely from cnm , while our measured @xmath1 includes the contributions from both cnm and wnm .
as we discussed in section 7.1.1 , our measured @xmath1 could be underestimated by a factor of two for high optical depth gas ( cnm ) .
while @xmath1 corrected for high optical depth would likely still be comparable to kmt09 s prediction , the importance of wnm in the envelopes of atomic - molecular complexes clearly needs to be taken into account in the models of h@xmath0 formation .
this is also supported by bolatto et al .
( 2011 ) who found a relatively good agreement between kmt09 s prediction and the observed @xmath22 vs @xmath248 in the small magellanic cloud , where wnm dominates the total hi ( dickey et al . 2000 ) . at the same time
, it is essential to observationally constrain the relative contribution from cnm and wnm to @xmath1 .
we are in the process of obtaining hi absorption measurements for perseus , which will allow us to separate the contributions from cold and warm gas .
as we showed in section 5.2 , in the case of ic348 and ngc1333 , the embedded b - type stars are the dominant sources of the radiation field .
s model does not take into account the role of internal radiation in determining @xmath22 , with the assumption that massive stars producing significant amounts of dissociating photons do not stay inside their parent molecular clouds for a long time .
while we clearly see the importance of the b - type stars for dust heating , we find no significant difference in our analysis of @xmath22 for the dark and star - forming regions .
this supports the idea that the internal radiation is not important for h@xmath0 formation .
however , as we show below , this is valid for perseus because it does not contain any stars earlier than the b5 v - type .
when o- or b - type stars suddenly turn on in a molecular cloud , uv photons from the stars photodissociate h@xmath0 and create hi dissociation regions around hii regions .
this variation of the uv radiation field has a much shorter timescale ( @xmath310@xmath249 yr ) than h@xmath0 formation ( @xmath310@xmath16 yr ) and therefore totally changes the thermal and chemical structures of the molecular cloud ( e.g. , hollenbach & natta 1995 ; bertoldi & draine 1996 ) .
however , the sphere of photodissociation influence depends greatly on the stellar type .
roger & dewdney ( 1992 ) studied the properties of hi dissociation regions around o- and b - type main sequence stars .
their source list includes svs3 , the b5 v - type star in ngc1333 ( red star in figure [ f : dust_map ] ) .
an hi region associated with svs3 was measured to have @xmath250 = 0.017 m@xmath6 and @xmath251 = 0.03 pc ( rodrguez et al .
this is consistent with an hi dissociation region created by a star of @xmath252 k in a dense medium of @xmath253 @xmath128 .
the size of the hi dissociation region around this particular b5 star is only @xmath31/10 of our spatial resolution .
therefore , the hi dissociation regions around the individual b - type stars in perseus are below our spatial resolution and therefore are negligible for our consideration of @xmath22 .
however , the internal radiation is still important for dust heating inside ic348 and ngc1333 .
this can be understood by the fact that not every uv photon produces photodissociation of h@xmath0 . while uv and visible photons heat dust grains , only the uv photons in the narrower lw band ( 9121120 ) photodissociate h@xmath0 .
in addition , only @xmath310% of the lw photons actually photodissociate h@xmath0 ( @xmath254 ) .
therefore , the uv photons from the b - type stars in ic348 and ngc1333 could be dominant for dust heating , while relatively ineffective for photodissociation of h@xmath0 .
if any stars earlier than b3 v resided in perseus , on the other hand , the role of the internal radiation in determining @xmath22 would be no longer negligible .
for example , roger & dewdney ( 1992 ) found that an hi region associated with iras 23545@xmath356508 has @xmath250 = 1.4 m@xmath6 and @xmath251 = 0.4 pc and these properties are consistent with an hi dissociation region created by a b3 v type star in a medium of @xmath255 @xmath128 .
the size of the hi dissociation region is now comparable to our spatial resolution and therefore @xmath22 around the stars earlier than b3 v would be significantly modified .
therefore , depending on the stellar type of stars residing inside gmcs , the internal radiation should not be ignored .
in this paper , we study the hi h@xmath0 transition across the perseus molecular cloud on sub - pc scales ( @xmath30.4 pc ) .
we use the highest resolution hi data for perseus from the galfa hi survey , which allow us to probe the outskirts as well as the main body of the cloud .
the velocity range of the hi emission associated with perseus is determined based on the correlation between @xmath18(hi ) and the dust column density traced by the 2mass @xmath48 image .
we then derive the @xmath18(hi ) image under the assumption of optically thin hi gas .
the dust column density is estimated using the iris 60 @xmath21 m and 100 @xmath21 m images and is calibrated with the 2mass @xmath48 image . in combination with the @xmath18(hi ) image and the derived dgr , we use the dust column density to derive the @xmath18(h@xmath0 ) image .
the @xmath18(hi ) and @xmath18(h@xmath0 ) images and the existing co data for perseus allow us to test two recent models of h@xmath0 and co formation in equilibrium . for several dark and star - forming regions ( b5 , b1e , b1 , ic348 , and ngc1333 )
, we investigate @xmath1 and @xmath22 as a function of @xmath9 and compare the results to kmt09 s predictions .
we also compare the h@xmath0 and co distributions and estimate the fraction of h@xmath0 mass in the `` co - dark '' component , @xmath14 .
we investigate @xmath14 as a function of @xmath221 and compare the result to whm10 s prediction .
several of the main results can be summarized as follows : \1 . for both dark and star - forming regions in perseus
, we find an almost uniform @xmath1 @xmath3 68 m@xmath6 pc@xmath7 .
this is consistent with kmt09 s prediction of the hi shielding surface density for solar metallicity ,
@xmath256 m@xmath6 pc@xmath7 . \2 .
the purely atomic envelopes are located more than @xmath320 pc from the centers of the dark and star - forming regions .
we probe the @xmath1 and @xmath9 distributions up to @xmath32030 pc from the centers of the regions and barely detect the turnover between atomic- and molecular - dominated zones for ic348 , b1e , and b1 .
the relation between @xmath22 and @xmath9 on a log - linear scale is remarkably consistent for all dark and star - forming regions , having a very sharp rise of @xmath22 at low @xmath9 , a turnover around @xmath22 @xmath3 1 , and a slow increase toward higher @xmath22 .
the dark and star - forming regions have similar @xmath156 = 610 , which implies that these regions are embedded in the atomic envelopes with @xmath139 = 2030 @xmath128 .
in addition , @xmath156 = 610 indicates a similar normalized radiation field @xmath257 , suggesting that dust shielding and h@xmath0 self - shielding are equally important for h@xmath0 formation .
we measure the hi
h@xmath0 transition gas column density of @xmath258 = ( 814 ) @xmath11 10@xmath12 cm@xmath7 or @xmath248 = 612 m@xmath6 pc@xmath7 at @xmath22 @xmath3 0.25 .
this is consistent with the previous estimates for the galaxy .
we estimate the fraction of `` co - dark '' gas , @xmath259 , for perseus , a factor of three lower than whm10 s prediction .
this implies that a significant amount of h@xmath0 gas is not traced by galactic co surveys .
we investigate the dependence of @xmath14 on @xmath221 and find a relatively large variation of @xmath14 @xmath3 0.10.3 for similar @xmath260 mag .
this is not consistent with whm10 s prediction that @xmath14 is a strong function of @xmath221 .
while we generally find good agreement with kmt09 s predictions , several outstanding questions remain .
our estimated @xmath1 @xmath3 68 m@xmath6 pc@xmath7 is likely dominated by wnm , while kmt09 s model predicts shielding by cnm alone .
clearly , the wnm in the atomic envelopes of gmcs needs to be considered in the models of h@xmath0 formation as a source of shielding against photodissociation .
the importance of gravitationally bound vs diffuse gas is another important aspect , as explored in ostriker et al .
in addition , kmt09 s model does not take into account the role of internal radiation for h@xmath0 formation .
this is valid for perseus as it does not contain early - type stars to photodissociate a significant amount of h@xmath0 that could be detected at our spatial resolution .
however , if gmcs are young enough to harbor early - type stars that are able to change the thermal and chemical structures of the gmcs , the internal radiation can not be ignored .
most importantly , our perseus data on sub - pc scales are consistent with kmt09 s model for h@xmath0 formation in equilibrium and suggest that turbulence may not be of primary importance for h@xmath0 formation . yet , the timescale for equilibrium h@xmath0 formation is longer than the estimated lifetime of most gmcs .
this problem remains to be resolved .
future detailed comparisons of observations with non - equilibrium , turbulence - driven h@xmath0 formation models will be important to investigate the role of turbulence for formation of h@xmath0 and various molecular species .
we thank the anonymous referee for suggestions that improved this work .
we also thank ron allen , joss bland - hawthorn , alberto bolatto , bruce elmegreen , fabian heitsch , mordecai mac low , chris mckee , eve ostriker , rene plume , and mark wolfire for stimulating discussions .
y.l . , s.s . , m.e.p . , c.h . , e.j.k . , j.e.g.p . , a.b . , j.g . , n.p . , and d.s
. acknowledge support from nsf grants ast-0707597 , 0917810 , 0707679 , and 0709347 .
y.l . and s.s
. thank the research corporation for science advancement for their support . k.a.d . acknowledges funding from the european community s seventh framework program under grant agreement no piif - ga-2008 - 221289 .
j.e.g.p . was supported by hst - hf-51295.01a , provided by nasa through a hubble fellowship grant from stsci , which is operated by aura under nasa contract nas5 - 26555 .
the arecibo observatory is part of the national astronomy and ionosphere center , which is operated by cornell university under a cooperative agreement with the national science foundation .
the national radio astronomy observatory is a facility of the national science foundation operated under cooperative agreement by associated universities , inc .
we credit the use of the karma visualization software ( gooch 1996 ) .
abdo , a. a. , ackermann , m. , ajello , m. et al .
2010 , apj , 710 , 133 + bally , j. , walawender , j. , johnstone , d. , kirk , h. , & goodman , a. a. 2008 , handbook of star forming regions , volume i : the northern sky , ed . b. reipurth ( asp monograph publications , vol .
4 ; san francisco , ca : asp ) , 308 + barriault , l. , joncas , g. , lockman , f. j. , & martin p. g. 2010 , mnras , 407 , 2645 + beichman , c. a. , wilson , r. w. , langer , w. d. , & goldsmith , p. f. et al .
1988 , apj , 332 , l81 + bergin , e. a. , hartmann , l. w. , raymond , j. c. , & ballesteros - paredes , j. 2004 , apj , 612 , 921 + bertoldi , f. , & draine , b. t. 1996 , apj , 458 , 222 + bigiel , f. , leroy , a. k. , walter , f. et al .
2008 , aj , 136 , 2846 + blitz , l. , & rosolowsky , e. 2004 , apj , 612 , 29 + blitz , l. , & rosolowsky , e. 2006 , apj , 650 , 933 + bolatto , a. , leroy , a. k. , jameson , k. et al .
2011 , apj , 741 , 12 + browning , m. k. , tumlinson , j. , & shull , j. m. 2003 , apj , 582 , 810 + bohlin , r. c. , savage , b. d. , & drake , j. f. 1978 , apj , 224 , 132 + boulanger , f. , abergel , a. , bernard , j
1996 , a&a , 312 , 256 + ernis , k. 1990 , ap&ss , 166 , 315 + dame , t. m. , hartmann , d. , & thaddeus , p. 2001
, apj , 547 , 792 + dsert , f .- x . , boulanger , f. , & puget , j. l. 1990 , a&a , 237 , 215 + dickey , j. m. , mebold , u. , stanimirovi , s. , staveley - smith , l. 2000 , apj , 536 , 756 + douglas , k. a. , & taylor , a. r. 2007 , apj , 659 , 426 + draine , b. t. , & bertoldi , f. 1996 , 468 , 269 + fukui , y. , & kawamura , a. ara&a , 2010 , 48 , 547 + fumagalli , m. , krumholz , m. r. , prochaska , j. x. , gavazzi , g. , & boselli , a. 2009 , apj , 697 , 1811 + fumagalli , m. , krumholz , m. r. , & hunt , l. k. 2010 , apj , 722 , 919 + gibson , s. j. 2010 , aspc , 438 , 111 + gillmon , k. , shull , j. m. , tumlinson , j. , & danforth , c. 2006 , apj , 636 , 891 + giovanelli , r. , haynes , m. p. , kent , b. r. et al .
2005 , aj , 130 , 2598 + glover , s. o. c. , & mac low , m .- m .
2007 , apj , 659 , 1317 + glover , s. o. c. , federrath , c. , mac low , m .-
m . , & klessen , r. s. 2010 , mnras , 404 , 2 + goldsmith , p. f. , li , d. , & kro , m. 2007 , apj , 654 , 273 + gonzlez hernndez , j. , iglesias - groth , s. , rebolo , r. et al .
2009 , apj , 706 , 866 + gooch , r. 1996 , astronomical data analysis software and systems v ( asp conf .
101 ) , ed .
g. h. jacoby & j. barnes ( san francisco , ca : asp ) , 80 + goodman , a. a. , pineda , j. e. , & schnee , s. l. 2009 , apj , 692 , 91 ( gps09 ) + grenier , i. a. , casandjian , j .- m . , & terrier , r. 2005 , science , 307 , 1291 + guram , s. s. , & taylor , a. r. 2009 , the low - frequency radio universe ( asp conf . ser .
407 ) , ed .
d. j. saikia et al .
( san francisco , ca : asp ) , 282 + hartmann , l. w. , ballesteros - paredes , j. , & bergin , e. a. 2001 , 562 , 852 + heithausen , a. , & thaddeus , p. 1990
, apj , 353 , l49 + heiles , c. 2005 , obtaining the if bandpass using least - squares frequency switching ( lsfs ) , http://www.naic.edu/ astro / aotms / performance / lsfs.ps + heiles , c. , & troland , t. h. 2003a , apjs , 145 , 329 + heiles , c. , & troland , t. h. 2003b , apj , 586 , 1067 + herbig , g. h. , & jones , b. f. 1983 , aj , 88 , 1040 + herbig , g. h. 1998 , apj , 497 , 736 + hollenbach , d. , & natta , a. 1995 , apj , 455 , 133 + hollenbach , d. , & tielens , a. g. g. m. 1997 , ara&a , 35 , 179 + imara , n. , & blitz , l. 2011 , apj , 732 , 78 + kalberla , p. m. w. , burton , w. b. , & hartmann , d. et al .
2005 , a&a , 440 , 775 + klochkova , v. g. , & kopylov , i. m. 1985 , astrofiz .
izv . spets .
obs . , 20 , 3 + kim , s .- h .
, & martin , p. g. 1996 , apj , 462 , 296 + krumholz , m. r. , mckee , c. f. , & tumlinson , j. 2008 , apj , 689 , 865 ( kmt08 ) + krumholz , m. r. , mckee , c. f. , & tumlinson , j. 2009 , apj , 693 , 216 ( kmt09 ) + lada , c. j. , alves , j. , & lada , e. a. 1996 , aj , 111 , 1964 + lequeux , j. 2005 , the interstellar medium ( berlin : springer ) + leroy , a. k. , bolatto , a. d. , bot , c. et al .
2009 , apj , 702 , 352 + lesh , j. r. 1969 , aj , 74 , 891 + li , a. & draine , b. t. 2001 , apj , 554 , 778 + liszt , h. s. , & lucas , r. 2000 , a&a , 355 , 333 + liszt , h. s. 2002 , a&a , 389 , 393 + liszt , h. s. 2007 , a&a , 461 , 205 + lombardi , m. & alves , j. 2001 , a&a , 377 , 1023 + luhman , k. l. , stauffer , j. r. , muench , a. a. et al .
2003 , apj , 593 , 1093 + markwardt , c. b. 2009 , asp conf
. ser . 411 , astronomical data analysis software and systems xviii , ed .
d. a. bohlender , d. durand , & p. dowler ( san francisco , ca : asp ) , 251 + mckee , c. f. , & ostriker , e. c. 2007 , ara&a , 45 , 565 + meyerdierks , h. & heithausen , a. 1996 , a&a , 313 , 929 + miville - desch@xmath38nes , m .- a . & lagache , g. 2005 , apjs , 157 , 302 + morton , d. c. , & adams , t. f. 1968 , apj , 151 , 611 + nelder , j. , & meade , r. 1965 , comput .
j. , 7 , 308 + narayanan , g. , heyer , m. h. , brunt , c. et al .
2008 , apjs , 177 , 341 + ostriker , e. c. , mckee , c. f. , & leroy , a. k. 2010 , apj , 721 , 9750 + peek , j. e. g. , heiles , c. , douglas , k. a. et al .
2011 , apjs , 194 , 20 + porter , t. a. & strong , a. w. 2005 , icrc ( pune ) , arxiv : astro - ph/0507119 + rachford , b. l. , snow , t. p. , tumlinson , j. et al .
2002 , apj , 577 , 221 + rachford , b. l. , snow , t. p. , destree , j. d. et al .
2009 , apjs , 180 , 125 + reach , w. t. , koo , b .-
c . , & heiles , c. 1994 , apj , 429 , 672 + ridge , n. a. , di francesco , j. , kirk , h. et al .
2006a , aj , 131 , 2921 + ridge , n. a. , schnne , s. l. , goodman , a. a. , & foster , j. b. 2006b , apj , 643 , 932 + rodrguez , l. f. , escalante , v. , lizano , s. , canto , j. , & mirabel , i. f. 1990 , 365 , 261 + roger , r. s. , & dewdney , p. e. 1992 , apj , 385 , 536 + sancisi , r. goss , w. m. , anderson , c. , johansson , l. e. b. , & winnberg , a. 1974 , a&a , 35 , 445 + savage , b. d. , bohlin , r. c. , drake , j. f. , & budich , w. 1977 , apj , 216 , 291 + schlegel , d. j. , finkbeiner , d. p. , & davis , m. 1998 , apj , 500 , 525 ( sfd98 ) + schnee , s. l. , ridge , n. a. , goodman , a. a. , & li , j. g. 2005 , apj , 634 , 442 ( s05 ) + snow , t. p. , hanson , m. m. , seab , g. c. , & saken , j. m. 1994 , apj , 420 , 632 + spaans , m. , tielens , a. g. g. m. , van dishoeck , e. f. , & bakes , e. l. o. 1994 , apj , 437 , 270 + spaans , m. , & neufeld , d. a. 1997 , apj , 484 , 785 + stepnik , b. , abergel , a. , bernard , j .-
2003 , a&a , 398 , 551 + tielens , a. g. g. m. , & hollenbach , d. 1985 , apj , 291 , 772 + ungerechts , h. , & thaddeus , p. 1987
, apjs , 63 , 645 + van dishoeck , e. f. , & black , j. h. 1986 , apjs , 62 , 109 + van dishoeck , e. f. , & black , j. h. 1988 , apj , 334 , 771 + wannier , p. g. , lichten , s. m. , & morris , m. 1983 , apj , 268 , 727 + wannier , p. g. , lichten , s. m. , andersson , b .-
g . , & morris , m. 1991 , apjs , 75 , 987 + wolfire , m. g. , mckee , c. f. , hollenbach , d. , & tielens , a. g. g. m. 2003 , apj , 587 , 278 + wolfire , m. g. , hollenbach , d. , & mckee , c. f. 2010 , apj , 716 , 1191 ( whm10 ) + wong , t. , & blitz , l. 2002 , apj , 569 , 157 + wong , t. , hughes , a. , fukui , y. et al .
2009 , apj , 696 , 370 + yamamoto , h. , kawamura , a. , tachihara , k. et al .
2006 , apj , 642 , 307 + 2mass @xmath48 image of perseus .
the contours of cfa @xmath39 are overlaid in black .
the contour levels range from 10% to 90% of the peak ( 69 k km s@xmath27 ) with 10% steps .
the angular resolution of the 2mass @xmath48 and cfa @xmath39 images is 5@xmath26 and 8.4@xmath26 , respectively .
several prominent dark and star - forming regions are labeled . ]
hi ( thick ) and co ( thin ) spectra at the position of ic348 . for comparison
, co is amplified twice .
the main component of hi peaks at @xmath262 km s@xmath27 and is offset by 23 km s@xmath27 from the peak of co. ] correlation coefficient between @xmath18(hi ) and 2mass @xmath48 as a function of velocity width @xmath54 .
data points with snr > 5 in the 2mass @xmath48 image are used to calculate the correlation coefficient . ]
@xmath18(hi ) image at an angular resolution of 4.3@xmath26 .
hi emission is integrated from @xmath29 = @xmath263 km s@xmath27 to @xmath41 km s@xmath27 and @xmath18(hi ) is derived under the assumption of optically thin hi gas . ] ( left ) @xmath66 as a function of @xmath18(hi ) .
( right ) @xmath67 as a function of @xmath18(hi ) .
the gray data points are assumed to mostly trace perseus and the black data points are likely from the taurus molecular cloud .
see section 4.2.,title="fig : " ] ( left ) @xmath66 as a function of @xmath18(hi ) .
( right ) @xmath67 as a function of @xmath18(hi ) .
the gray data points are assumed to mostly trace perseus and the black data points are likely from the taurus molecular cloud . see section 4.2.,title="fig : " ] @xmath73 image derived from the iris 60 @xmath21 m and 100 @xmath21 m images after accounting for the contribution from vsgs .
several b - type stars as potential dust heating sources are marked ; b8 v - type stars ( bd@xmath264540 and bd@xmath264549 ) are shown as blue crosses ( ernis 1990 ) ; b5 v - type stars ( bd@xmath265643 and svs3 ) are shown as red stars ( ernis 1990 ; luhman et al .
2003 ) ; b4 iv v - type star ( bd@xmath266566 ) is shown as a red triangle ( ernis 1990 ) . ]
@xmath48 as a function of @xmath102 .
the contour levels correspond to 50% , 30% , and 10% of the total data points .
the black circles are median @xmath48 values in 3 @xmath11 10@xmath110 cm@xmath7 wide bins .
the linear least - squares fit for the median @xmath48 values is shown as a black dashed line . ]
histograms of @xmath73 .
the median @xmath73 is given in the top right corner of each plot .
( left top ) b5 .
( right top ) ic348 .
( left bottom ) b1e.,title="fig : " ] histograms of @xmath73 .
the median @xmath73 is given in the top right corner of each plot .
( left top ) b5 .
( right top ) ic348 .
( left bottom ) b1e.,title="fig : " ] histograms of @xmath73 .
the median @xmath73 is given in the top right corner of each plot .
( left top ) b5 .
( right top ) ic348 .
( left bottom ) b1e.,title="fig : " ] @xmath1 as a function of @xmath9 .
all data points in the rectangular boxes ( see figure [ f : rh2 ] ) are used for plotting .
the median 3@xmath109 values of @xmath1 and @xmath9 for the whole perseus cloud are shown as the black dashed lines .
the red curves are the best - fit model curves determined in section 7.2.1 .
the best - fit parameters are summarized in the top right corner of each plot .
( left top ) b5 .
( right top ) ic348 .
( left bottom ) b1e.,title="fig : " ] @xmath1 as a function of @xmath9 .
all data points in the rectangular boxes ( see figure [ f : rh2 ] ) are used for plotting .
the median 3@xmath109 values of @xmath1 and @xmath9 for the whole perseus cloud are shown as the black dashed lines .
the red curves are the best - fit model curves determined in section 7.2.1 .
the best - fit parameters are summarized in the top right corner of each plot .
( left top ) b5 .
( right top ) ic348 .
( left bottom ) b1e.,title="fig : " ] @xmath1 as a function of @xmath9 .
all data points in the rectangular boxes ( see figure [ f : rh2 ] ) are used for plotting .
the median 3@xmath109 values of @xmath1 and @xmath9 for the whole perseus cloud are shown as the black dashed lines .
the red curves are the best - fit model curves determined in section 7.2.1 .
the best - fit parameters are summarized in the top right corner of each plot .
( left top ) b5 .
( right top ) ic348 .
( left bottom ) b1e.,title="fig : " ] @xmath22 as a function of @xmath9 .
all data points in the rectangular boxes ( see figure [ f : rh2 ] ) are used for plotting .
the median 3@xmath109 values of @xmath22 and @xmath9 for the whole perseus cloud are shown as the black dashed lines .
the best - fit model curves determined in section 7.2.1 . are shown in red .
the best - fit parameters are summarized in the lower right corner of each plot .
( left top ) b5 .
( right top ) ic348 .
( left bottom ) b1e.,title="fig : " ] @xmath22 as a function of @xmath9 .
all data points in the rectangular boxes ( see figure [ f : rh2 ] ) are used for plotting .
the median 3@xmath109 values of @xmath22 and @xmath9 for the whole perseus cloud are shown as the black dashed lines .
the best - fit model curves determined in section 7.2.1 . are shown in red .
the best - fit parameters are summarized in the lower right corner of each plot .
( left top ) b5 .
( right top ) ic348 .
( left bottom ) b1e.,title="fig : " ] @xmath22 as a function of @xmath9 .
all data points in the rectangular boxes ( see figure [ f : rh2 ] ) are used for plotting .
the median 3@xmath109 values of @xmath22 and @xmath9 for the whole perseus cloud are shown as the black dashed lines .
the best - fit model curves determined in section 7.2.1 . are shown in red .
the best - fit parameters are summarized in the lower right corner of each plot .
( left top ) b5 .
( right top ) ic348 .
( left bottom ) b1e.,title="fig : " ] @xmath22 radial profiles .
the data points with snr of @xmath22 > 1 are selected from the rectangular boxes ( see figure [ f : rh2 ] ) and the mean @xmath22 values in three - pixel size bins are calculated .
the results from the power - law fitting are summarized in the top right corner of each plot .
the light gray data points are excluded from the power - law fitting .
1@xmath26 corresponds to 0.09 pc at the distance of 300 pc .
( left top ) b5 .
( right top ) ic348 .
( left bottom ) b1e.,title="fig : " ] @xmath22 radial profiles .
the data points with snr of @xmath22 > 1 are selected from the rectangular boxes ( see figure [ f : rh2 ] ) and the mean @xmath22 values in three - pixel size bins are calculated .
the results from the power - law fitting are summarized in the top right corner of each plot .
the light gray data points are excluded from the power - law fitting .
1@xmath26 corresponds to 0.09 pc at the distance of 300 pc .
( left top ) b5 .
( right top ) ic348 .
( left bottom ) b1e.,title="fig : " ] @xmath22 radial profiles .
the data points with snr of @xmath22 > 1 are selected from the rectangular boxes ( see figure [ f : rh2 ] ) and the mean @xmath22 values in three - pixel size bins are calculated .
the results from the power - law fitting are summarized in the top right corner of each plot .
the light gray data points are excluded from the power - law fitting .
1@xmath26 corresponds to 0.09 pc at the distance of 300 pc .
( left top ) b5 .
( right top ) ic348 .
( left bottom ) b1e.,title="fig : " ] @xmath2 image .
the contour for snr of @xmath2 = 3 is overlaid in gray .
the contour for snr of cfa @xmath39 = 3 is also overlaid in black .
the angular resolution of the @xmath2 and cfa @xmath39 images is 4.3@xmath26 and 8.4@xmath26 , respectively .
the @xmath2 image is divided into four regions and each region has either a dark or a star - forming region . ]
@xmath2 ( gray ) and cfa @xmath39 ( light gray ) normalized radial profiles . for each radial profile ,
the data points with snr > 1 are selected from the rectangular boxes ( see figure [ f : rh2 ] ) and the mean @xmath2 and @xmath39 values in two - pixel size bins are calculated .
( left top ) b5 .
( right top ) ic348 .
( left bottom ) b1e.,title="fig : " ] @xmath2 ( gray ) and cfa @xmath39 ( light gray ) normalized radial profiles . for each radial profile ,
the data points with snr > 1 are selected from the rectangular boxes ( see figure [ f : rh2 ] ) and the mean @xmath2 and @xmath39 values in two - pixel size bins are calculated .
( left top ) b5 .
( right top ) ic348 .
( left bottom ) b1e.,title="fig : " ] @xmath2 ( gray ) and cfa @xmath39 ( light gray ) normalized radial profiles . for each radial profile ,
the data points with snr > 1 are selected from the rectangular boxes ( see figure [ f : rh2 ] ) and the mean @xmath2 and @xmath39 values in two - pixel size bins are calculated .
( left top ) b5 .
( right top ) ic348 .
( left bottom ) b1e.,title="fig : " ]
|
to investigate the fundamental principles of h@xmath0 formation in a giant molecular cloud , we derive the hi and h@xmath0 surface density ( @xmath1 and @xmath2 ) images of the perseus molecular cloud on sub - pc scales ( @xmath30.4 pc ) .
we use the far - infrared data from the improved reprocessing of the _ iras _ survey and the @xmath4-band extinction image provided by the complete survey to estimate the dust column density image of perseus . in combination with the hi data from the galactic arecibo l - band feed array hi survey and an estimate of the local dust - to - gas ratio
, we then derive the @xmath2 distribution across perseus .
we find a relatively uniform @xmath5 68 m@xmath6 pc@xmath7 for both dark and star - forming regions , suggesting a minimum hi surface density required to shield h@xmath0 against photodissociation . as a result
, a remarkably tight and consistent relation is found between @xmath8 and @xmath9 .
the transition between the hi- and h@xmath0-dominated regions occurs at @xmath10 @xmath3 ( 814 ) @xmath11 10@xmath12 cm@xmath7 .
our findings are consistent with predictions for h@xmath0 formation in equilibrium , suggesting that turbulence may not be of primary importance for h@xmath0 formation . however , the importance of warm neutral medium for h@xmath0 shielding , internal radiation field , and the timescale of h@xmath0 formation still remain as open questions .
we also compare h@xmath0 and co distributions and estimate the fraction of `` co - dark '' gas , @xmath14 @xmath3 0.3 .
while significant spatial variations of @xmath14 are found , we do not find a clear correlation with the mean @xmath4-band extinction .
|
the boundary between classical planets ( e.g. , jupiter ) and stars has been blurred by the discovery of brown dwarfs , objects which presumably form like stars but lack sufficient mass to sustain core hydrogen fusion @xcite .
known brown dwarfs exhibit a number of features similar to those of giant planets , including size = 7.15@xmath810@xmath9 cm and m@xmath1 = 1.90@xmath810@xmath10 g = 0.00095 m@xmath11 throughout this article @xcite . ]
( r @xmath5 r@xmath1 ) , atmospheric composition ( e.g. , ch@xmath12 , condensate clouds ) , and possibly mass ( of order 10 m@xmath1 ) ; but are typically found in stellar environments as isolated field and cluster objects and as wide companions to main sequence stars .
while brown dwarfs appear to be rare as close companions to stars @xcite , a regime occupied by solar and extrasolar planets , the possibility of planetary ejection @xcite or brown dwarf capture @xcite , and the identification of low mass ratio binaries with substellar secondaries ( e.g. , hr 7329ab ; lowrance et al .
2000 ) , makes it impossible to unambiguously deduce the origin of a particular substellar object .
this situation has led to considerable debate over the definition of a planet and its distinction from a low - mass brown dwarf @xcite . recently , ( * ? ? ? * hereafter z02 ) have identified a brown dwarf in the direction of @xmath2 orionis particularly relevant to this debate .
the object , s ori 053810.1 - 023626 ( hereafter s ori 70 ) , is classified as a t dwarf based on the presence of ch@xmath12 absorption in its near - infrared spectrum @xcite , and therefore has an effective temperature ( t@xmath13 ) less than @xmath512001400 k @xcite . z02
identify this object as a candidate member of the young ( age @xmath5 1 - 8 myr ) @xmath2 orionis cluster , and deduce an estimated mass of only 3@xmath0 m@xmath1 , potentially the least massive brown dwarf known . both z02 and ( * ? ? ?
* hereafter mz03 ) claim that cluster membership for s ori 70 is verified by ( 1 ) comparative near - infrared spectroscopy with the presumably older ( higher surface gravity ) field t dwarf 2mass j05591914@xmath31404488 @xcite ; ( 2 ) spectral fits to theoretical models from @xcite that yield a low surface gravity consistent with a young , low mass brown dwarf ; and ( 3 ) low foreground contamination by field t dwarfs in the search area imaged by z02 . in this article
, we address flaws in each of these arguments .
specifically , in @xmath1424 we show that ( 1 ) comparison with the spectrum of 2mass 0559@xmath31404 is invalidated by the fact that mz03 observed the wrong comparison star , and in fact the spectrum of s ori 70 is an excellent match to those of field t6t7 dwarfs ; ( 2 ) @xcite spectral model fits tend to be skewed toward lower surface gravities , and thus lower ages and masses , for late - type t dwarfs ; and ( 3 ) the identification of one foreground t dwarf in the z02 field is statistically consistent with the expected contamination .
this analysis indicates that s ori 70 may simply be a foreground field t dwarf with a much older age and higher mass than reported by z02 and mz03 .
we summarize our results in @xmath155 .
young , low - mass brown dwarfs have much lower surface gravities than their evolved field counterparts ; @xmath6(cgs ) @xmath5 34 is typical for a 110 myr , 150 m@xmath1 object , versus 4.55.5 for a @xmath51 gyr field brown dwarf with similar t@xmath13 @xcite . as the abundances and band strengths of molecular species that dominate the photospheric opacity of cool brown dwarfs ( e.g. , h@xmath16o , ch@xmath12 , collision - induced h@xmath16 absorption ) are pressure - sensitive , the photospheric gas pressure , and hence surface gravity , can substantially influence the emergent spectral energy distribution .
the @xmath17-band spectrum of s ori 70 presented in mz03 exhibits a distinct triangular - shaped spectral morphology which these authors argue is due to its low surface gravity .
they draw an analogy with the peaked spectral morphologies of low - gravity l dwarfs in the trapezium cluster @xcite , although the latter data were obtained at @xmath18- and @xmath19-bands and not at @xmath17 . the low surface gravity argument of mz03
is supported by a comparison with the spectrum of the presumably older , and hence higher surface gravity , field t dwarf 2mass 0559@xmath31404 , which appears in their figure 1 to have a very different @xmath17-band spectral morphology .
however , their reported spectrum is not that of 2mass 0559@xmath31404 , but rather an adjacent background source .
figure 1 compares the reported mz03 spectrum of 2mass 0559@xmath31404 to data obtained as part of the nirspec brown dwarf spectroscopic survey ( * ? ? ?
* hereafter bdss ) ; and to a re - extraction of the raw data from mz03 using the redspec package ( figure 1 ) .
the re - reduced spectrum of the comparison object observed by mz03 exhibits none of the hallmark features of mid- and late - type t dwarfs .
thus , the differences noted by mz03 between s ori 70 and their field comparison star have nothing to do with surface gravity effects . in fact , the triangular shape of the @xmath17-band spectrum of s ori 70 is common amongst mid- and late - type t dwarfs , the result of h@xmath16o and ch@xmath12 absorption wings on either side of the @xmath17-band peak @xcite .
figure 2 compares the mz03 spectrum of s ori 70 to bdss data for the t7 field brown dwarf 2mass j15530228 + 1532369 @xcite .
this object is presumably much older than the @xmath2010 myr @xmath2 orionis cluster given the similarity of its 12.5 @xmath4 spectrum to other t6t7 field dwarfs @xcite , which have space motions consistent with 15 gyr disk dwarfs @xcite , and the absence of any star forming regions in its vicinity .
the match between the s ori 70 and 2mass 1553 + 1532 data is excellent , and similar good agreement is found with other t6 and t7 bdss spectra .
hence , the @xmath17-band spectrum of s ori 70 is consistent with that of an old , high - gravity , late - type field t dwarf .
both z02 and mz03 determine the surface gravity and t@xmath13 for s ori 70 by fitting their spectra to the most recent theoretical models from @xcite . in the absence of dynamical mass , radius , and bolometric
flux measurements , the use of spectral models is required to derive these physical parameters , as the complexity of the spectra prohibit classical techniques such as curve of growth @xcite . however , despite substantial improvements over the past decade , these models do not yield reliable fits to the observed data , particularly when disentangling t@xmath13 and gravity @xcite .
these limitations are largely due to incomplete or inaccurate molecular line lists and the uncertain treatment of photospheric condensates @xcite .
for instance , @xcite find t@xmath13 discrepancies as high as 400 k between spectral model fits and luminosity - determined temperatures for the latest - type field l dwarfs . to assess whether similar discrepancies occur in the t dwarf regime , we performed minimum @xmath21 fits of bdss @xmath17-band data for seven t6t7 field dwarfs using the same @xcite cond models as those used by mz03 .
we sampled a grid of models spanning @xmath22 and 600 @xmath20 t@xmath13 @xmath20 1400 k in intervals of 0.5 dex and 100 k , respectively .
both empirical and model spectra were gaussian smoothed to the instrumental resolution ( r @xmath5 2000 ) and interpolated onto identical wavelength scales .
as parallax measurements are unavailable for most of these objects and s ori 70 , we performed the fits by normalizing the empirical and theoretical spectra at the 1.27 @xmath4 spectral peak .
the @xmath21 deviation between the spectra was computed over the range 1.161.34 @xmath4 , minimizing over relative scalings of 0.81.2 .
this procedure is similar to that employed by mz03 for s ori 70 .
table 1 lists the best - fit model parameters for the spectral data , along with associated mass and age estimates from the @xcite evolutionary models .
given that the objects examined in these fits are all field dwarfs , it is readily apparent that the derived parameters tend to be skewed toward lower gravities , and hence younger ages and lower masses , particularly for the t7 dwarfs .
the most deviant case is that of gliese 570d , a widely - separated t8 brown dwarf companion to the nearby gliese 570 triple star system @xcite .
assuming coevality , gliese 570d has a well - defined age of 25 gyr @xcite . on the contrary , the spectral model fits predict an age @xmath20 1 myr and mass @xmath20 1 m@xmath1 .
figure 3 compares the minimum @xmath21 fit model for this object to one based on the more widely - adopted parameters @xmath23 and t@xmath13 = 800 k @xcite .
the lower gravity model matches the overall j - band morphology better than the higher - gravity model , but predicts spiked features around 1.255 @xmath4 .
these features do not coincide with ( presumably noise ) spikes seen at 1.265 @xmath4 in the s ori 70 spectrum , as is evident in figure 3 of mz03 .
the inferior fit of the high gravity model is generally confined to the spectral peak , and may be the result of deep condensate opacity in this relatively transparent spectral region @xcite . regardless of the reason , it is clear that the best fit spectral models can yield skewed gravities for late - type field t dwarfs , resulting in underestimated ages and masses .
hence , the spectral fit parameters for s ori 70 , which itself appears to be a late - type t dwarf , must be considered with skepticism .
there are also inconsistencies between the @xmath17-band spectral model fits for s ori 70 from mz03 and the @xmath24 spectral model fits from z02 .
the latter yield a substantially lower t@xmath13 ( 800 versus 1100 k ) and a somewhat higher surface gravity ( 4.0 versus 3.5 dex ) , which is in fact consistent with a 50 myr , not 8 myr , brown dwarf @xcite . while the differences between the parameter fits are within the reported uncertainties , they yield significantly incompatible spectral morphologies .
figure 4 demonstrates this , comparing the @xmath24 spectrum from z02 to a smoothed @xcite model using the best fit parameters from mz03 .
it is clear that this model grossly underestimates the strength of the 1.6 @xmath4 ch@xmath12 band and somewhat overestimates the flux at @xmath19-band . on the other hand , empirical spectral data for the field t6.5 2mass j10475385 + 2124234 @xcite , obtained with identical instrument settings , does fairly well at matching the @xmath18-band region , although it somewhat underestimates the flux at @xmath19-band .
the latter disagreement , possibly due to differences in collision - induced h@xmath16 opacity , may be indicative of a somewhat lower surface gravity for s ori 70 relative to 2mass 1047 + 2124 @xcite , although signal - to - noise may also be an issue .
nevertheless , the fact that empirical data for field t6t7 dwarfs reproduce the 1.11.3 @xmath4 and 1.52.4 @xmath4 spectra for s ori 70 more consistently than the theoretical models indicates both the unreliability of physical parameters derived from the model fits and the likelihood that s ori 70 is a foreground brown dwarf .
a third argument used by z02 to establish cluster membership for s ori 70 is the low foreground contamination by field t dwarfs ( 0.080.3 objects ) expected in their 55.4 arcmin@xmath25 search area .
we have computed an independent estimate of the foreground contamination using substellar luminosity function simulations from burgasser ( 2001 ) ; absolute photometry from @xcite and @xcite ; and an assumed limiting magnitude of @xmath26 ( z02 ) . for a substellar mass function @xmath27 and @xmath28 , we find 0.10.2 t dwarfs with 700 @xmath29 t@xmath13 @xmath29 1100 k ( typical for late - type t dwarfs ; burgasser 2001 ) expected in the z02 field , consistent with their estimates and certainly less than unity . however , the relevant quantity is the likelihood of detecting one foreground object in the imaged area given the expected contamination rate .
this confidence limit , cl , can be derived from poisson statistics : @xmath30 @xcite for @xmath31 source detected and an expected contamination of @xmath32 . for @xmath33 , eqn .
1 yields @xmath34 , equivalent to 0.61.4 @xmath2 on a gaussian scale .
hence , the presence of one foreground t dwarf in the z02 field is a 1@xmath2 event , and can not be ruled out statistically .
based on the arguments above , we find compelling evidence that s ori 70 is not a member of the @xmath2 orionis cluster but rather a foreground field brown dwarf .
this is not an unexpected result given the predicted @xmath530% contamination rate amongst l dwarfs in the @xmath2 orionis sample @xcite .
indeed , quite a few @xmath2 orionis brown dwarf candidates have been ruled out as foreground objects in follow - up observations @xcite , and @xcite find the fraction of spectroscopically - verified candidates in @xmath2 orionis decreases toward fainter magnitudes .
it should come as no surprise that the faintest candidate to date , s ori 70 , may be a contaminant field source .
uncertainties in the spectral models and difficulty in obtaining high signal - to - noise spectra for this faint t dwarf ( @xmath35 ; z02 ) imply that unambiguous verification of cluster membership for s ori 70 will require measurement of its parallax and proper motion to establish spatial and kinematic association .
such observations are hindered by the distance of this object even if it is a foreground brown dwarf .
z02 measure an upper limit proper motion of @xmath36 yr@xmath37 for s ori 70 , which only restricts its distance to @xmath38 40 pc , roughly half the spectrophotometric distance if it is a field t6t7 dwarf ( @xmath5 75100 pc ; tinney , burgasser , & kirkpatrick 2003 ) .
we conclude that s ori 70 has not been rigorously proven to be a member of the young @xmath2 orionis cluster .
rather , observations obtained thus far are consistent with this object being a massive ( m @xmath5 3060 m@xmath1 assuming an age of 15 gyr ) foreground field brown dwarf .
this interpretation should be seriously considered before assuming the existence of a `` cluster planet '' population in the @xmath2 orionis cluster .
the authors would like to thank e. l. martn and m. r. zapatero osorio for candid discussions regarding their results and for providing their reduced and raw data for s ori 70 and 2mass 0559@xmath31404 ; f. allard for providing access to theoretical spectral models ; and t. ayres and l. hillenbrand for very useful discussions .
we also thank our anonymous referee for her / his very prompt review .
a. j. b. acknowledges support provided by nasa through hubble fellowship grant hst - hf-01137.01 awarded by the space telescope science institute , which is operated by the association of universities for research in astronomy , incorporated , under nasa contract nas5 - 26555 . this research has made use of the simbad database , operated at cds , strasbourg , france .
kenyon , m. j. , jeffries , r. d. , & naylor , t. 2002 , in the future of cool star astrophsics , proceedings of the 12th cambridge workshop on cool stars , stellar systems , and the sun , eds .
a. brown , g. m. harper , & t. r. ayres , http://origins.colorado.edu/cs12/index.html , p. 645
leggett , s. k. , golimowski , d. a. , fan , x. , geballe , t. r. , & knapp , g. r. 2003 , in proceedings of the 12th cambridge workshop on cool stars , stellar systems , and the sun , ed .
a. brown , t. r. ayres , & g. m. harper ( boulder : univ .
colorado press ) , p. 120
llcccccccc 2mass j23565477 + 1553111 & t6 & 1100 & 5.0 & 0.8 & 0.03 & & 1000 & 5.0 & 0.030.06 + sdssp j162414.37 + 002915.6 & t6 & 900 & 5.0 & 2 & 0.03 & & 1000 & 5.0 & 0.030.06 + 2mass j12373919 + 6526148 & t6.5 & 900 & 5.0 & 2 & 0.03 & & 950 & 5.0 & 0.030.05 + 2mass j15530228 + 1532369 & t7 & 1200 & 3.5 & 0.004 & 0.003 & & 900 & 5.0 & 0.030.05 + 2mass j07271824 + 1710012 & t7 & 1200 & 3.5 & 0.004 & 0.003 & & 900 & 5.0 & 0.030.05 + gliese 570d & t8 & 1100 & 3.0 & @xmath20 0.001 & @xmath20 0.001 & & 784824 & 5.05.3 & 0.030.05 + 2mass j04151954@xmath30935066 & t8 & 700 & 4.5 & 0.7 & 0.014 & & 800 & 4.55.0 & 0.020.04 +
|
we examine recent claims that the t - type brown dwarf s ori 053810.1 - 023626 ( s ori 70 ) is a spectroscopically verified low mass ( 3@xmath0 m@xmath1 ) member of the 18 myr @xmath2 orionis cluster .
comparative arguments by martn & zapatero osorio asserting that s ori 70 exhibits low surface gravity spectral features indicative of youth and low mass are invalidated by the fact that their comparison object was not the field t dwarf 2mass 0559@xmath31404 but rather a nearby background star .
instead , we find that the 12.5 @xmath4 spectra of s ori 70 are well - matched to older ( age @xmath5 few gyr ) field t6t7 dwarfs .
moreover , we find that spectral model fits to late - type field t dwarf spectra tend to yield low surface gravities ( @xmath6 = 3.03.5 ) , and thus young ages ( @xmath7 5 myr ) and low masses ( @xmath7 3 m@xmath1 ) , inconsistent with expected and/or empirical values .
finally , we show that the identification of one t dwarf in the field imaged by zapatero osorio et al .
is statistically consistent with the expected foreground contamination .
based on the re - examined evidence , we conclude that s ori 70 may simply be an old , massive ( 3060 m@xmath1 ) field brown dwarf lying in the foreground of the @xmath2 orionis cluster .
this interpretation should be considered before presuming the existence of so - called `` cluster planets . ''
|
it has recently become accepted that quasar activity and black hole growth are an integral part of galaxy evolution , however a detailed understanding of what triggers quasar activity and how they are fueled still eludes us . the leading contender for the identity of luminous
, high redshift qsos is that they are black holes fed by by major mergers of gas - rich galaxies @xcite .
recent incarnations of such models @xcite provide a good description of many observed properties of the qso population .
the situation is particularly interesting at high redshift , where the population of supermassive black holes that powers the qsos is growing rapidly ( see e.g. * ? ? ? * for a recent review ) . to further understand this important phase of black hole and galaxy evolution we would like to build a model in which qso activity is tied to the evolving cosmic web of dark matter halos .
the relationship between qsos and dark matter halos , their environments and duty cycles , can be constrained via observations of their space density and large - scale clustering @xcite .
these constraints become particularly sensitive if the qsos inhabit the rarest , most massive halos for which the spatial clustering depends strongly on halo mass @xcite . at redshifts @xmath4 the advent of large optical surveys for qsos
has led to firm constraints on the clustering as a function of luminosity and redshift @xcite . with the sloan digital sky survey ( sdss )
we are now able to measure the clustering of qsos well even at @xmath5 ( * ? ? ?
* see also earlier work by @xcite ) .
interestingly , the correlation length of the qso population increases rapidly with redshift , from @xmath6 at @xmath7 to @xmath8 at @xmath9 .
@xcite demonstrate that they can fit the observed clustering and space density of @xmath0 qsos with the model of @xcite provided the @xmath0 qsos are relatively long lived ( @xmath10myr ) and inhabit halos more massive than about @xmath11 .
for this calculation @xcite assume that there is a monotonic relationship between instantaneous qso luminosity and halo mass , with no scatter .
the actual relation between instantaneous qso luminosity and host halo mass is expected to include some scatter .
scatter is expected in several of the relationships linking the qso luminosity and the host halo mass : the relationship between the host halo mass and galaxy bulge , in the relationship between galaxy bulge and black hole mass , in the relationship between black hole mass and peak luminosity and in the relationship between peak and instantaneous qso luminosity . here
we demonstrate that the very high correlation length measured by @xcite , when combined with the rapid increase in bias for the most massive halos , strongly constrains the scatter between instantaneous qso luminosity and halo mass .
the essential idea is that any scatter increases the contribution from lower mass ( and less highly biased ) halos , so that a measurement of large clustering amplitude limits the contribution from lower mass objects .
this constraint is as strong as direct , observational constraints at lower redshift on the amount of scatter in the various relationships mentioned above .
we shall work throughout in the @xmath12cdm framework and adopt the following cosmological parameters : @xmath13 , @xmath14 , @xmath15 and @xmath16 .
where appropriate we shall comment on how our results depend on these particular choices .
the next section outlines our formalism and applies it to the data of @xcite , while [ sec : discussion ] interprets our results and discusses future observations .
we will focus our attention on the space density and large - scale bias of qsos , as these are the easiest to interpret , observationally measurable properties of the population .
if the mean number of qsos in a halo of mass @xmath17 is @xmath18 then their number density and large - scale bias are @xmath19 where @xmath20 is the ( comoving ) number density of halos per mass interval and @xmath21 is the bias associated with halos of that mass . for @xmath20 we use the fitting function from @xcite .
the large - scale bias is slightly more problematic , as our results depend upon massive halos being biased and different fits to @xmath21 from simulations have appeared in the literature . @xcite
derived the bias appropriate to their mass function fit using the peak background split .
a slightly higher bias at fixed mass comes from assuming ellipsoidal collapse @xcite .
the largest bias for halos of a given mass is given by the bias of the @xcite mass function , as computed by @xcite , which is very similar to the fit of @xcite for the masses of interest .
these two forms give biases 30% larger , in the mass and redshift range of interest , than the lowest fit ( of * ? ? ?
the fit of @xcite lies between that of @xcite and @xcite . for mass - thresholded samples of dark matter halos in a large n - body simulation particles in a @xmath22gpc box using the _ treepm _ code described in @xcite .
see @xcite for a recent comparison with other n - body codes .
] , we found the @xcite and @xcite forms bracketed the halo correlation function so this should represent the level of current uncertainty . as it provides the most conservative estimate of the maximum allowed scatter , and the numerical data lay closer to the form of @xcite , we adopt this as our fiducial bias . however , we caution that the statistics in the simulation are poor and we also illustrate the effect of choosing the @xcite form below . in several recent simulations at other redshifts and masses
the @xcite form appears to be a worse fit @xcite , but unfortunately there are few direct n - body calibrations of @xmath23 at the number densities and redshifts of interest to us .
one could imagine that for extremely rare peaks , whose collapse is much closer to spherical @xcite , these newer fits actually perform worse than the original formulation ( see e.g. * ? ? ?
* ; * ? ? ?
this challenging numerical problem deserves further investigation . to interpret the large - scale clustering measurements of @xcite
we then must specify @xmath18 .
we wish to choose as simple a model as possible to illustrate the effect of scatter on the @xmath24 relation , so we assume that the probability that a qso is seen with instantaneous luminosity @xmath25 is log - normally distributed can be interpreted as a fractional scatter .
the log - base-10 based @xmath26 would be @xmath27 times smaller . ] around a central value @xmath28 with a width @xmath29 @xmath30 \
d\ln l \qquad .\ ] ] if a given halo hosts a qso , the probability that it is above some limiting luminosity , @xmath31 , is then the integral of this expression from @xmath31 to infinity , which can be expressed as an error function .
if we assume that at most a fraction @xmath32 of halos host active qsos at any epoch and that @xmath33 in the region of the turn - on , the mean number of qsos above @xmath31 in halos of mass @xmath17 is @xmath34 \label{eqn : nofm}\ ] ] where @xmath35 , @xmath36 and we expect @xmath37 .
we shall assume eq .
( [ eqn : nofm ] ) in what follows and try to constrain @xmath26 .
figure [ fig : b_vs_n ] shows @xmath38 vs. @xmath39 for a range of @xmath40 and @xmath26 .
the larger the dispersion the lower the bias at fixed space density , as a larger fraction of the qsos are hosted in lower mass halos .
similarly , a decrease in @xmath32 must be compensated by a decrease in @xmath40 to hold @xmath41 fixed , again leading to a lower @xmath42 . in our assumed cosmology the universe is @xmath43gyr old at @xmath44 .
if we assume @xmath45 ( e.g. * ? ?
? * ) then the e - folding time is @xmath46 , so @xmath47 is not unexpected . while the inferred lifetimes are then weakly in conflict with the upper range of observational constraints ( @xmath48 years ; *
* ) , these constraints largely stem from lower - redshift qsos .
our model is not the most general one describing how qsos inhabit dark matter halos . in particular
, there is no reason in principle why halos could not host more than one qso . to constrain the functional form of @xmath18
would require modeling both the large- and small - scale clustering , and additional assumptions about the statistics of qsos in halos .
this is beyond the scope of this paper , so we merely note that the mass function is very steeply falling in the range of halo masses that are of interest .
our statistics are thus dominated by the lowest mass halos in our sample .
in particular , the number density is approximately linear in @xmath49 .
simply allowing @xmath50 has no impact on @xmath51 , but it would raise @xmath41 and slightly weaken our constraints . however to increase @xmath41 significantly above that for @xmath52 would require almost every halo of mass @xmath40 to host more than one high luminosity qso which would lead to an unacceptably high close pair fraction .
for example , if every halo hosted 2 qsos the volume averaged correlation function within the halo radius would be @xmath53 . given the extreme rarity of these halos , this would be in conflict with observations ( * ? ? ?
* ; * ? ? ?
* see e.g. fig .
17 in @xcite ) . at this point
it is also easy to see how a change in cosmological parameters affects @xmath24 .
for example , if we raise @xmath54 , halos of a fixed mass correspond to less rare peaks .
these halos are therefore slightly less biased and significantly more numerous .
to hold @xmath41 fixed we would need to increase @xmath40 . to hold the bias fixed requires that we increase @xmath40 yet further and therefore to match both the @xmath42 and @xmath41
the duty cycle must increase . for @xmath55 ,
a 10% increase in @xmath54 leads to a 4% decrease in @xmath42 , which is insignificant for our purposes . in general , for a power - law power spectrum with index @xmath56 , the fractional change in number density at fixed bias is @xmath57 times the fractional change in @xmath54 .
small changes in the other cosmological parameters have even smaller effects .
this formalism allows us to use the observations of high-@xmath58 qso clustering from @xcite to constrain @xmath26 and @xmath32 in our model . to do
so we need to estimate the large - scale bias , @xmath42 , from the measurements provided in @xcite and this requires making several assumptions .
first we must convert from the clustering quoted by @xcite to large - scale bias .
there are several ways to do this , and we have chosen to use the amplitude of their fits to @xmath59 power - laws over the range @xmath60mpc . at lower @xmath58
the bias is relatively constant over this range and the slope of the dark matter correlation function is close to @xmath61 , so converting the fit into a measurement of @xmath51 is straightforward .
however at @xmath5 the qsos are hosted by increasingly rare halos , whose bias is becoming more scale dependent . using the mass and halo catalogs from the n - body simulation
described previously we find that the mass correlation function is shallower than @xmath59 below @xmath62 but the halo correlation function is still quite close to @xmath59 down to @xmath63mpc .
we therefore convert from @xmath64 to @xmath51 by matching @xmath65 at @xmath66mpc , where the bias is close to constant .
( this is also the scale where we compared the simulations to the @xcite fitting form to obtain the bias . ) for convenience we use the non - linear power spectrum of @xcite when computing @xmath67 , but we checked that this agrees well with the results of the n - body simulation on the scales of interest and the correction for non - linearity is relatively small .
the clustering measurements of @xcite are averaged over bins in redshift , over which the bias and mass correlations are evolving strongly .
fortuitously , the effects approximately cancel , leading to a slow evolution in @xmath68 .
we estimate the effects of clustering evolution on our constant - time constraints by considering 3 models for the evolution : passive evolution ( which for massive , rare halos corresponds almost to @xmath69constant ) , constant bias and constant halo mass .
each model predicts @xmath70 from which we can compute the average value measured by @xcite as @xmath71 where @xmath72 is the redshift distribution , @xmath73 is the hubble parameter at redshift @xmath58 and @xmath74 is the comoving angular diameter distance to redshift @xmath58 .
using @xmath72 from table 1 of @xcite in the two bins @xmath75 and @xmath76 we find that for the passive evolution model the correlation length inferred from @xmath77 is within 1% of that inferred from @xmath78 and @xmath79 respectively .
for the constant halo mass case @xmath80 increases with @xmath58 , but the inferred @xmath64 is still within 5% of the constant @xmath58 value for both samples .
for the constant bias case , @xmath80 decreases with @xmath58 , but the inferred @xmath64 is within 4% of the constant @xmath58 value for both samples . since the quoted errors on @xmath64 are larger than this , we shall interpret the quoted correlation lengths as measurements at @xmath81 and @xmath44 respectively .
assuming @xmath82 , @xcite quote @xmath83mpc for their `` good '' sample with @xmath75 . using the dark matter correlation function for our cosmology at @xmath81
this corresponds to @xmath84 .
the quoted space density is @xmath85 .
if we use the ` all ' sample both @xmath64 and hence @xmath51 are 14% lower .
we will find that this point is relatively unconstraining for either choice . at @xmath44
the best fit is @xmath86mpc , again using the `` good '' sample , corresponding to @xmath87 , and @xmath88 .
the other clustering fits provided by @xcite are discussed below .
note however , as pointed out by @xcite , this space density is actually an underestimate of the true space density because the @xcite parameterization of the lf underestimates the measured space density from @xmath89 ( see figure 20 of * ? ? ?
* ) . at @xmath9
this underestimate is approximately a factor of two and we therefore use @xmath90 .
the higher space density leads to a smaller predicted @xmath51 , requiring smaller dispersion @xmath26 between luminosity and halo mass to fit the observed clustering ( i.e. this provides a more stringent constraint ) . in fig .
[ fig : b_vs_sig ] we show @xmath38 as a function of log - normal scatter , @xmath26 , at fixed @xmath41 for @xmath52 , 0.3 and 0.1 .
the @xmath91 and @xmath92 confidence ranges in @xmath42 ( above ) are also shown .
as can be seen , in order to get the large @xmath42 seen by @xcite , the scatter at @xmath9 has to be less than 80% and the fraction of halos containing quasars , @xmath32 , must be larger than @xmath93 .
the constraints will change if different fits to the clustering measurements , different fits for @xmath94 , or different number densities are used .
as the constraints are coming from @xmath44 , we focus on this case .
@xcite gives 4 fits to power - law correlation functions , for `` all '' or `` good '' qso s , with fixed power law @xmath95 or varying power law .
the fiducial calculation above was for the `` good '' qsos sample fit to @xmath95 , which has a @xmath96 .
the other combinations give : ( sample , power - law index , @xmath97 , @xmath98)= ( `` all '' , @xmath99 , 0.52 , -15% ) , ( `` good '' , @xmath100 , 0.32 , + 6% ) , ( `` all '' , @xmath101 , 0.50 , -6% )
. the central value in fig .
[ fig : b_vs_sig ] then just moves up and down by the shift in bias ( the error bars do change slightly in width ) .
our example in fig .
[ fig : b_vs_sig ] is one of the best @xmath97 cases but also gives a high bias ( and thus is very constraining ) .
as mentioned previously , our fiducial @xmath94 was chosen to give the least stringent constraint .
the functional forms of @xcite , and @xcite predict lower @xmath94 at fixed @xmath17 and hence tighter constraints on @xmath26 and @xmath32 .
we illustrate the range in figure [ fig : bias_cmp ] , which compares the results using our fiducial bias to that of @xcite . as we discussed before the fits to @xmath21 from n - body simulations , where they are available , tend to lie between those of @xcite and @xcite which differ from each other by @xmath102 in the mass and redshift range of interest ( see also * ? ? ?
* ; * ? ? ?
neither the sheth - tormen or sheth - mo - tormen biases are able to fit the measured bias and number density .
the most unconstraining estimate would be to use the lower amplitude of clustering , e.g. the `` all '' sample , at @xmath44 .
we did not choose this because there are a number of systematics which could lower the measured clustering amplitude and the `` all '' sample has a worse @xmath97 .
but if taken in conjunction with the ( most conservative ) press - schechter bias and the highest allowed number density , the limit can be weakened to allow @xmath103 at 95% confidence for @xmath104 .
it is also possible that qsos may inhabit a special subclass of halos for which the bias is larger than the average for that mass .
for instance there are indications that halo history affects clustering in some cases @xcite . for the very rare and highly biased halos hosting qsos at @xmath9
this effect may be large enough to weaken our constraint .
unfortunately the bias of the relevant objects has not yet been measured in simulations .
it appears that the most luminous , highest redshift qsos have instantaneous luminosities which are well correlated with their host halo masses . the already strong constraint shown here
should be improved by measurements of high - redshift qso clustering with yet larger samples .
new observations will broadly improve the constraint in two ways : through wider area observations to identify a larger number of qsos similar in redshift to the sdss sample , and through deeper observations over the same area .
deeper observations will identify slightly fainter qsos at all redshifts and may provide a sufficient sample to measure the clustering amplitude at even higher redshift than @xmath105 , where we expect the bias to be even larger . going wider in area will decrease the fractional error on @xmath42 for high - redshift qsos .
as these measurements of high - redshift qso bias will remain poisson limited , the fractional error on @xmath42 will scale as @xmath106 for a survey of @xmath107 qsos .
going deeper to measure luminous qsos at yet higher redshifts is also expected to be useful , as the data indicate strong evolution in the clustering amplitude with redshift .
figure [ fig : b_vs_z ] illustrates how @xmath42 increases with redshift for three values of the scatter between qso luminosity and host halo mass assuming @xmath108 . at each redshift @xmath42 was calculated to match the qso space density evolution parametrized by @xcite , although converted to our cosmological parameters and consequently the predicted bias is slightly higher . ] .
the steepest relation between @xmath42 and redshift is for the case of the smallest scatter . if @xmath108 and the scatter is minimal , the relative value of @xmath42 for different values of the dispersion is a good measure of the fractional uncertainty in @xmath42 required to improve on our constraint .
these ratios demonstrate that the most effective constraint from clustering arises from the lowest redshift at which @xmath108 and minimal scatter is a reasonable approximation .
for example , the ratio of @xmath109 to @xmath110 increases by less than 10% from @xmath44 to @xmath111 , while the @xmath107 above the sdss flux limit drops by approximately an order of magnitude . therefore while the clustering amplitude increases , the decline in @xmath107 for a flux - limited , high - redshift survey more than offsets this gain .
by contrast , color selection of just very high - redshift ( e.g. @xmath112 ) qsos could produce a substantial improvement for a given total number of spectra .
the immediate prospect for improvement in this constraint is completion of the sdss observations beyond the 4000 square degrees employed by @xcite .
these observations will simply increase @xmath107 and the error bars should scale as @xmath113 . on a somewhat longer timescale , the proposed baryon oscillation spectroscopic survey ( boss ) should approximately double the number of @xmath76 qsos , although it will mostly achieve this by going deeper .
provided @xmath108 and small scatter are still reasonable approximations to this slightly fainter population ( that will still be well above the break in the qso lf ) , boss factor of two improvement in sample size should decrease the error bars by approximately a factor of two and produce a powerful improvement in the constraint .
for example , if the measured bias stayed fixed and the errorbars decreased by a factor of two , the upper limit on the observed scatter would improve to less than 0.5 ( i.e. 0.2 dex ) for @xmath114 .
there are also upcoming photometric surveys that go even wider in area and deeper : pan - starrs , the dark energy survey ( des ) , and the large synoptic survey telescope ( lsst ) .
while these surveys do not include a dedicated plan for spectroscopic observations of qsos , they will provide the necessary candidate database of these extremely rare objects for spectroscopy . on the theoretical side , the calculations presented here could also be improved .
better modeling could allow us to use a wider range of scales .
as large - scale , high - resolution simulations become increasingly feasible , it will become possible to precisely measure both the bias of all halos and the bias as a function of halo history .
the scatter between these various relationships can also be measured in self - consistent numerical simulations of merger - driven black hole growth .
@xcite presented a recent analysis of the observed and simulated scatter between black hole mass and various host ( bulge ) galaxy properties ( @xmath26 , host galaxy mass , effective radius ) at lower redshifts than we are considering here .
they find that the intrinsic scatter is generally lower than the observed ( upper limit ) and as small as @xmath115dex for combinations that produce a fundamental plane for black holes .
however , they also note that the strong correlation of black hole mass with bulge properties implies that the connection to halo mass is only indirect and dominated by the evolution in the typical gas fractions .
measurement of the evolution of these quantities in full cosmological simulations ( e.g. * ? ? ?
* ) could provide measurements of the expected scatter at the very high redshifts where qsos are observed to be so strongly biased .
regardless of these future prospects , the strength of the present constraint on scatter in a monotonic relation between qso luminosity and halo mass is already surprising .
as noted in the introduction , we expect scatter to be present because we expect scatter in at least four relationships that combine to determine the instantaneous luminosity of a qso in a halo of a given mass : the relationship between instantaneous luminosity and peak luminosity ; the relationship between peak luminosity and black hole mass ; the relationship between black hole mass and bulge properties ; and the relationship between bulge and halo mass .
these expectations arise from the observed scatter in these individual relations at lower redshifts intriguingly , in all cases the observed scatter only sets an upper limit on the intrinsic scatter . for the relation between black hole mass and bulge velocity dispersion known as the @xmath116 relation @xcite , @xcite showed that the intrinsic dispersion is no larger than 0.250.3 dex ( or @xmath117 in our natural log notation ) .
@xcite explored the relation between @xmath118 by using rotation curve data and argued that the dispersion between black hole and halo mass could be even less than the @xmath3 dispersion , although she notes that the uncertainty in the transformation between circular velocity and halo mass is not well characterized .
finally , @xcite measured the distribution in eddington ratio , @xmath119 , for a sample of @xmath120 qsos and find the distribution is well described as log - normal with a peak at 0.25 and a dispersion of 0.3 dex .
as this observed dispersion must account for the uncertainty in the mass estimator and bolometric correction in addition to @xmath119 , this measurement also sets an upper limit to the intrinsic dispersion in @xmath119 .
in addition to the small scatter in @xmath121 , the data also suggest that the duty cycle is approximately unity
. this may be less surprising , given their extremely high luminosities and masses of their supermassive black holes .
if qsos are limited to accreting at no more than the eddington rate , the luminosities of @xmath122 qsos imply that their central , supermassive black holes already exceeded @xmath123 when the universe was less than @xmath124gyr old @xcite .
subsequent near - infrared spectroscopy provides further evidence that their central black holes are indeed this massive @xcite . to create such massive black holes at early times
requires @xmath125 , modulo the uncertainty in the initial seed mass .
together , these strong constraints on the duty cycle and scatter in @xmath126 for high - redshift qsos provide important new information on how supermassive black holes and their host halos grew at early times .
we thank yue shen and cris porciani for helpful comments on an earlier draft of this paper .
the simulations used in this paper were analyzed at the national energy research scientific computing center .
we thank ccapp at the ohio state university and the participants in the quasar mini - workshop for a stimulating meeting .
jdc and mw also are grateful to ccapp for hosting them for an extended visit .
mw is supported by nasa .
|
recent measurements of high - redshift qso clustering from the sloan digital sky survey indicate that qsos at @xmath0 have a bias @xmath1 .
we find that this extremely high clustering amplitude , combined with the corresponding space density , constrains the dispersion in the @xmath2 relation to be less than 50% at 99% confidence for the most conservative case of a 100% duty cycle .
this upper limit to the intrinsic dispersion provides as strong a constraint as current upper limits to the intrinsic dispersion in the local @xmath3 relation and the ratio of bolometric to eddington luminosity of luminous qsos .
-1 cm dark matter large - scale structure of universe quasars : general
|
a part of this work was done at the department of physics engineering , mie university , tsu , japan .
the authors are grateful to professor y. abe and assistant professor m. matsunaga for their hospitality .
one of the authors ( t. m. ) is supported in part by a grant - in - aid for scientific research from the ministry of education , science , sports and culture , japan ( no . 10640256 ) .
99 e. witten , nucl .
b188 * ( 1981 ) , 513 ; * b202 * ( 1982 ) , 253 .
+ i. affleck , m. dine and n. seiberg , nucl . phys .
* b256 * ( 1985 ) , 557 .
+ k. izawa and t. yanagida , prog .
* 95 * ( 1996 ) , 829 .
+ k. intriligator and s. thomas , nucl . phys .
* b473 * ( 1996 ) , 121 .
m. dine and a. e. nelson , phys . rev . * d48 * ( 1993 ) , 1277 .
+ m. dine , a. e. nelson and y. shirman , phys .
rev . * d51 * ( 1995 ) , 1362 .
+ m. dine , a. e. nelson , y. nir and y. shirman , phys . rev .
* d53 * ( 1996 ) , 2658 .
+ s. p. martin , phys . rev .
* d55 * ( 1997 ) , 3177 .
+ c. kolda , nucl .
* 62 * ( 1998 ) , 266 .
+ n. haba , n. maru and t. matsuoka , nucl . phys .
* b497 * ( 1997 ) , 31 ; phys . rev .
* d56 * ( 1997 ) , 4207 .
+ e. poppitz and s. p. trivedi , phys
* b401 * ( 1997 ) , 38 .
h. p. nilles , m. olchowski and m. yamaguchi , phys . lett .
* b415 * ( 1997 ) , 24 ; nucl . phys .
* b530 * ( 1998 ) , 43 . + a. font , l. e. ib@xmath112ez , d. l@xmath113st and f. quevedo , phys . lett . * b245 * ( 1990 ) , 401 .
v. s. kaplunovsky and j. louis , phys . lett .
* b306 * ( 1993 ) , 269 .
+ s. ferrara , c. kounnas and f. zwirner , nucl .
b429 * ( 1994 ) , 589 .
+ a. brignole , l. e. ib@xmath110ez and c. mu@xmath115 oz , nucl . phys . *
b422 * ( 1994),125 .
+ c. mu@xmath115 oz , nucl .
suppl , * 49 * ( 1996 ) , 96 ; nucl . phys .
* b267 * ( 1986 ) , 365 .
l. e. ib@xmath110ez and h. p. nilles , phys
b169 * ( 1986 ) , 354 .
+ k. choi and j. e. kim , phys
b165 * ( 1985 ) , 71 .
+ l. j. dixon , v. s. kaplunovsky and j. louis , nucl . phys . *
b355 * ( 1991 ) , 649 .
+ h. p. nilles and s. stieberger , nucl .
b499 * ( 1997 ) , 3 .
|
in the context of the weakly coupled heterotic string , we propose a new model of mediating supersymmetry breaking . the breakdown of supersymmetry in the hidden sector is transmitted to anti - generation fields via gravitational interactions .
subsequent transmission of the breaking to the mssm sector occurs via gauge interactions .
it is shown that the mass spectra of superparticles are phenomenologically viable .
kgku-00 - 01 + hep - ph/0001323 * mediation of supersymmetry breaking + via anti - generation fields * masato ito@xmath0 and takeo matsuoka@xmath1 _ @xmath0department of physics , nagoya university , nagoya , japan 464 - 8602 + @xmath1kogakkan university , nabari , japan 518 - 0498 _ the dynamical mechanism of supersymmetry(susy ) breaking and its mediation to our visible world form the major unsolved problems of particle physics . as for the former ,
many models have been proposed@xcite . for phenomenological implications
it is of more importance to clarify how susy breaking is communicated to the visible sector .
two ways of transmitting susy breaking to the visible sector have been proposed .
the first is that gravitational interactions in @xmath2 supergravity theory play this role@xcite .
the second is that gauge interactions play the role of the messenger@xcite . in the framework of superstring theory
it seems that the gravity - mediated susy breaking scenario is plausible . in the context of the weakly coupled heterotic @xmath3 superstring ,
@xmath4 is considered to be the hidden sector gauge symmetry . if supersymmetric gauge dynamics in the hidden sector , such as gaugino condensation , cause susy breaking to occur , the effects of the susy breaking will be transmitted to the observable sector through gravitational interactions . in these types of simple models@xcite
, the @xmath5-components of moduli fields other than the dilaton are expected to develop their vacuum expectation values ( vevs ) .
the gauginos acquire their soft susy breaking masses at one loop , but not at the tree level . as a consequence ,
the gaugino masses turn out to be too small compared to the soft scalar masses .
namely , we have the relations @xmath6 where @xmath7 , @xmath8 and @xmath9 are the gaugino mass , the soft scalar mass and the gravitino mass , respectively .
this mass spectrum of superparticles is phenomenologically unacceptable .
although it has been pointed out that this problem can be solved in strongly coupled heterotic @xmath3 superstring theory ( m - theory)@xcite@xcite , there remains another problem regarding non - universality of the soft breaking mass parameters . since in string theory the moduli fields are generally charged under the flavor symmetry , the soft scalar masses become flavor - dependent . as a result of these non - universal properties for the scalar masses ,
we are confronted with the so - called fcnc problem . with regard to this point
, the gauge - mediated susy breaking scenario has the advantage of producing the universal soft scalar masses and then ensuring sufficient suppression of the fcnc . in this paper
we propose a new model based on the weakly coupled heterotic string on the calabi - yau compactification , whose superparticle mass spectrum is phenomenologically viable and which is free from the fcnc problem .
the breakdown of susy coming from the hidden sector gaugino condensation begins by being transmitted to anti - generation fields in the observable sector via gravitational interactions .
this is due to the assumption that only vevs of the khler class moduli fields @xmath10 acquire nonvanishing @xmath5-components .
subsequent transmission of susy breaking to the low - energy mssm sector occurs via gauge interactions .
then the gaugino masses are of the same order as the soft scalar masses .
in addition , due to the flavor - blind nature of gauge interactions , we obtain the universal soft scalar masses .
the four - dimensional effective theory in the observable sector from the weakly coupled calabi - yau string is characterized by @xmath2 susy , the @xmath11 gauge group , and massless matter fields which belong to @xmath12 and @xmath13 representations in @xmath11 .
the massless chiral superfields apart from @xmath11 singlets consist of @xmath14 where @xmath15 denotes the family number at low energies .
@xmath16 sets of vector - like multiplets are included in the massless sector .
the numbers @xmath17 and @xmath16 represent the generation number and the anti - generation number , respectively .
these numbers coincide with the hodge numbers @xmath18 and @xmath19 for the compactified manifold , respectively .
we will assume @xmath20 and @xmath21 for the sake of simplicity .
the particles beyond the mssm are contained in @xmath22 .
namely , in @xmath22 we have color - triplet higgses @xmath23 and @xmath24 and a singlet @xmath25 as well as the quark superfield @xmath26 lepton superfield @xmath27 and
higgs doublets @xmath28 .
moreover , there appear a dilaton field @xmath29 , @xmath19 khler class moduli fields @xmath30 , and @xmath18 complex structure moduli fields @xmath31 .
the vev of a dilaton field @xmath29 determines the gauge coupling constant , and the vevs of the moduli fields @xmath31 and @xmath30 parametrize the size and shapes of the compactified manifold . in the observable sector from the calabi - yau string ,
the superpotential @xmath32 contains trilinear terms in @xmath33 and @xmath34 : @xmath35 here @xmath36 and @xmath37 are the appropriate holomorphic functions of the moduli fields @xmath38 and @xmath10 , respectively . in eq .
( [ eqn3 ] ) we have omitted the generation indices . in ref.@xmath39
it is shown that the cubic terms of @xmath34 in the superpotential are corrected by instantons on the string worldsheet .
it should be noted that in the calabi - yau string , @xmath40 and @xmath34 couple separately to the @xmath38 moduli and the @xmath10 moduli , respectively @xcite . as mentioned above , in the observable sector
, the seed of susy breaking is expected to be a nonvanishing vev of @xmath41 @xcite , which is the @xmath5-component of the moduli field @xmath10 .
the @xmath5-components of the dilaton @xmath29 and the moduli @xmath38 are assumed to have vanishing vevs .
this situation is the same as that in the moduli - dominated susy breaking scenario@xcite .
hereafter we denote the vev of the moduli field @xmath10 by the same letter as the field .
for example , we write @xmath42 . in eq .
( [ eqn3 ] ) the trilinear coupling of @xmath34 contains the terms @xmath43 if the singlet field @xmath44 acquires a nonvanishing vev @xmath45 @xcite , then @xmath46 , @xmath47 , @xmath48 and @xmath49 in @xmath34 gain supersymmetric masses @xmath50 also , we have the soft susy breaking scalar masses , which are given by @xmath51 where @xmath52 denotes the derivative with respect to @xmath10 . therefore , the scalar particles in @xmath46 , @xmath47 , @xmath48 and @xmath49 have the masses squared @xmath53 . since only the anti - generation fields @xmath34 couple to the @xmath10 moduli , the fields @xmath34 behave as messenger fields in the framework of the gauge - mediated susy breaking scenario .
it should be noted that the flavor dependence of @xmath54 is not transmitted to the generation fields @xmath40 . in the gut - type models which accommodate the mssm at low energies , @xmath45 and @xmath55
are expected to be larger than @xmath56 gev @xcite . in this case
these messenger fields will become sufficiently heavy compared to the electroweak scale .
the other extra fields belonging to one set of the vector - like multiplet also become massive near or slightly above the intermediate energy scale via nonrenormalizable interactions @xcite . on the other hand , due to vanishing @xmath5-terms of the @xmath38 moduli , the generation fields @xmath40 can not behave as messenger fields .
the messenger fields @xmath34 have the quantum number of the standard model gauge group .
therefore , integrating out the messenger sector gives rise to gaugino masses at one loop .
estimating one - loop diagrams , one finds that the gaugino masses induced are @xmath57 where @xmath58 and the @xmath59s are the gauge coupling constants ( @xmath60 ) .
the soft scalar masses in the low - energy mssm sector , in which @xmath61 sets of the generation fields @xmath62 remain light , arise at leading order from two - loop diagrams .
messenger fields , gauge bosons and gauginos take part in the internal lines of the two - loop diagrams . consequently ,
the soft scalar masses induced are of the same order as the gaugino masses : @xmath63 provided that @xmath64 is @xmath65 , the mass spectrum of superparticles is consistent with that in the mssm .
as mentioned above , the significant features of this scenario are that there are sufficient degeneracies among squarks ( sleptons ) to ensure adequate suppression of fcnc and that no new @xmath66 phases are induced in soft susy breaking parameters . as a consequence of the worldsheet instanton effects , @xmath37 is of the form @xmath67 where @xmath68 is a constant .
@xmath69 is related to the compactification radius @xmath70 as @xmath71 in units of the string scale .
the string scale @xmath72 is defined by @xmath73 , where @xmath74 is the string tension . from eq .
( [ eqn10 ] ) , @xmath64 in eq .
( [ eqn8 ] ) becomes @xmath75 . in view of the fact that @xmath76 corresponds to an axion - like field , it is adequate for us to require the periodicity of @xmath37 in units of the string scale .
in this context we postulate that @xmath77 .
thus , we have @xmath78 with the scale @xmath79 , we find that the gaugino masses are on the order of a few hundred gev .
now we proceed to discuss the gravitino mass . in the
weakly coupled heterotic string , the khler potential is given by @xcite @xmath80 here the dots represent the higher - order terms . using the above khler potential and assuming a vanishing cosmological constant , we obtain the gravitino mass @xmath81 in string theory , the compactification radius is expected to be around the inverse string scale . from eqs .
( [ eqn9 ] ) , ( [ eqn11 ] ) and ( [ eqn13 ] ) we obtain the phenomenologically viable relations @xmath82 since the gaugino mass and the scalar mass are also induced by gravitational interactions , we must study whether or not the gauge - mediated contribution is dominated by a gravity - mediated contribution . in the
weakly coupled heterotic string , the gauge - kinetic function in the observable sector is of the form @xmath83 , and we have @xmath84 in the weak coupling region .
the second term in @xmath85 comes from the string one - loop correction and @xmath86 is a small constant fixed by gauge and gravitational anomaly@xcite@xcite .
the gravity - mediated contribution to the gaugino mass is given by @xmath87 since we have @xmath88 at the unification scale , eq .
( [ eqn15 ] ) cac be rewritten as @xmath89 from eqs .
( [ eqn11 ] ) and ( [ eqn16 ] ) the ratio of @xmath90 to @xmath91 becomes @xmath92 further , the gravity - mediated contribution to the scalar mass is given by @xmath93 up to a one - loop correction @xcite .
then the ratio of @xmath94 to @xmath8 becomes @xmath95 where we have taken @xmath96 at the unification scale . as can be deduced from eqs .
( @xmath97 ) and ( @xmath98 ) , the gauge - mediated contribution to the gaugino mass and scalar mass are dominant compared to the gravity - mediated contribution .
finally , we discuss the @xmath99 problem . in the present model the @xmath99 term comes from the first term in eq .
( [ eqn3 ] ) with @xmath100 .
the @xmath101 term is generated in the two - loop diagram as shown in fig.[fig1 ] .
the messenger fields ( @xmath34 ) propagate in the innermost loop .
the external lines are the scalar fields @xmath102 and @xmath103 and the remaining internal line has the fermion mass insertion @xmath104 .
the ratio of @xmath105 to @xmath106 becomes @xmath107 provided that @xmath108 . introducing a certain type of flavor symmetry ,
we can obtain @xmath109 @xcite .
thus , the @xmath99 problem is solved . in conclusion
, we propose a new model of susy breaking mediation on the basis of the weakly coupled heterotic string on the calabi - yau compactification . in this model
, susy breaking is mediated to the mssm sector by two phases . in the first phase ,
susy breaking in the hidden sector is transmitted to anti - generation fields via gravitational interactions .
subsequent transmission of susy breaking occurs via gauge interactions .
although the moduli fields are charged under the flavor symmetries , due to the flavor - blind nature of gauge interactions , we obtain the universal soft scalar masses .
the mass spectra of superparticles obtained here are phenomenologically viable .
|
Subsets and Splits
Data Mining Abstracts
This query searches for and retrieves 2000 documents containing the phrase 'data mining' in the abstract, providing limited but specific document samples.